I'm sure that a lot of the l33t h4x0rs here think that Supabase sucks and is only for amateurs but I'll say that as a former engineer who's getting back into building fun side projects again, Supabase has been incredible and just what I wanted. It's my favorite new product that I've started using in the last year. I hope they build out an enormous TAM of people who don't want to live inside a terminal and make a ton of money.
A non-technical family member is working on a tech project, and giving them Lovable.dev with Supabase as a backend was like complete magic. No amount of fiddling with terminals or propping up postgres is too little.
We technical people always underestimate how fast things change when non-technical users can finally get things done without opening the hood.
Back in the day we'd call this phase a design and workflow prototype as to not have to deal with all the technical components until the actual flow and concept is done.
Feels we're skipping these steps and "generating" prototypes that may or may not satisfy the need and moving forward with that code into final.
One of the huge benefits of things like Invision, Marvel, Avocode, Figma, etc. was to allow the idea and flow to truly get its legs and skip the days where devs would plop right into code and do 100s of iterations and updates in actual code. This was a huge gain in development and opened up roles for PMs and UI/UX, while keeping developer work more focused on the actual implementation.
Feels these generate design & code tools are regressing back to direct-Code prototypes without all that workflow and understanding of what should actually be happening BEFORE the code, and instead will return to the distractions of the "How", and its millions of iterations and updates, rather than "What".
Some of this was already unfortunately happening due to Figma's loss of focus on workflow and collaboration, but seems these AI generation tools have made many completely lose sight of what was nice about the improved workflow of planning, simply because we CAN now generate the things we think we want, doesn't mean we should, especially before we know what we actually want / need.
there is no need to this tedious, boring phase which you miss, especially since it still requires a significant of coding effort (eg to stitch a backend to figma).
you can vibe code a fully working UI+backend that requires way less effort so why bother with planning and iterating on the UI separately at all?
anybody who actually knows what they are doing gets 10x from these tools plus they enable non-coders to bring ideas to the market and do it fast.
That's always been the justification to skip this phase :). Tools have just changed. One-person to small-team wonders that could code and build directly made the same arguments.
My point isn't to stitch things to Figma, that's abhorrent to me as well. My point is to not get bogged down on the implementation details, in this case an actually working DB, those tables, etc, but rather less fidelity actual full flow concepts that can be generated and iterated.
Then that can be fed into a magic genie GPT that generates the front-end, back-end, and all that good jazz.
If the effort to produce websites goes tends to zero, the value of websites will surely tend to zero. Either issues with security and maintainability will be a break on this tendency or we will get to a point where generating a custom website will be something trivial that will be done on demand.
The thing is, the cost of producing websites is already pretty low, but the value of websites mostly derives from network effects. So a rising flood of micro crud saas products will not be likely to generate much added value. And since interoperability will drive complexity, and transformer based LLMs are inherently limited at compositional tasks, any unforeseen value tapped by these extra websites will likely be offset by the maintainability and security breaks I mentioned. And because there is a delay in this signal, there is likely to be a bullwhip effect: an explosion of sites now and a burnout in a couple of years in which a lot of people will get severely turned off by the whole experience.
I couldn't agree more. "Vibe coding" is pretty cool, but it's not sustainable at least with with current technology. You're much better off being a knowledgeable developer who can guide an an LLM to write code for you.
One thing I will agree on though is that LLMs make it easier to iterate or try ideas and see if they'll work. I've been doing that a ton in my projects where I'll ask an LLM to build an interface and then if I like it I'll clean it up and or rebuild it myself.
I doubt that I'll ever use Figma to design, it's just too foreign to me. But LLMs let me work in a medium that I understand (code) while iterating quickly and trying ideas that I would never attempt because I wouldn't and be sure if they'd work out and it would take me a long time to implement them visually.
Really, that's where LLMs shine for me. Trying out an idea that you would even be capable of doing, but it would take you a long time. I can't tell you how many scripts I've asked ChatGPT or similar to write that I am fully capable of writing, but the return on investment just would not be there if I had to write it all by by hand. Additionally, I will use them to write scripts to debug problems or analyze logs/files. Again, things that I am perfectly capable of doing but would never do in the middle of a production issue because they would take too long and wouldn't necessarily yield results. With an LLM, I feel comfortable trying it out because at worst I'd burn a minute or two of of time and at best I can save myself hours. The return on investment just isn't there if it would take me 30 minutes to write that script and only then find out if it was useful or not.
LLMs are better search. Google burned down the house to keep itself warm and held off on LLMs until it was inevitable and are now pulling up ahead. This is the logical conclusion. LLMs will be monetized and enshitified by ads.
Soon, some free smart LLM code generators will stop generating certain outputs and instead suggest using commercial components that have paid for promotion.
This can only be true of some products. Often there are a lot of concerns like privacy, white labeling, legal consequences that need to be considered _before_ you vibe code.
> We technical people always underestimate how fast things change when non-technical users can finally get things done without opening the hood.
This is good and bad. Non-technical users throwing up a prototype quickly is good. Non-technical users pushing that prototype into production with its security holes and non-obvious bugs is bad. It's easy for non-technical users to get a false sense of confidence if the thing they make looks good. This has been true since the RAD days of Delphi and VisualBasic.
Knowing the industry I'm pretty sure they will all push those AI prototypes to production - because they did the same with non-AI prototypes before. Now the question is once they inevitably pull in experienced folk for maintenance, refactoring and debugging, will it be easier or harder than working with that retired solo devs spaghetti codebase?
From looking at "vibe coding" tools their output is about the quality of bad body shop contractors. It's entirely possible for experienced devs to come in and fix it.
I think there's going to be the same problems as there are fixing bad body shop code. The companies that pushed their "vibe code" for a few dollars worth of AI tokens will expect people to work for pennies and/or have unreasonable time demands. There's also no ability to interview the original authors to figure out what they were thinking.
Meanwhile their customers are getting screwed over with data leaks if not outright hacks (depending on the app).
It's not a whole new issue, shitty contractors have existed for decades, but AI is pushing down the value of actual expertise.
I think this is just another correction. The software market is worth several trillion dollars now. Enterprise is pushing against the rise in labor costs. It will backfire as it did every single time and in a few years competent developers will be worth their weight in platinum.
For nearly 50 years now, software causes disruption, demand drives labor costs, enterprise responds with some silver bullet, haircuts in expensive suits collect bonuses, their masters pocket capital gains, and the chicken come home to roost with a cycle of disruption and labor cost increases. LLMs are being sold as disruption but it's actually another generation of enterprise tech. Hence the confusion. Vibe coding is just PR. Karpathy knows what he's doing.
> Non-technical users pushing that prototype into production with its security holes and non-obvious bugs is bad.
I beg to differ. Non-technical users pushing anything into production is GREAT!
For many, that's the only way they can get their internal tool done.
For many others, that's the only way they might get enough buyers and capital to hire a "real" developer to get rid of the security holes and non-obvious bugs.
I mean, it's not like every "senior developer" is immune from having obvious-in-retrospect security holes. Wasn't there a huge dating app recently with a glaring issue where you could list and access every photo and conversation ever shared, because nobody on their professional tech team secured the endpoints against enumeration of IDs?
What about users who sign up for these insecure apps and have their data and possibly their identity stolen due to the misplaced trust? That this already happens is no excuse to encourage even less security by encouraging novices to believe they are experts.
I agree it is great that more people can build software, but let's not pretend there are zero downsides.
This is a contrived situation. Most of the apps in discussion see little to no use and go dead soon after launch. The vast majority are collecting little data of negligible risk.
If a user is confident enough about a no name company that they give them enough info to make identity theft a possibility, it was only a matter of time before a spammer/phishing attack gets them anyway
> Most of the apps in discussion see little to no use and go dead soon after launch
That's not convincing. Of the apps that do get used, the vibe-coded ones will likely be unsafe.
> If a user is confident enough about a no name company that they give them enough info to make identity theft a possibility
That's completely unrelated. You can give a company very little information. Any of it being leaked is unacceptable. You can find a lot from an email, or a phone number.
People are taught, by CNBC, by suits, by hacks, that you can trust the apps on your commercials and it will be fine. It likely won't be, and your response is exactly why. Many of you are apathetic to the idea of doing right by people.
So people are manipulated, and some of them are elderly and don't even understand how computers work. This is reason enough to care about what they are exposed to, not say "let's burn it all down with shitty vibe-coding because users are dumb anyway."
My feeling is that this is similar to saying, "non-professional AirBnB hosts are a terrible security nightmare, and the fact that people are not much safer in regulated hotels is no excuse to encourage even less security by encouraging novices to play in the hospitality business".
AirBnB externality is not the safety risk for guests (although I personally ended up in some sketchy situations years ago, I don't use it anymore, mainly because:) the real externality is imposed on the inhabitants of popular tourist destinations.
There was a reason the industry was regulated, and circumventing these reasons with an app has been a net negative to society.
Even us entrepreneurially minded technical devs cut corners on our personal projects that we just want to through a Stripe integration or Solana Wallet connect on
And large companies with FTC and DOJ involved data breaches just wind up offering credits to users as compensation
so for non-technical creators to get into the mix, this just expands how many projects there are that get big enough to need dedicated UX and engineers
Having worked with it quite a bit I'm still not sure I really understand what it is, which sounds like a bizarre sentence but:
It's Postgres, but bundled with some extensions and Postgrest. And a database UI. But hosted and it runs locally also by pulling the separate parts. Running it locally has issues though, so much so that I found it easier to run a docker compose of the separate parts from scratch and at that point just carry that through to a deployment, at which point is there still a reason to use Supabase rather than another hosted Postgres with the extensions?
The product story is that people want to build apps and naturally find themselves having to handle:
- remote state
- authoritative logic that can't run solely on the user's device because you can't trust it
- authentication
each of which is annoying when you're focused on building the user-facing app experience. Supabase solves all three without you needing to touch any infrastructure. The self-hosting thing just provides insurance that users are not completely locked in to their platform, which is a big concern when you're outsourcing basically your entire backend stack.
It's also implied, and proven by some, that having access to Postgres means you can up and leave Supabase if you want to later. It won't be snap-your-fingers easy, but it's more direct than other hosted SaaS where you can't access your data or the schemas.
Realistically 99% of the users would still be screwed if they ever shut down, regardless of if it's open (see: Parse)... but it gives people a some confidence to hear they're building on a platform that they could (strictly in theory) spin up their own instance of should a similar rug pull ever occur
They have also been giving back to postgres some of their extra work, and also their real time stuff i think is on erlang?
I agree you might prefer to choose the stack yourself, but for total n00bs and vibe coders supabase is a great start / boilerplate vs say the MEAN stack that was a hit 5y ago
I’ve been using Hasura and PostgREST for a few years now with real big production apps, in enterprise and in startups, and honestly the only problem with them is that backend engineers feel threatened.
They are great products that cover 95% of what a CRUD API does without hacks. They’re great tools in the hands of engineers too.
To me it’s not about vibe coding or AI. It is that it's pointless to reinvent the wheel on every single CRUD backend once again.
Experienced backend dev here who also uses Hasura for work at a successful small business. I think it's great at getting a prototype to production and solves real business problems that a solo dev could do by himself. As engineer #2 it's a mess, and it doesn't seem like a viable long term strategy.
I've only worked with Hasura, but I can say it's an insecure nightmare that forces anti-patterns. Your entire schema is exposed. Business logic gets pushed into your front end because where else do you run it unless you make an API wrapper. Likewise you can't easily customize your API without building an API on top of your API. You're doing weird extra network hops if you have other services that need the data but can't safely access it directly. You're pushed into fake open source where you can't always run the software independently. Who knows what will happen when the VC backers demand returns or the company deems the version you're on as not worth it to maintain compared to their radically different but more lucrative next version.
I think the people who write this off as "backend engineers feel threatened" aren't taking the time to understand the arguments they're hearing
Backends are far messier (especially when built over time by a team), more expensive and less flexible than a GraphQL or PostgREST's api.
> I've only worked with Hasura, but I can say it's an insecure nightmare that forces anti-patterns
Writing backend code without knowing what you're doing is also an insecure nightmare that forces anti-patterns. All good engineering practices still need to apply to Hasura.
Nothing says that "everything must go through it". Use it for the parts it fits well, use a normal backend for the non-CRUD parts. This makes securing tables easier for both Hasura and PostgREST.
> Business logic gets pushed into your front end because where else do you run it unless you make an API wrapper. You're doing weird extra network hops if you have other services that need the data but can't safely access it directly
I'm gonna disagree a bit with the sibling post here. If you think that going through Hasura for everything is not working: just don't.
This is 100% a self-imposed limitation. Hasura and PostgREST still allow you to have a separate backend that goes around it. There is nothing forbidding you from accessing the DB directly from another backend. This is not different from accessing the same database from two different classes. Keep the 100% CRUD part on Hasura/PostgREST, keep the fiddly bits in the backend.
The kind of dogma that says that everything must be built with those tools produces worse apps. You're describing it yourself.
> I think the people who write this off as "backend engineers feel threatened" aren't taking the time to understand the arguments they're hearing
I have heard the arguments and all I hear is people complaining about how hard it is to shove round pieces in square holes. These tools can be used correctly, but just like anything else they have a soft spot that you have to learn.
Once again: "use right tool for the job" doesn't mean you can only use a single tool in your project.
I've only played with these kinds of plug and play databases, but mixing and matching seems like the worst of both worlds. The plug and play is gone, because some things might me in API 1, some others in API 2, and maybe worst of all, their domains might overlap. So you need to know that the "boring" changes happen via the postgREST, but the fancier ones via some custom API. The APIs will probably also drift apart in small ways, making everything even more error prone.
As a long-time Hasura stan, I can't agree with this in any way.
> Your entire schema is exposed
In what sense? All queries to the DB go thru Hasura's API, there is no direct DB access. Roles are incredibly easy to set up and limit access on. Auth is easy to configure.
If you're really upset about this direct access, you can just hide the GQL endpoint and put REST endpoints that execute GQL queries in front of Hasura.
> Business logic gets pushed into your front end because where else do you run it unless you make an API wrapper
> Likewise you can't easily customize your API without building an API on top of your API. You're doing weird extra network hops
... How is an API that queries Hasura via GQL any different than an API that queries PG via SQL? Put your business logic in an API. Separating direct data access from API endpoints is a long-since solved problem.
Colocating Hasura and PG or Hasura and your API makes these network hops trivial.
Since Hasura also manages roles and access control, these "extra hops" are big value adds.
> You're pushed into fake open source where you can't always run the software independently
... Are you implying they will scrub the internet of their docker images? I always self-host Hasura. Have for years.
> I think the people who write this off as "backend engineers feel threatened" aren't taking the time to understand the arguments they're hearing
I think your arguments pretty much sum up why people think it's just about backend engineers feeling threatened - your sole point with any merit is that there's one extra network leg, but in a microservices world that's generally completely inconsequential.
I like PostgREST for some of it's use cases (views mostly), but the issue I have with it is that I don't often want a user to have direct access to the database, even if it's limited to their own data.
Mike can edit his name and his bio. He could edit some karma metric that he's got view access to but no write access to. That's fine, I can introduce an RLS policy to control this. Now Mike wants to edit his e-mail.
Now I need to send a confirmation e-mail to make sure the e-mail is valid, but at this point I can't protect the integrity of the database with RLS because the e-mail/receipt/confirm loop lives outside the database entirely. I can attach webhooks for this and use pg_net, but I could quickly have a lot of triggers firing webhooks inside my database and now most of my business logic is trapped in SQL and is at the mercy of how far pg_net will scale the increasing amount of triggers on a growing database.
Even for simple CRUD apps, there's so much else happening outside of the database that makes this get really gnarly really fast.
> Now I need to send a confirmation e-mail to make sure the e-mail is valid, but at this point I can't protect the integrity of the database with RLS because the e-mail/receipt/confirm loop lives outside the database entirely
Congratulations: that's not basic CRUD anymore, so you ran into the 5% of cases not covered by an automatic CRUD API.
And I don't see what's the dilemma here. Just use a normal endpoint. Keep using PostgREST to save time.
You don't have to throw the baby away with the bathwater just because it doesn't cover 5% of cases the way you want.
It's a rite of passage to realize that "use the right tool for the job" means you can use two tools at the same time for the same project. There are nails and screws. You can use a hammer and a screwdriver at the same time.
>You can use a hammer and a screwdriver at the same time
How do you balance the nail and screw? I'm serious, I'm trying to picture this, hammer in one hand, screwdriver in the other, and the problem I see here is the nail and screw need to be set first, which implies I can't completely use them both at the same time.
Perhaps my brain is too literal here, but I can't figure how to do this without starting with one or the other first
I have used them too, and I would say that at least for Hasura, performance can be poor for the generated queries. You have to be careful. Especially since they gate metrics behind their enterprise offering.
This is the same for any GraphQL backend. And even REST backends can be misused: I've fixed way too many joins-in-the-frontend that were causing N+1 queries in lists.
World still needs a replacement for Microsoft Access on the web.
It’s been so long that new ideas are solving parts on the access spectrum without seemingly being aware of it.
Supabase and others would have a smaller footprint to add an app layer and reporting layer to their tool since it is data as the cornerstone not an afterthought
What’s Supabase’s exit strategy? Are they sustainable long term as a standalone business?
You can also see how money is starting to chase “vibe coding” — as long as you say the magic words, even if your product is only tangentially related to it, you can get funding!
Reading the tea leaves, Series D means they opted for more funding vs IPO. They claim to have 2 million users, but they're open core so how many are paying? Maybe their books aren't looking that great. Wall street doesn't understand database vendors outside of "big data", so they're probably hoping for acquisition. Not sure who would buy them though, as PostgreSQL vendors are kind of a dime-a-dozen these days...
If lovable, bolt.new, etc kept integrating with them, that's a money maker without needing to do much sales. There's a wave of AI tools that require somehow save state and Supabase provides that. I'm absolutely amazed others haven't jumped in the same ship yet.
That definitely seems to be the play. Keep funneling in users from Lovable/bolt.new and keep building revenue or hope to be acquired if one of those vibe coding tools gets huge.
A lot are paying, including me for multiple projects. They have a pretty good offering. I used to use them for dev and prod, but now using neon for dev. Supabase still for prod. I had switched from mongo to supabase. I may switch to neon for prod but not in a rush.
They also offer so much more than just postgres. Though I use them only for postgres myself.
Since you use both supabase and neon, any particular strength or weakness to keep supabase for prod? I just moved my app to neon today (easy enough to test it!) and am enjoying the auto-scaling features and UI is great on neon. But I'm curious about how supabase stacks up.
it's an aggressive preemptive round, so i'd guess 2b/50 = 40M of revenue. Probably low margins since the free tier/ hosting postgres nature of the business.
Sure but ultimately they're still just selling something that is already free and wrapping AWS. These business models aren't sustainable unless you trash your free product, which also isn't sustainable. Presumably they have a good deal with their cloud vendor, AWS I think, but I think its safe to assume they lose A LOT on their free products.
This is like if Google Spanner were open sourced tomorrow morning: realistically how many people are going to learn how to deploy a thing that was built by Google for Google to serve an ultra-specific persona?
Maybe you might get some Amazon-sized whale peeking at it for bits to improve their own product, but the entire value prop is that it's a managed service: you're probably going to continue paying for it to be managed for you.
IMO it also depends on how the whole process is tied together.
I always loved Vercel for their easy hosting of Next.js with included CI/CD, but I recently switched to self-hosting - their pricing switched from a flat, worry-free $20/month to an unpredictable whatever-it-may-cost plus it sent me 10+ emails every single month about hitting some quotas that they introduced and I couldn't find a good way to stop that.
> Are they sustainable long term as a standalone business?
It's bananas to me that questions like these could be unanswered even 5 years after the business started. This possibly cannot be the most efficient way for finding new solutions and "disrupting" stale industries?
It's very common for tech companies to go public without being profitable. As long as their growth is good and they have a reasonable story for how they will achieve profitability, then it typically makes sense. Of course every company is different and not all will reach their profitability goals post-IPO, but in many cases, it wouldn't make sense to wait for profitability before going public.
Discord has the user growth to make up for it. It’s practically a household name on the level of Google now. I see many people boosting servers too, so clearly they are making a bunch of money.
Discord has a fairly successful subscription product that is generating tens of millions in revenue. They most certainly have made more than 0 dollars. Profitable? Less likely.
To be fair, their $2B valuation is probably the most reasonable valuation we've seen in years. That doesn't negate the question of how they plan to turn a profit.
If they truly have 3.5 million databases, that's only ~$500 per database to recoup the investments, that doesn't seem to crazy. Companies like OpenAI or Twitter/X are never going to be profitable enough to cover what they've already spend/cost. Supabase could because the amount is so much lower and they have paying customer, but I'd like to emphasize the "could".
and setting up postgresql on a simple VPS is so easy... You can literally ask Gemini 2.5 Pro or o3 or Sonnet 3.7 and do it in 15-30 minutes...
Learned helplessness is really something and vibe coding is overrated imo:
https://www.lycee.ai/blog/why-vibe-coding-is-overrated
> What’s Supabase’s exit strategy? Are they sustainable long term as a standalone business?
Acquisition best case, Private Equity worst case.
Do you see Supabase going public on the stock market? Perhaps unless they do what Cloudflare done and are replicating AWS, it may be hard to see a stock market debut.
Supabase is basically AWS Postgres under the hood. It's popular amongst hobbyists and small teams but I'm not sure whether any large teams actively use it. Once you're past the point of serious business, it's much more cost effective to host everything by yourself.
What is serious business? I think supabase can scale brilliantly, and it doesn't lock you in, if you have need for some special infra you can build and integrate it, I don't know but you could possibly even use FDW to access a postgres you run yourself.
Also they can't run on AWS postgres with all their postgres plug-ins AFAIK.
The point of "cheaper to host everything yourself" is a lot higher than what most estimate.
My only concern is that is supabase goes out of business or go evil you're gonna have a bad time, however everything is open-source
Serious business is when you need to maintain uptime and stability. Not just me, but a lot of folks on the Supabase reddit have complained often about the insane downtimes that we've experienced at times with the platform. I would 100% use it for prototypes and MVPs, but for production? Neither me nor a lot of others would touch it with a pole, even though I'm sure your experience might be different.
Supabase at is minimum providing a PostgreSQL server, pooler (they started on pgbouncer, how their own) and a PostREST API, and support, backup, logging etc. You can be doing serious business and not have the time/people to run these reliably self-hosted.
They also provide Auth almost to the Auth0 level and Edge functions like Vercel, S3 like storage (sharing the db's permission system), and websocket/presence backed by Elixir. TBH they are a compelling value, at least for us.
Granted, Supabase does provide a lot under the hood, and it's excellent for whipping out a quick MVP, but I wouldn't count it production ready even today, simply because of the downtime I've experienced at times (even if their dashboards say otherwise). And it's not just an isolated case - a lot of folks on reddit complain with the same issues. Perhaps your treatment is different because of the high spend you guys have.
My former place ran a lot of RDS Postgres but also loved Supabase. It's more than just hosted DB because it has loads of value adds like web-based table editing, auth, edge functions, row-level security, easy hooks and triggers. We were capable of operating RDS but the cost of operations in dev hours was high. Supabase was super easy for moderate price and readily compatible with our RDS and Redshift.
I don't disagree. Supabase does provide a lot of functions under the hood that one would have to build out individually otherwise. I really like their lock-in model - you're not locked in with your data, but because of the extra functionality that they provide to your database.
Acquisition. All of these VC companies raise unsustainable levels of money in hopes of acquisition or IPO. Supabase seems to be leaning towards acquisition.
What an absolute joke. Their exit strategy is presumably to keep chasing the high and find more ways to integrate AI. The era of building good software for fun and profit is coming to an end.
I think their product is sound, they build essentially a backend as a service platform on open-source software. That doesn't make it easy to run at scale, so you probably wanna use their paid offering unless you plan to hire a lot of staff to maintain it, but it is possible and they support small scale dev envs
Have you thought about possibly hiring a junior engineer to make the transition possible? I don't know what your use case but creating a solution that fits your needs would be worth it IMO. You're already spending a years worth of dev salary which can get you plenty of good talent around the world with.
>Supabase is currently used by two million developers who manage more than 3.5 million databases. The startup supports Postgres, the most popular developer database system that’s an alternative to Google’s Firebase. Supabase’s goal: To be a one-stop backend for developers and "vibe coders."
How many of those users are paid. You can sign up for free without a credit card.
It's cool, for certain use cases. I ended up trying it for a few months before switching to Django.
If you ONLY need to store data behind some authentication, and handle everything else on the frontend, it's great. Once you need to try some serverside logic it gets weird. I'm open to being wrong, but I found firebase phenomenally more polished and easier to work with particularly when you get to firebase functions compared to edge functions.
Self hosting requires magical tricks, it's clearly not a focus for them right now.
I hope they keep the free tier intact. While it's not perfect, if your in a situation where you can spend absolutely no money you can easily use it learning ( or for portfolio piece).
Has anything changed recently? ~1year ago I installed a local instance (that I still use today for logging LLM stats) and IIRC all I had to do was "docker compose up". All the dockers are still starting for me at boot from that 1yo install, to this day. (I use it on 127.0 so no SSE & stuff, perhaps that's where the pain points are? Dunno, but for my local logging needs it's perfect).
Hosting it on an actual server with a URL is not a fun experience. You need to generate a specific type of string to get it to work.
This isn't documented anywhere. Deep deep in their GitHub issues you'll find a script for generating this magic string which needs to be set as an environment variable.
Looks like it is just an issue of correctly making a jwt token, if you are not using their client libraries, but you can also just do it via their docs https://supabase.com/docs/guides/self-hosting/docker#generat... now (not sure how long you've been able to do in the docs)
Sure "it's in the docs" but last time our devops tried the compose file with ~10 or so services it took several days of fiddling with all sorts of different issues. It is just not made for selfhosting at all. It can be so much simpler but JS devs like it different.
It’s normal Postgres. There’s no need to handle everything on the front end. The tutorials nudge you to learn RLS and use their SDKs for the client, but you can write perfectly normal server side code as well.
Yeah, it's a bit wonky, especially when you are dealing with configuring specific combination of supabase/deno/typescript features (e.g. stage 2 vs stage 3 decorators)
For my current project I basically need a backend server for processing some basic game logic.
I had done something similar in Firebase and it was easy. Supabase wasn't straightforward here. It got to a point where I'm sure I could eventually get it working, but I also think I'm outside the expected usecase.
Is the new valuation multiplier number of developers on platform instead of revenue? Valuing at $1000/developer is kind of insane. Valuing at $570/database is also nuts. It's a cool product but I hope the founders can find a win in what must be a pretty cramped cap table.
Their 2024 revenue was estimated at $16.8M [1] and $10.5M in 2023. If you extrapolate that growth rate +1 year you can assume it's now $26.9M. Another source estimates it at $15M in 2025 [2].
So if you assume their revenue is in that range, you're looking at 66x to 133x ARR multiple. In today's market that's quite a big markup. Standard SaaS right now is probably more like 5-15x. AI is a lot more (but Supabase isn't AI). But they are a key leader in their market, so probably get a meaningful bonus for that. And I'm sure a lot of big industry investors were competing against each other for the Supabase deal, so that definitely would have helped valuation too. Also, at their maturity today, they are probably showing some great success signing big enterprise deals and telling a story about how that will grow.
That being said, those factors alone don't answer 66-133x. Perhaps Supabase's strongest angle is their opportunity for product-led growth:
- They have a huge number of people on a free tier
- The growth rate of free tier users might be accelerating
- The conversion rate of free tier users to paid users might also be increasing
- They're adding more things that people can pay for, increasing LTV of customers. e.g., for my business, we probably 20x our Supabase cost in the last 6 months - most of that is due to our growth but also there are a lot of things we can buy from Supabase beyond compute.
So I would assume, in addition to the above, they're telling a story about their actual revenue growth rate will accelerate meaningfully because of all of these factors working together.
Lots of assumptions in here, but you can start to see how a lot of different factors + a hype multiple could lead to such a valuation.
I was a speaker in a local Supabase event just few weeks ago, https://shorturl.at/JwWMk. We had a local event in Abuja, Nigeria. There we promoted their Launch Week 14 series, highlighting new features from Supabase. In reality, it became an event to show people how to bootstrap a quick backend for their SME business in a weekend.
Nice, I gotta admit, I like seeing new stuff get funded like this - but man, whenever I read about crazy big rounds, I start wondering how these companies ever make real money vs just chasing hype till the next exit. Always makes me think about what actually lasts long term besides just burn and churn.
Personally think that supabase and vercel are giving AWS a run for their money, and will start to eat into their market share. The developer experience is far superior to AWS, in a way that cannot be argued away by AWS handling "more complexity and scale". Supabase/Vercerl products are superior to large cloud providers, and while they target a narrower aspect of the tech stack, and to smaller customers, they will expand into more enterprise as their users grow.
AWS needs to get their act together and start prioritizing developer experience
Also, supabase is looking like the go to database for ai created apps. Which will be a major tailwind
But it doesn't matter because they are transitively being used by virtue of Vercel/Supabase being on them. They could definitely charge more, that's for sure, but it's not like they don't get a pretty penny from others building atop them
I've been testing out PostgREST, RLS, and PL/pgSQL functions for a new app and I'm not wild about it. It's pretty complicated to grok the permission model, it's awkward to load up your DB with logic, and the LLMs kinda suck at reliably generating working queries, policies, functions, etc. So I'm not really convinced it's ideal for 'vibe coders'.
I built a 'serverless' platform that's similar, smaller but more focus on low-code than Supabase https://saasufy.com/
All the components are declarative HTML and update in realtime. Similar concept as HTMX but doesn't require any backend code. You can still implement complex UX, authentication, access control and filtered views (indexing and all).
Raising a very late funding stage in the private market to avoid an IPO in the violent public market I presume? I’m guessing this is the new startup norm for a while, along with reducing burn or exiting at a big discount if not just dying
Supabase is quite well integrated with vibe coding tools - cursor, replit, v0, etc. I agree that Firebase is a superior well-integrated product, but IMO the vibes are on Supabase side.
As an observation, the quotations and the article itself doesn’t appear very thoughtful or well reasoned. It sounds more like a VC wanted a big deal (or at least to represent it as such) and made a bet, invited his buddies(?) along, and the company has seen a recent bump from “vibe coding”. Some allusion to Larry Ellison. Very light on information to form an opinion…
I'm sure the team might not be expecting this level of valuation. These trends won't last long. Make hay while the hype shines. Who knows how soon people would forget that there was something called vibe-coding and a back-end development.
I've been a lukewarm user of Supabase for my side projects. Unfortunately the amount of work to get off of it has been too high for me to leave.
The major issue is - cost. It is way more expensive than I realized as they have so many little ways they charge you. It's almost like death by thousands of paper cuts. My bill for my app with just a few thousand users was $70 last month.
I do like the tooling and all, but the pricing has been very confusing.
Kind of the same feeling, I don't use all of services they offer either and when I looked at self-hosting, it all seemed kind of heavy and fragile to self-host. I ended up replicating the parts I used with a small API layer connected to a managed postgres db for a tenth of the cost or something. I'd say it's pretty handy for prototyping but not sure I'd want to build a business on the back of it.
Few Thousand!?! Sound very reasonable to me. Monetize just two of those users at $35 per month and your server costs are covered. Or run it yourself, there's a lot of moving parts but it's all open source.
Not to be negative, but that seems super huge for its worth. How can you even recoup that with barely anyone paying? Founders trying to exit is comprehensible, but how can VCs sign this deal?
I'm a happy Supabase customer. It's not just postgres, it's postgres + auth + functions + storage all in one place, with a very good local Docker environment and a variety of nice client SDKs.
It's a quick/convenient way to get a "full" backend up and running. The other option (Firebase) can't be self-hosted and has some absurd pricing footguns.
It has a nice web interface, comes with support for auth (which almost all apps need), and has a couple other nice things like storage buckets and serverless functions. When doing a hobby project, I don't want to rent a server from hetzner, set up SSH, connect, figure out how to install postgres, make sure everything is configured and so on. It's much nicer to click one button in the supabase UI and have everything I need
While Supabase does support many equivalent Firebase features like a database, auth, functions, storage, etc, Supabase doesn't support website hosting whereas Firebase does. Supabase is mainly for backend stuff (BaaS) whereas Firebase is for both frontend and backend (PaaS).
To get the equivalent functionality of Firebase, you'd need to add something like Netlify or Vercel to pair it with your Supabase backend.
The only that hasn't changed in this industry is that engineers apparently can't stand new, green beginners getting into the field.
"They ship buggy, insecure messes"
"They don't know how to fix what AI gave them"
etc etc etc
Right. Like that same thing hasn't been happening literally during the entire existence of programming. I, for one, welcome the vibe coders. I hope it grows their interest in the field and encourages them to go deeper and learn more. Will some be lazy and not even try? Of course! Will some get curious and learn the ins and outs? Absolutely.
The thing I believe least in this article is that the investor is betting their career on this investment. Maybe they think that, but they’re still multi-millionaires making bets with other people’s money. They’ll be fine.
I don't quite understand the high valuation. I use Supabase for the database and file storage (legacy choice). But those seem interchangeable. Are people using many many more features, if so how?
I'm from the future again, and I predict that either Replicate or fal.ai will get acquired by one of these so-called 'Vibe-coding' companies. Supabase included.
When I see valuations like this, they are overvalued until they use that money to acquire another company for a total addressable market expansion.
It's really sad that nobody is actually trying to be a real company these days. When Steve Jobs was asked that if he will ever sell Apple, he physically recoiled and scolded the person asking such things. Nowadays, all of these companies are being set up just to be acquired. Nobody has any ambition anymore to be long lasting, profitable and an actual business.
So basically the nobody wants to work of the business longevity world? What is the actual difference? If their goal is profitability and eventually retirement why not do it on an accelerated timescale and exit with millions in their 30s instead of hundreds of thousands in their 70s?
This seems like another one of those VC companies which gets pushed by the VCs to the other companies in their portfolios as a probable integration point.
The whole growth of vibe coding really did help them because I don't think actual developers use it because putting things like functions in the database and authorization in the database is something that we learnt a few decades ago is a bad idea.
So I would guess they are used by massive amounts of developers who are new to coding or do not fully know how to code, but are becoming developers and who love the free databases Supabase provides.
> The whole growth of vibe coding really did help them because I don't think actual developers use it because putting things like functions in the database and authorization in the database is something that we learnt a few decades ago is a bad idea.
Why are those things a bad ideas? You could be right but if you insist on making value judgements without explanation or elaboration, you're going to sound like a whiny old crank who is scared of becoming obsolete.
What I mean is that, you know, we tried all of these things back in the day. Like putting your business logic in a database was something that was tried. It's not that the first time that these ideas have shown up. And they were basically tried and tested and found faulty. But now there's this whole new generation of sub-bar developers. And they are being fooled into thinking that all of this works because they just want to get their hands dirty. I guess when normies flood the scene the wolves make a killing.
We're currently using supabase in production. I was already planning to leave them. I feel like it's close to good, but still really does have a _lot_ of bugs. It really doesn't feel like it's been tested thoroughly, and the documentation, while present, leaves a lot to be desired.
My experience of supabase really demonstrates to me that the ideals of all of the postgres layer technologies - postrest, realtime via wal, jwt auth in the db -, just don't make for an easy experience. It all works (mostly) but I find it more annoying than useful and have to work around it more often than I'd like. I suppose I'm old school, but just building the things that one needs is often more robust and less work than trying to plug into what they've provided.
I really don't know what they're going to do with a series D. It seems they now _have_ to go for a high-value exit, but I really don't see which company would provide that exit.
I had similar experience with everything missing that final attention to detail and polish and having to write issues and ask other how they got past certain problems. I ended up switching to Pocketbase, and while it is not a complete or drop in replacement hosted service, it is light weight and approachable to feel comfortable that it can scale and be more stable long term.
We had a pretty bad experience with it on a NextJS app and had to gut it out (I decided not to charge the client for this multi-week refactoring effort to pull out Supabase since it was upon my recommendation that they went with Supabase, so that decision actually cost us thousands of dollars).
It is good to get started and no doubt useful for simple CRUD apps. But once you want to start doing more complicated stuff, a lot of the RLS primitives become very hard to maintain and test, for example. You could say that that's Postgres's fault, but Supabase strongly pushes you in that direction.
The tooling, while looking quite polished, just felt pretty half baked along with docs (at least a year ago when we pulled the plug). Try to implement even a halfway complicated permissions scheme with it and RLS and you are in for a world of hurt and seemingly unmaintainable code.
So we ditched Supabase Auth for AuthJS, and are using vanilla postgres with Prisma. That's worked well for us. All the tooling is relatively mature, it's easy to write tests, etc.
Maybe if AI is writing some of the code, it might get easier, but for now, I'm avoiding Supabase like the plague until I see a project that's relatively complex that's actually easy to maintain.
Agree, also the logging is very limited across the board,, database functions are impossible to debug without good logging, I think a lot of people don't realize its not a direct copy of Postgres, for the cloud version several Postgres functions are disabled by default.
Most of the heavy hitters in Postgres development are already sponsored by companies to do their work. Supabase has at least one Postgres committer on staff for instance.
> I wouldn't break into someone's house and tell them they painted their house an uninformed color.
but if you, an experienced professional painter, were hired to repaint somebody's house, and that person used bad paint (or something else entirely, like... i dunno, shellac, or white glue), your scope of work changes from "light prep and paint" to a much more involved job of undoing the previous work (or mistakes) and then getting to the original scope of work.
> If something sucks or won't scale, it will sort itself out in the market.
i truly don't mean this in a derogatory way but that sounds incredibly naive.
At the end of the day we are really just disagreeing on how seriously we should take AI-ification of software engineering.
I take a lighter stance that it’s mostly harmless and has a lot of educational positives. I guess you foresee a lot of potential harm.
I do appreciate people who are willing to be the canary in the coal mine. I just find that a lot of the criticism appears to be more gatekeeping than legitimate worries.
A lot of comments seem to take issue with Supabase being aligned to Vibe-coding, which is understandable, but I do think it's related! I think vibe-coding could very well be a real market-force, and as best as I can tell it exists within a specific type of tech circle. No one is vibe coding Elixir, no one is vibe coding Rust, people are vibe coding React + Node/Python, and more specifically I think people are watching streamers and YT videos on hot-new-frameworks-near-you and want to try them out! Supabase is absolutely a darling of this tech-circle, and I think a few other companies could be as well:
I did :) I made a browser-based MMO with Phoenix to test out liveview and learn the language: https://shopkeep.gg
And it was pretty annoying. Elixir doesn't really lend itself to vibe coding due to namespacing and aliasing of modules, pattern matching, all without static typing (I know, Dialyzer...). It also struggles to understand the difference between LiveComponents and LiveViews, where to send/handle messages between layers.
Without references to filenames, the agent perpetually decides "this doesn't exist, so I'll write it :)". I found it to be pretty challenging before figuring out I could force the agent to run `mix xref callers <Some.Module>` when trying to do cross-module refs.
But you can just give them a prompt to teach them. Case in point, Transact has a prompt that you can put into an LLM that enables it to one-shot some simple programs:
> The startup supports Postgres, the most popular developer database system that’s an alternative to Google’s Firebase. Supabase’s goal: To be a one-stop backend for developers and "vibe coders."
It's all fun and games until you realize the cool thing you made is locked into an overpriced cloud stack that will bleed you dry. "Vibe coders" seem to invariably end up offloading as much heavy lifting as possible to ultra-expensive PaaS/SaaS vendors, and those vendors are encouraging it (e.g. Vercel v0).
If you are trying to commercialize something, a popular project with bad margins is a better spot to be in than an unsuccessful project with good margins. If it's a personal learning project, that might not be the case.
These cloud back and stacks are very cheap at low volume and honestly I expect them to remain so or even go down in price.
I've seen many colleagues bootstrap something - even if they're not themselves very technical - because they've leveraged these well integrated low cost platforms.
I do think it’s binary. The project either shows potential to meet your goal or it doesn’t.
I think it’s rare that fails to show potential because of the underlying technology that’s chosen.
Sure, Vercel is relatively expensive. But I just don’t see how you’d throw in the towel because the costs are too high without first evaluating how to lower them.
If you’re saying that the evaluation is likely to show that you’re stuck - I have never seen that be the case personally.
As a "real engineer" you'd likely use LLMs differently. I save my conversations, have chats and codebase exegesis summarized into .txt files, and constantly refactor LLM output. I have an increasingly reliable sense of when to dip in and write things myself and when to let the LLM rip. My LLM-assisted code is better than my hand-written code; how could it not be? I'd have to be committing raw LLM output without even reading it to end up somewhere worse. If I did that, how much of a "real engineer" would I be?
All this is to say: even if all progress on AI halted today, it would remain the case that, after the Internet, LLMs are the most impactful thing to happen to software development in my career. It would be weird if companies like Supabase weren't thinking about them in their product plans.
Have you found any good resources on how to get a good process going? That would be an interesting read.
I have two main issues, first the tooling is changing so rapidly that as I start to hone in on a process it changes out from under me. The second is verifying the output. I’m at like 90% success rate on getting code generated correctly (if not always faster than I could do it) but boy does that final 10% bite when I don’t notice.
An aside, I think the cloud ought to make your (perhaps especially your) list. At least for me that changed the whole economy of building new software enterprises.
For “real work” done by a “real engineer”, I approach it almost exactly as you say.
For side projects/personal software that I most likely would have never started pre-llms? I’ll just go full vibe code and see how far I get. Sometimes I have to just delete it, but sometimes it works. That’s cool.
I don’t think it happens because the “vibe coders” are necessarily clueless, it can very much be a calculated risk or tradeoff.
Based on the “vibe coders” crowd I see on X, they are a superset of indie hackers with lower barrier to entry when it comes to coding skills and less patience for mediocre success. They seem to have the “go big or go home” mindset.
As long as they have a popular product, they don’t mind forking over some of their profit to OpenAI or a hosting provider. None of the Ghibli generator app creators complained about paying OpenAI… If the product is not popular, no outrageous costs, and the product will be abandoned anyway very fast.
Heroku is the OG "vibe coding" platform and, honestly, it was awesome. The first platform where you could deploy with one CLI command, sure it was expensive, but for years it was my favorite place to prototype or MVP stuff.
As far as cost, 200/month is nothing, but those are not the numbers we hear about when things spiral out of controll due to a ddos or sudden surge in popularity.
We serve thousands of customers off Supabase for $250/month including PITR. It’s a STEAL. Technical support is also responsive and knowledgeable. Outside of the auth SSR upgrade, it’s technically also been great to work with. Supabase is some of the best value we get from any vendor.
If you hire someone full time, you can do the math that way.
But the market rate for a freelance midlevel US-based engineer would be about double per hour what you'd pay a full-time employee of the same level, to account for taxes/PTO/health care/etc.
Would be happy to find something at half that, but there’s no work right now. Lots of ghost jobs and even upwork has twenty+ applicants fighting over $30/hr scraps.
I’m assuming that’s the fully burdened rate, i.e., salary, benefits, taxes, overhead, profit, etc that employees never see, even though they should.
I remember getting a sheet from an employer early in my career that fully broke down the cost of benefits and taxes and showed me the full cost of just my employment, not including overhead, profit, etc. it was rather eye opening because although I kid of knew it from accounting and finance, it never really impacted me quite as much before seeing the numbers.
If the system didn't structure it that way, everyone would know the numbers and protest. As it is people pay up for the most part, they even defend the concept of taxing, supposedly going to public services.
I think the real "vibe coders" in the end will be people like me. Making neat little personal projects that don't use many resources and can be relatively insecure because they don't matter much. I made wannawatchsomething.com with no knowledge of how it works and full knowledge that it's insecure and dumb. It still does what I want and I couldn't have done it a year ago so it's a net win.
Honestly I love that vercel and Superbase and these companies are making enormous amounts of money by only targeting JavaScript developers who have such a phobia of doing actual development work that they're willing to pay 10 times, 20 times the price for infrastructure to actually host their apps. They are not real developers and it's great to see them being squeezed and I'm glad that companies are making them pay through the nose and locking them in. They should suffer and face the consequences of their lack of skill.
“Real developer” label or not, it is now easier than ever to dream up, build, and ship an app. And at the end of the day, that’s all that matters—-what you ship. Just seems incredibly gatekeepy to devalue someone’s work based on the tools they used to build their product.
Yes, “vibecoding” still has issues (and likely will for the forseeable future). I’m sure the next decade will be an absolute boon for security researchers working with new companies. But you shouldn’t dismiss people based on their use of these tools.
And other commenters are right that these expensive infra tools can be replaced later when the idea has actually been validated.
The day before yesterday I got a technical assignment from a company I was interviewing with to build a Next.js app. Normally I would build it myself, but that day it just felt so tedious, so I gave Claude Code a try. To my surprise, with little to no guidance (probably cuz it was React), it built the app and it even looked very similar to mock-ups the company gave. I changed some things here and there and submitted the task. The whole thing was done in about half an hour. Yesterday they emailed me to schedule the next interview. I'm hooked.
We still do take-homes. You just design them assuming people are going to use LLMs. They're going to do that on the job anyways, so why tie a hand behind their back?
LLMs have also allowed us to significantly expand the scope of what we ask candidates to look at. Previously, we were constrained by time budgets (the candidate's, not ours) to a relatively small project, from which we had to read tea leaves; minor variations, objective but still small-bore, were determinative of how candidates ranked. Now we can drop a pretty ambitious project, which creates a lot of variation and room to demonstrate approach.
I turn 50 tomorrow and I love vibe coding. In the hands of an expert with decades of experience in all the internal corners of C, Python and Postgres I find AI tools to be miracles of technology. I know how to ask them exactly what I want and I know how to separate the goodness from the bullshit. If Supabase is bringing AI closer to the developer at the database level then that is a great thing.
Vibe coding is excellent if you have the experience to understand what the AI is churning out and then what to do with it.
The problem we have now is we have people who aren't engineers trying to make an app and they end up creating insecure and buggy messes, then struggle with figuring out how to deploy, or they end up destroying all their code with no recovery because they didn't know anything about version control.
Do you have any writings or materials that show your process in depth? I’m interested in learning from those who know how to really squeeze the juice out of these tools.
I think what they are saying is the 'secret sauce' to successfully vibe coding is being an expert with all the languages, frameworks and tools yourself.
Makes sense to me, vibe coding basically shifts your burden to specification and review, which are traditionally things a senior developer should be good at.
I agree with this - I hear a lot of hate towards vibe coding but my experience with voice dictation and using 20 years experience in the trenches and so being very specific telling the model what to do has been, well, refreshing to say the least.
I used to pride myself of knowing all the little ins and outs of the tech stack, especially when it comes to ops type stuff. This is still required, the difference is you don't need to spend 4 hours writing code - you can use the experience to get to the same result in 4 minutes.
I can see how "ask it for what you want and hope for the best" might not end well but personally - I am very much enjoying the process of distilling what I know we need to do next into a voice dictated prompt and then watching the AI just solve it based on what I said.
Part of what I think is rough is the framing of Postgres being "an alternative to Google’s Firebase" in the article. I mean, ok yeah they are both databases in a sense, but they are not the same thing at all. Firebase is a service and Postgres a database technology, and certainly not well-described as an alternative to Firebase by anyone competent in the industry. Lol.
I don't mean to demean "vibe coders" exactly either, but rather jumping on the hype train of using that term for your funding pitch. You're using AI to learn to become a software developer? Great! No problem with that.
But also — if you now have a database involved and you're handling people's data, you better learn what you're doing. A database provider pushing "vibe coding" is not a good look imo.
I'm sure that a lot of the l33t h4x0rs here think that Supabase sucks and is only for amateurs but I'll say that as a former engineer who's getting back into building fun side projects again, Supabase has been incredible and just what I wanted. It's my favorite new product that I've started using in the last year. I hope they build out an enormous TAM of people who don't want to live inside a terminal and make a ton of money.
I was looking for this comment.
A non-technical family member is working on a tech project, and giving them Lovable.dev with Supabase as a backend was like complete magic. No amount of fiddling with terminals or propping up postgres is too little.
We technical people always underestimate how fast things change when non-technical users can finally get things done without opening the hood.
Back in the day we'd call this phase a design and workflow prototype as to not have to deal with all the technical components until the actual flow and concept is done.
Feels we're skipping these steps and "generating" prototypes that may or may not satisfy the need and moving forward with that code into final.
One of the huge benefits of things like Invision, Marvel, Avocode, Figma, etc. was to allow the idea and flow to truly get its legs and skip the days where devs would plop right into code and do 100s of iterations and updates in actual code. This was a huge gain in development and opened up roles for PMs and UI/UX, while keeping developer work more focused on the actual implementation.
Feels these generate design & code tools are regressing back to direct-Code prototypes without all that workflow and understanding of what should actually be happening BEFORE the code, and instead will return to the distractions of the "How", and its millions of iterations and updates, rather than "What".
Some of this was already unfortunately happening due to Figma's loss of focus on workflow and collaboration, but seems these AI generation tools have made many completely lose sight of what was nice about the improved workflow of planning, simply because we CAN now generate the things we think we want, doesn't mean we should, especially before we know what we actually want / need.
Maybe I'm just getting old, but that's my .02 :).
there is no need to this tedious, boring phase which you miss, especially since it still requires a significant of coding effort (eg to stitch a backend to figma).
you can vibe code a fully working UI+backend that requires way less effort so why bother with planning and iterating on the UI separately at all?
anybody who actually knows what they are doing gets 10x from these tools plus they enable non-coders to bring ideas to the market and do it fast.
That's always been the justification to skip this phase :). Tools have just changed. One-person to small-team wonders that could code and build directly made the same arguments.
My point isn't to stitch things to Figma, that's abhorrent to me as well. My point is to not get bogged down on the implementation details, in this case an actually working DB, those tables, etc, but rather less fidelity actual full flow concepts that can be generated and iterated.
Then that can be fed into a magic genie GPT that generates the front-end, back-end, and all that good jazz.
If the effort to produce websites goes tends to zero, the value of websites will surely tend to zero. Either issues with security and maintainability will be a break on this tendency or we will get to a point where generating a custom website will be something trivial that will be done on demand.
The thing is, the cost of producing websites is already pretty low, but the value of websites mostly derives from network effects. So a rising flood of micro crud saas products will not be likely to generate much added value. And since interoperability will drive complexity, and transformer based LLMs are inherently limited at compositional tasks, any unforeseen value tapped by these extra websites will likely be offset by the maintainability and security breaks I mentioned. And because there is a delay in this signal, there is likely to be a bullwhip effect: an explosion of sites now and a burnout in a couple of years in which a lot of people will get severely turned off by the whole experience.
Don’t get me wrong, I love Supabase, but
> you can vibe code a fully working UI+backend
…is gonna bring a lot of houses crashing down sooner or later.
I couldn't agree more. "Vibe coding" is pretty cool, but it's not sustainable at least with with current technology. You're much better off being a knowledgeable developer who can guide an an LLM to write code for you.
One thing I will agree on though is that LLMs make it easier to iterate or try ideas and see if they'll work. I've been doing that a ton in my projects where I'll ask an LLM to build an interface and then if I like it I'll clean it up and or rebuild it myself.
I doubt that I'll ever use Figma to design, it's just too foreign to me. But LLMs let me work in a medium that I understand (code) while iterating quickly and trying ideas that I would never attempt because I wouldn't and be sure if they'd work out and it would take me a long time to implement them visually.
Really, that's where LLMs shine for me. Trying out an idea that you would even be capable of doing, but it would take you a long time. I can't tell you how many scripts I've asked ChatGPT or similar to write that I am fully capable of writing, but the return on investment just would not be there if I had to write it all by by hand. Additionally, I will use them to write scripts to debug problems or analyze logs/files. Again, things that I am perfectly capable of doing but would never do in the middle of a production issue because they would take too long and wouldn't necessarily yield results. With an LLM, I feel comfortable trying it out because at worst I'd burn a minute or two of of time and at best I can save myself hours. The return on investment just isn't there if it would take me 30 minutes to write that script and only then find out if it was useful or not.
LLMs are better search. Google burned down the house to keep itself warm and held off on LLMs until it was inevitable and are now pulling up ahead. This is the logical conclusion. LLMs will be monetized and enshitified by ads.
Soon, some free smart LLM code generators will stop generating certain outputs and instead suggest using commercial components that have paid for promotion.
So they say; I still haven't seen any high quality vibe coded software, and I'm pretty sure I never will.
I’ve seen tons of low quality successful software (concur anyone). Clearly being well built is not a requirement for success.
I'd split the difference and say the cons that low quality software creates only matter sometimes.
E.g. Concur is primarily feature-complete and will only ever need to evolve gradually.
So the drawbacks of being brittle, kludged-together, and incapable of making rapid feature changes doesn't really matter.
In some other products, that matters a huge deal.
So the tl;dr is, as always, optimize for the things that actually matter for your particular situation.
This can only be true of some products. Often there are a lot of concerns like privacy, white labeling, legal consequences that need to be considered _before_ you vibe code.
> We technical people always underestimate how fast things change when non-technical users can finally get things done without opening the hood.
This is good and bad. Non-technical users throwing up a prototype quickly is good. Non-technical users pushing that prototype into production with its security holes and non-obvious bugs is bad. It's easy for non-technical users to get a false sense of confidence if the thing they make looks good. This has been true since the RAD days of Delphi and VisualBasic.
Knowing the industry I'm pretty sure they will all push those AI prototypes to production - because they did the same with non-AI prototypes before. Now the question is once they inevitably pull in experienced folk for maintenance, refactoring and debugging, will it be easier or harder than working with that retired solo devs spaghetti codebase?
From looking at "vibe coding" tools their output is about the quality of bad body shop contractors. It's entirely possible for experienced devs to come in and fix it.
I think there's going to be the same problems as there are fixing bad body shop code. The companies that pushed their "vibe code" for a few dollars worth of AI tokens will expect people to work for pennies and/or have unreasonable time demands. There's also no ability to interview the original authors to figure out what they were thinking.
Meanwhile their customers are getting screwed over with data leaks if not outright hacks (depending on the app).
It's not a whole new issue, shitty contractors have existed for decades, but AI is pushing down the value of actual expertise.
> From looking at "vibe coding" tools their output is about the quality of bad body shop contractors.
Genuinely, it's a lot better.
I think this is just another correction. The software market is worth several trillion dollars now. Enterprise is pushing against the rise in labor costs. It will backfire as it did every single time and in a few years competent developers will be worth their weight in platinum.
For nearly 50 years now, software causes disruption, demand drives labor costs, enterprise responds with some silver bullet, haircuts in expensive suits collect bonuses, their masters pocket capital gains, and the chicken come home to roost with a cycle of disruption and labor cost increases. LLMs are being sold as disruption but it's actually another generation of enterprise tech. Hence the confusion. Vibe coding is just PR. Karpathy knows what he's doing.
> Non-technical users pushing that prototype into production with its security holes and non-obvious bugs is bad.
I beg to differ. Non-technical users pushing anything into production is GREAT!
For many, that's the only way they can get their internal tool done.
For many others, that's the only way they might get enough buyers and capital to hire a "real" developer to get rid of the security holes and non-obvious bugs.
I mean, it's not like every "senior developer" is immune from having obvious-in-retrospect security holes. Wasn't there a huge dating app recently with a glaring issue where you could list and access every photo and conversation ever shared, because nobody on their professional tech team secured the endpoints against enumeration of IDs?
What about users who sign up for these insecure apps and have their data and possibly their identity stolen due to the misplaced trust? That this already happens is no excuse to encourage even less security by encouraging novices to believe they are experts.
I agree it is great that more people can build software, but let's not pretend there are zero downsides.
This is a contrived situation. Most of the apps in discussion see little to no use and go dead soon after launch. The vast majority are collecting little data of negligible risk.
If a user is confident enough about a no name company that they give them enough info to make identity theft a possibility, it was only a matter of time before a spammer/phishing attack gets them anyway
> Most of the apps in discussion see little to no use and go dead soon after launch
That's not convincing. Of the apps that do get used, the vibe-coded ones will likely be unsafe.
> If a user is confident enough about a no name company that they give them enough info to make identity theft a possibility
That's completely unrelated. You can give a company very little information. Any of it being leaked is unacceptable. You can find a lot from an email, or a phone number.
People are taught, by CNBC, by suits, by hacks, that you can trust the apps on your commercials and it will be fine. It likely won't be, and your response is exactly why. Many of you are apathetic to the idea of doing right by people.
So people are manipulated, and some of them are elderly and don't even understand how computers work. This is reason enough to care about what they are exposed to, not say "let's burn it all down with shitty vibe-coding because users are dumb anyway."
We're supposed to be better than this.
My feeling is that this is similar to saying, "non-professional AirBnB hosts are a terrible security nightmare, and the fact that people are not much safer in regulated hotels is no excuse to encourage even less security by encouraging novices to play in the hospitality business".
I agree with you on the downsides.
AirBnB externality is not the safety risk for guests (although I personally ended up in some sketchy situations years ago, I don't use it anymore, mainly because:) the real externality is imposed on the inhabitants of popular tourist destinations.
There was a reason the industry was regulated, and circumventing these reasons with an app has been a net negative to society.
I don't think its bad enough
Even us entrepreneurially minded technical devs cut corners on our personal projects that we just want to through a Stripe integration or Solana Wallet connect on
And large companies with FTC and DOJ involved data breaches just wind up offering credits to users as compensation
so for non-technical creators to get into the mix, this just expands how many projects there are that get big enough to need dedicated UX and engineers
Having worked with it quite a bit I'm still not sure I really understand what it is, which sounds like a bizarre sentence but:
It's Postgres, but bundled with some extensions and Postgrest. And a database UI. But hosted and it runs locally also by pulling the separate parts. Running it locally has issues though, so much so that I found it easier to run a docker compose of the separate parts from scratch and at that point just carry that through to a deployment, at which point is there still a reason to use Supabase rather than another hosted Postgres with the extensions?
It's a bit of a confusing product story.
I really love supabase. And I’m glad they are getting some funding because I’m terrified they’ll get bought by Amazon or google and completely ruined.
The developer experience is first rate. It’s like they just read my mind and made everything I need really easy.
- Deals with login really nicely
- Databases for data
- Storage for files
- Both of those all nicely working with permissions
- Realtime is v cool
- Great docs
- Great SDK
- Great support peeps
Please never sell out.
The product story is that people want to build apps and naturally find themselves having to handle:
- remote state
- authoritative logic that can't run solely on the user's device because you can't trust it
- authentication
each of which is annoying when you're focused on building the user-facing app experience. Supabase solves all three without you needing to touch any infrastructure. The self-hosting thing just provides insurance that users are not completely locked in to their platform, which is a big concern when you're outsourcing basically your entire backend stack.
it's just a firebase competitor, that's based on postgres and you can run sql against it if you want.
It's also implied, and proven by some, that having access to Postgres means you can up and leave Supabase if you want to later. It won't be snap-your-fingers easy, but it's more direct than other hosted SaaS where you can't access your data or the schemas.
that "just" is carrying a lot of weight there
exactly this
You are not wrong that it’s a postgres + extensions. However, the tech market is very big now and that can sustain these valuations.
Not really a confusing story: it's a PaaS that wants to beat fears of becoming another Parse (https://www.willowtreeapps.com/craft/parse-shutdown-what-it-...)
Realistically 99% of the users would still be screwed if they ever shut down, regardless of if it's open (see: Parse)... but it gives people a some confidence to hear they're building on a platform that they could (strictly in theory) spin up their own instance of should a similar rug pull ever occur
They have also been giving back to postgres some of their extra work, and also their real time stuff i think is on erlang?
I agree you might prefer to choose the stack yourself, but for total n00bs and vibe coders supabase is a great start / boilerplate vs say the MEAN stack that was a hit 5y ago
I’ve been using Hasura and PostgREST for a few years now with real big production apps, in enterprise and in startups, and honestly the only problem with them is that backend engineers feel threatened.
They are great products that cover 95% of what a CRUD API does without hacks. They’re great tools in the hands of engineers too.
To me it’s not about vibe coding or AI. It is that it's pointless to reinvent the wheel on every single CRUD backend once again.
Experienced backend dev here who also uses Hasura for work at a successful small business. I think it's great at getting a prototype to production and solves real business problems that a solo dev could do by himself. As engineer #2 it's a mess, and it doesn't seem like a viable long term strategy.
I've only worked with Hasura, but I can say it's an insecure nightmare that forces anti-patterns. Your entire schema is exposed. Business logic gets pushed into your front end because where else do you run it unless you make an API wrapper. Likewise you can't easily customize your API without building an API on top of your API. You're doing weird extra network hops if you have other services that need the data but can't safely access it directly. You're pushed into fake open source where you can't always run the software independently. Who knows what will happen when the VC backers demand returns or the company deems the version you're on as not worth it to maintain compared to their radically different but more lucrative next version.
I think the people who write this off as "backend engineers feel threatened" aren't taking the time to understand the arguments they're hearing
I completely disagree.
Backends are far messier (especially when built over time by a team), more expensive and less flexible than a GraphQL or PostgREST's api.
> I've only worked with Hasura, but I can say it's an insecure nightmare that forces anti-patterns
Writing backend code without knowing what you're doing is also an insecure nightmare that forces anti-patterns. All good engineering practices still need to apply to Hasura.
Nothing says that "everything must go through it". Use it for the parts it fits well, use a normal backend for the non-CRUD parts. This makes securing tables easier for both Hasura and PostgREST.
> Business logic gets pushed into your front end because where else do you run it unless you make an API wrapper. You're doing weird extra network hops if you have other services that need the data but can't safely access it directly
I'm gonna disagree a bit with the sibling post here. If you think that going through Hasura for everything is not working: just don't.
This is 100% a self-imposed limitation. Hasura and PostgREST still allow you to have a separate backend that goes around it. There is nothing forbidding you from accessing the DB directly from another backend. This is not different from accessing the same database from two different classes. Keep the 100% CRUD part on Hasura/PostgREST, keep the fiddly bits in the backend.
The kind of dogma that says that everything must be built with those tools produces worse apps. You're describing it yourself.
> I think the people who write this off as "backend engineers feel threatened" aren't taking the time to understand the arguments they're hearing
I have heard the arguments and all I hear is people complaining about how hard it is to shove round pieces in square holes. These tools can be used correctly, but just like anything else they have a soft spot that you have to learn.
Once again: "use right tool for the job" doesn't mean you can only use a single tool in your project.
I've only played with these kinds of plug and play databases, but mixing and matching seems like the worst of both worlds. The plug and play is gone, because some things might me in API 1, some others in API 2, and maybe worst of all, their domains might overlap. So you need to know that the "boring" changes happen via the postgREST, but the fancier ones via some custom API. The APIs will probably also drift apart in small ways, making everything even more error prone.
> As engineer #2 it's a mess
As a long-time Hasura stan, I can't agree with this in any way.
> Your entire schema is exposed
In what sense? All queries to the DB go thru Hasura's API, there is no direct DB access. Roles are incredibly easy to set up and limit access on. Auth is easy to configure.
If you're really upset about this direct access, you can just hide the GQL endpoint and put REST endpoints that execute GQL queries in front of Hasura.
> Business logic gets pushed into your front end because where else do you run it unless you make an API wrapper
> Likewise you can't easily customize your API without building an API on top of your API. You're doing weird extra network hops
... How is an API that queries Hasura via GQL any different than an API that queries PG via SQL? Put your business logic in an API. Separating direct data access from API endpoints is a long-since solved problem.
Colocating Hasura and PG or Hasura and your API makes these network hops trivial.
Since Hasura also manages roles and access control, these "extra hops" are big value adds.
> You're pushed into fake open source where you can't always run the software independently
... Are you implying they will scrub the internet of their docker images? I always self-host Hasura. Have for years.
> I think the people who write this off as "backend engineers feel threatened" aren't taking the time to understand the arguments they're hearing
I think your arguments pretty much sum up why people think it's just about backend engineers feeling threatened - your sole point with any merit is that there's one extra network leg, but in a microservices world that's generally completely inconsequential.
I like PostgREST for some of it's use cases (views mostly), but the issue I have with it is that I don't often want a user to have direct access to the database, even if it's limited to their own data.
Mike can edit his name and his bio. He could edit some karma metric that he's got view access to but no write access to. That's fine, I can introduce an RLS policy to control this. Now Mike wants to edit his e-mail.
Now I need to send a confirmation e-mail to make sure the e-mail is valid, but at this point I can't protect the integrity of the database with RLS because the e-mail/receipt/confirm loop lives outside the database entirely. I can attach webhooks for this and use pg_net, but I could quickly have a lot of triggers firing webhooks inside my database and now most of my business logic is trapped in SQL and is at the mercy of how far pg_net will scale the increasing amount of triggers on a growing database.
Even for simple CRUD apps, there's so much else happening outside of the database that makes this get really gnarly really fast.
> Now I need to send a confirmation e-mail to make sure the e-mail is valid, but at this point I can't protect the integrity of the database with RLS because the e-mail/receipt/confirm loop lives outside the database entirely
Congratulations: that's not basic CRUD anymore, so you ran into the 5% of cases not covered by an automatic CRUD API.
And I don't see what's the dilemma here. Just use a normal endpoint. Keep using PostgREST to save time.
You don't have to throw the baby away with the bathwater just because it doesn't cover 5% of cases the way you want.
It's a rite of passage to realize that "use the right tool for the job" means you can use two tools at the same time for the same project. There are nails and screws. You can use a hammer and a screwdriver at the same time.
>You can use a hammer and a screwdriver at the same time
How do you balance the nail and screw? I'm serious, I'm trying to picture this, hammer in one hand, screwdriver in the other, and the problem I see here is the nail and screw need to be set first, which implies I can't completely use them both at the same time.
Perhaps my brain is too literal here, but I can't figure how to do this without starting with one or the other first
I have used them too, and I would say that at least for Hasura, performance can be poor for the generated queries. You have to be careful. Especially since they gate metrics behind their enterprise offering.
This is the same for any GraphQL backend. And even REST backends can be misused: I've fixed way too many joins-in-the-frontend that were causing N+1 queries in lists.
Elite hacker here. Supabase is excellent.
Not until/unless it has proper offline-first support. Check out InstantDB and Triplit.
How much are you paying per month?
So you're saying it's something like an updated version of Yahoo Small Business?
I’m with you, supabase is a fantastic product.
It is really good for getting started but ultimately our companies transition off of it.
World still needs a replacement for Microsoft Access on the web.
It’s been so long that new ideas are solving parts on the access spectrum without seemingly being aware of it.
Supabase and others would have a smaller footprint to add an app layer and reporting layer to their tool since it is data as the cornerstone not an afterthought
It's all fun and games until you need caching - something that comes at unspecified cost from when I looked into it.
That’s a lot of money.
What’s Supabase’s exit strategy? Are they sustainable long term as a standalone business?
You can also see how money is starting to chase “vibe coding” — as long as you say the magic words, even if your product is only tangentially related to it, you can get funding!
Reading the tea leaves, Series D means they opted for more funding vs IPO. They claim to have 2 million users, but they're open core so how many are paying? Maybe their books aren't looking that great. Wall street doesn't understand database vendors outside of "big data", so they're probably hoping for acquisition. Not sure who would buy them though, as PostgreSQL vendors are kind of a dime-a-dozen these days...
If lovable, bolt.new, etc kept integrating with them, that's a money maker without needing to do much sales. There's a wave of AI tools that require somehow save state and Supabase provides that. I'm absolutely amazed others haven't jumped in the same ship yet.
That definitely seems to be the play. Keep funneling in users from Lovable/bolt.new and keep building revenue or hope to be acquired if one of those vibe coding tools gets huge.
A lot are paying, including me for multiple projects. They have a pretty good offering. I used to use them for dev and prod, but now using neon for dev. Supabase still for prod. I had switched from mongo to supabase. I may switch to neon for prod but not in a rush.
They also offer so much more than just postgres. Though I use them only for postgres myself.
Since you use both supabase and neon, any particular strength or weakness to keep supabase for prod? I just moved my app to neon today (easy enough to test it!) and am enjoying the auto-scaling features and UI is great on neon. But I'm curious about how supabase stacks up.
it's an aggressive preemptive round, so i'd guess 2b/50 = 40M of revenue. Probably low margins since the free tier/ hosting postgres nature of the business.
> Not sure who would buy them though, as PostgreSQL vendors are kind of a dime-a-dozen these days...
Supabase defo has a much higher mindshare.
Sure but ultimately they're still just selling something that is already free and wrapping AWS. These business models aren't sustainable unless you trash your free product, which also isn't sustainable. Presumably they have a good deal with their cloud vendor, AWS I think, but I think its safe to assume they lose A LOT on their free products.
The whole premise of cloud hosting businesses is that people don't want to manage stuff themselves.
They end up managing stuff themselves anyway. Plus managing another kind of bills.
AWS seems to be doing fine.
> so how many are paying
This is like if Google Spanner were open sourced tomorrow morning: realistically how many people are going to learn how to deploy a thing that was built by Google for Google to serve an ultra-specific persona?
Maybe you might get some Amazon-sized whale peeking at it for bits to improve their own product, but the entire value prop is that it's a managed service: you're probably going to continue paying for it to be managed for you.
IMO it also depends on how the whole process is tied together.
I always loved Vercel for their easy hosting of Next.js with included CI/CD, but I recently switched to self-hosting - their pricing switched from a flat, worry-free $20/month to an unpredictable whatever-it-may-cost plus it sent me 10+ emails every single month about hitting some quotas that they introduced and I couldn't find a good way to stop that.
> Are they sustainable long term as a standalone business?
It's bananas to me that questions like these could be unanswered even 5 years after the business started. This possibly cannot be the most efficient way for finding new solutions and "disrupting" stale industries?
> It's bananas to me that questions like these could be unanswered even 5 years after the business started.
Those are rookie numbers, Discord is coming up on 10 years old and has made zero dollars to date, yet is supposedly considering an IPO soon.
It's very common for tech companies to go public without being profitable. As long as their growth is good and they have a reasonable story for how they will achieve profitability, then it typically makes sense. Of course every company is different and not all will reach their profitability goals post-IPO, but in many cases, it wouldn't make sense to wait for profitability before going public.
Also reasons they need to go public. Growth is costly.
Discord has the user growth to make up for it. It’s practically a household name on the level of Google now. I see many people boosting servers too, so clearly they are making a bunch of money.
Discord has a fairly successful subscription product that is generating tens of millions in revenue. They most certainly have made more than 0 dollars. Profitable? Less likely.
Yeah I meant zero profit, poor wording on my part.
The IPO is how (the shareholders) make money, by selling to bagholders.
What's really bananas is that your comment is just as relevant today as it would have been 15 years ago. It's been bananas for a while now.
Google bought firebase, so my guess is they are aiming for an Amazon or Microsoft acquisition.
To be fair, their $2B valuation is probably the most reasonable valuation we've seen in years. That doesn't negate the question of how they plan to turn a profit.
If they truly have 3.5 million databases, that's only ~$500 per database to recoup the investments, that doesn't seem to crazy. Companies like OpenAI or Twitter/X are never going to be profitable enough to cover what they've already spend/cost. Supabase could because the amount is so much lower and they have paying customer, but I'd like to emphasize the "could".
"ROI? Return on Investment?"
"No, Radio on the Internet."
and setting up postgresql on a simple VPS is so easy... You can literally ask Gemini 2.5 Pro or o3 or Sonnet 3.7 and do it in 15-30 minutes... Learned helplessness is really something and vibe coding is overrated imo: https://www.lycee.ai/blog/why-vibe-coding-is-overrated
i'm a bit brain fried right now but are you being sarcastic? typing out apt-get install postgresql is a lot less that 15-30min.
They are definitely creating some value. Managed database.
> What’s Supabase’s exit strategy? Are they sustainable long term as a standalone business?
Acquisition best case, Private Equity worst case.
Do you see Supabase going public on the stock market? Perhaps unless they do what Cloudflare done and are replicating AWS, it may be hard to see a stock market debut.
Could be wrong though.
Supabase is basically AWS Postgres under the hood. It's popular amongst hobbyists and small teams but I'm not sure whether any large teams actively use it. Once you're past the point of serious business, it's much more cost effective to host everything by yourself.
What is serious business? I think supabase can scale brilliantly, and it doesn't lock you in, if you have need for some special infra you can build and integrate it, I don't know but you could possibly even use FDW to access a postgres you run yourself.
Also they can't run on AWS postgres with all their postgres plug-ins AFAIK.
The point of "cheaper to host everything yourself" is a lot higher than what most estimate.
My only concern is that is supabase goes out of business or go evil you're gonna have a bad time, however everything is open-source
Serious business is when you need to maintain uptime and stability. Not just me, but a lot of folks on the Supabase reddit have complained often about the insane downtimes that we've experienced at times with the platform. I would 100% use it for prototypes and MVPs, but for production? Neither me nor a lot of others would touch it with a pole, even though I'm sure your experience might be different.
Supabase at is minimum providing a PostgreSQL server, pooler (they started on pgbouncer, how their own) and a PostREST API, and support, backup, logging etc. You can be doing serious business and not have the time/people to run these reliably self-hosted. They also provide Auth almost to the Auth0 level and Edge functions like Vercel, S3 like storage (sharing the db's permission system), and websocket/presence backed by Elixir. TBH they are a compelling value, at least for us.
Granted, Supabase does provide a lot under the hood, and it's excellent for whipping out a quick MVP, but I wouldn't count it production ready even today, simply because of the downtime I've experienced at times (even if their dashboards say otherwise). And it's not just an isolated case - a lot of folks on reddit complain with the same issues. Perhaps your treatment is different because of the high spend you guys have.
My former place ran a lot of RDS Postgres but also loved Supabase. It's more than just hosted DB because it has loads of value adds like web-based table editing, auth, edge functions, row-level security, easy hooks and triggers. We were capable of operating RDS but the cost of operations in dev hours was high. Supabase was super easy for moderate price and readily compatible with our RDS and Redshift.
I don't disagree. Supabase does provide a lot of functions under the hood that one would have to build out individually otherwise. I really like their lock-in model - you're not locked in with your data, but because of the extra functionality that they provide to your database.
Acquisition. All of these VC companies raise unsustainable levels of money in hopes of acquisition or IPO. Supabase seems to be leaning towards acquisition.
The greater fool strategy has worked well for unprofitable tech companies for decades, and shows no signs of slowing down
> Are they sustainable long term as a standalone business?
Was Meteor? They are exactly the same thing. And I really liked Meteor!
To me, the more money pouring in, the better. That said:
https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcRCVKYR...
(The Silicon Valley Economy cartoon)
What an absolute joke. Their exit strategy is presumably to keep chasing the high and find more ways to integrate AI. The era of building good software for fun and profit is coming to an end.
I think their product is sound, they build essentially a backend as a service platform on open-source software. That doesn't make it easy to run at scale, so you probably wanna use their paid offering unless you plan to hire a lot of staff to maintain it, but it is possible and they support small scale dev envs
100% this; we have a 4 digit monthly spend. I guess I will double-check when we reach 5 digits, but I can't afford my own time self-host it yet.
Have you thought about possibly hiring a junior engineer to make the transition possible? I don't know what your use case but creating a solution that fits your needs would be worth it IMO. You're already spending a years worth of dev salary which can get you plenty of good talent around the world with.
>Supabase is currently used by two million developers who manage more than 3.5 million databases. The startup supports Postgres, the most popular developer database system that’s an alternative to Google’s Firebase. Supabase’s goal: To be a one-stop backend for developers and "vibe coders."
How many of those users are paid. You can sign up for free without a credit card.
It's cool, for certain use cases. I ended up trying it for a few months before switching to Django.
If you ONLY need to store data behind some authentication, and handle everything else on the frontend, it's great. Once you need to try some serverside logic it gets weird. I'm open to being wrong, but I found firebase phenomenally more polished and easier to work with particularly when you get to firebase functions compared to edge functions.
Self hosting requires magical tricks, it's clearly not a focus for them right now.
I hope they keep the free tier intact. While it's not perfect, if your in a situation where you can spend absolutely no money you can easily use it learning ( or for portfolio piece).
> Self hosting requires magical tricks
Has anything changed recently? ~1year ago I installed a local instance (that I still use today for logging LLM stats) and IIRC all I had to do was "docker compose up". All the dockers are still starting for me at boot from that 1yo install, to this day. (I use it on 127.0 so no SSE & stuff, perhaps that's where the pain points are? Dunno, but for my local logging needs it's perfect).
Hosting it on an actual server with a URL is not a fun experience. You need to generate a specific type of string to get it to work.
This isn't documented anywhere. Deep deep in their GitHub issues you'll find a script for generating this magic string which needs to be set as an environment variable.
See https://github.com/supabase/supabase/issues/17164#issuecomme...
Looks like it is just an issue of correctly making a jwt token, if you are not using their client libraries, but you can also just do it via their docs https://supabase.com/docs/guides/self-hosting/docker#generat... now (not sure how long you've been able to do in the docs)
Sure "it's in the docs" but last time our devops tried the compose file with ~10 or so services it took several days of fiddling with all sorts of different issues. It is just not made for selfhosting at all. It can be so much simpler but JS devs like it different.
It’s normal Postgres. There’s no need to handle everything on the front end. The tutorials nudge you to learn RLS and use their SDKs for the client, but you can write perfectly normal server side code as well.
Yeah I’ve ran a small project where I just did everything with the “service account” credentials which operates like a normal Postgres connection.
If you're not supporting users, it's fine.
But if you usecase involves Supabase auth, using a service account to bypass RLS is kind of like hardcoding connection strings.
You can use both properly and together.
The service account should only be accessed on the service.
If using Auth+Server, you can check the verified user identity via Auth JWTs (or something, see the docs).
Yeah, don't use the server connection on the client, but they have many warnings against that.
Yeah, it's a bit wonky, especially when you are dealing with configuring specific combination of supabase/deno/typescript features (e.g. stage 2 vs stage 3 decorators)
I literally use it because it's a free hosted postgres database. I just connect via connection string on my backend and run the queries there.
How is Djanjo a replacement for Supabase?
For my current project I basically need a backend server for processing some basic game logic.
I had done something similar in Firebase and it was easy. Supabase wasn't straightforward here. It got to a point where I'm sure I could eventually get it working, but I also think I'm outside the expected usecase.
Django is much more flexibility in this regard.
Is the new valuation multiplier number of developers on platform instead of revenue? Valuing at $1000/developer is kind of insane. Valuing at $570/database is also nuts. It's a cool product but I hope the founders can find a win in what must be a pretty cramped cap table.
Their 2024 revenue was estimated at $16.8M [1] and $10.5M in 2023. If you extrapolate that growth rate +1 year you can assume it's now $26.9M. Another source estimates it at $15M in 2025 [2].
So if you assume their revenue is in that range, you're looking at 66x to 133x ARR multiple. In today's market that's quite a big markup. Standard SaaS right now is probably more like 5-15x. AI is a lot more (but Supabase isn't AI). But they are a key leader in their market, so probably get a meaningful bonus for that. And I'm sure a lot of big industry investors were competing against each other for the Supabase deal, so that definitely would have helped valuation too. Also, at their maturity today, they are probably showing some great success signing big enterprise deals and telling a story about how that will grow.
That being said, those factors alone don't answer 66-133x. Perhaps Supabase's strongest angle is their opportunity for product-led growth:
- They have a huge number of people on a free tier
- The growth rate of free tier users might be accelerating
- The conversion rate of free tier users to paid users might also be increasing
- They're adding more things that people can pay for, increasing LTV of customers. e.g., for my business, we probably 20x our Supabase cost in the last 6 months - most of that is due to our growth but also there are a lot of things we can buy from Supabase beyond compute.
So I would assume, in addition to the above, they're telling a story about their actual revenue growth rate will accelerate meaningfully because of all of these factors working together.
Lots of assumptions in here, but you can start to see how a lot of different factors + a hype multiple could lead to such a valuation.
[1] https://getlatka.com/companies/supabase.com#revenue
[2] https://leadiq.com/c/supabase/5ed1e4778a998f161ef62998
Their series C was in September 2024!
Congrats team!
I was a speaker in a local Supabase event just few weeks ago, https://shorturl.at/JwWMk. We had a local event in Abuja, Nigeria. There we promoted their Launch Week 14 series, highlighting new features from Supabase. In reality, it became an event to show people how to bootstrap a quick backend for their SME business in a weekend.
Good for them. They get so much hate online I think because it’s such a “why didn’t I think of that” idea. People are jealous of their success I feel.
Nice, I gotta admit, I like seeing new stuff get funded like this - but man, whenever I read about crazy big rounds, I start wondering how these companies ever make real money vs just chasing hype till the next exit. Always makes me think about what actually lasts long term besides just burn and churn.
$2B valuation at $16M revenue sounds nuts..
My prediction: They're banking on a big exit to OpenAI or Claude as the defacto backend for an AI IDE.
They're the only big alternative to Firebase, and Firebase just got pulled into Google AI Studio.
where did you get 16M revenue?
Personally think that supabase and vercel are giving AWS a run for their money, and will start to eat into their market share. The developer experience is far superior to AWS, in a way that cannot be argued away by AWS handling "more complexity and scale". Supabase/Vercerl products are superior to large cloud providers, and while they target a narrower aspect of the tech stack, and to smaller customers, they will expand into more enterprise as their users grow.
AWS needs to get their act together and start prioritizing developer experience
Also, supabase is looking like the go to database for ai created apps. Which will be a major tailwind
I think large companies/gov contractors will still prioritize compliance and control systems over DX.
And I believe both Supabase and Vercel run all their services on AWS anyways, so AWS gets paid no matter what.
Both use AWS under the hood
point still stands
How are they giving AWS a run for their money when they use AWS for their own service? AWS profits from Supabase growth.
AWS is unusable in alot of cases.
But it doesn't matter because they are transitively being used by virtue of Vercel/Supabase being on them. They could definitely charge more, that's for sure, but it's not like they don't get a pretty penny from others building atop them
Right but they run on AWS. So AWS is making money every time someone uses Vercel or Supabase
I've been testing out PostgREST, RLS, and PL/pgSQL functions for a new app and I'm not wild about it. It's pretty complicated to grok the permission model, it's awkward to load up your DB with logic, and the LLMs kinda suck at reliably generating working queries, policies, functions, etc. So I'm not really convinced it's ideal for 'vibe coders'.
I built a 'serverless' platform that's similar, smaller but more focus on low-code than Supabase https://saasufy.com/
All the components are declarative HTML and update in realtime. Similar concept as HTMX but doesn't require any backend code. You can still implement complex UX, authentication, access control and filtered views (indexing and all).
I built this app with it over a few months as a weekend project: https://www.insnare.net/app/#/onboarding/country/All
A 'serverless' platform that's "similiar". But not open source, and uses a token to 'own a piece of it'?
Care to explain what the idea behind this is and where you see the similarities?
Raising a very late funding stage in the private market to avoid an IPO in the violent public market I presume? I’m guessing this is the new startup norm for a while, along with reducing burn or exiting at a big discount if not just dying
Nevertheless congrats to the Supabase team!
Execs cashing out some of their shares to buy houses in Tahoe while the stock market is in the red.
I always felt like they’re the database dogman would use.
As someone with a seven year old, I appreciated this. Thank you.
Time to vibe code the data into "That supa awesome database over there"
I'm convinced the only profitable market for these DB companies is enterprise.
Either that or they need to add features and products alongside the DB to essentially replace the likes of Vercel.
Having said that Supabase is probably the best 'cloud DB' I've played around with so hope they succeed.
Why do they bring up vibe coding here. They are just a firebase alternative and Google has way superior ai code gen tools
Supabase is quite well integrated with vibe coding tools - cursor, replit, v0, etc. I agree that Firebase is a superior well-integrated product, but IMO the vibes are on Supabase side.
My guess is that they got that insane overvaluation because they sold themselves as an AI company
My humble guess is they sold themselves as the enabler for the AI vibe coding "revolution"
As an observation, the quotations and the article itself doesn’t appear very thoughtful or well reasoned. It sounds more like a VC wanted a big deal (or at least to represent it as such) and made a bet, invited his buddies(?) along, and the company has seen a recent bump from “vibe coding”. Some allusion to Larry Ellison. Very light on information to form an opinion…
I'm sure the team might not be expecting this level of valuation. These trends won't last long. Make hay while the hype shines. Who knows how soon people would forget that there was something called vibe-coding and a back-end development.
And the decline of a good product begins because it will be all about profits for board members going forward...
To be fair this is a series D
I've been a lukewarm user of Supabase for my side projects. Unfortunately the amount of work to get off of it has been too high for me to leave.
The major issue is - cost. It is way more expensive than I realized as they have so many little ways they charge you. It's almost like death by thousands of paper cuts. My bill for my app with just a few thousand users was $70 last month.
I do like the tooling and all, but the pricing has been very confusing.
Kind of the same feeling, I don't use all of services they offer either and when I looked at self-hosting, it all seemed kind of heavy and fragile to self-host. I ended up replicating the parts I used with a small API layer connected to a managed postgres db for a tenth of the cost or something. I'd say it's pretty handy for prototyping but not sure I'd want to build a business on the back of it.
> just a few thousand users was $70 last month.
Few Thousand!?! Sound very reasonable to me. Monetize just two of those users at $35 per month and your server costs are covered. Or run it yourself, there's a lot of moving parts but it's all open source.
> Few Thousand!?! Sound very reasonable to me. Monetize just two of those users at $35 per month and your server costs are covered
That's one way to look at it, but compared to any other way to run a server, it's objectively terrible. You can serve that many users with a $5 box.
So put Supabase on a $5 box or build out your own replacement backend if your time is that cheap.
Not to be negative, but that seems super huge for its worth. How can you even recoup that with barely anyone paying? Founders trying to exit is comprehensible, but how can VCs sign this deal?
I don't really understand supabase. Everyone talks about how great it is, but it's just slow expensive postgres? I must be missing something.
I'm a happy Supabase customer. It's not just postgres, it's postgres + auth + functions + storage all in one place, with a very good local Docker environment and a variety of nice client SDKs.
It's a quick/convenient way to get a "full" backend up and running. The other option (Firebase) can't be self-hosted and has some absurd pricing footguns.
It has a nice web interface, comes with support for auth (which almost all apps need), and has a couple other nice things like storage buckets and serverless functions. When doing a hobby project, I don't want to rent a server from hetzner, set up SSH, connect, figure out how to install postgres, make sure everything is configured and so on. It's much nicer to click one button in the supabase UI and have everything I need
It's slow expensive postgres _with_AI_
Retool uses Supabase for its out of the box database setup
> The startup supports Postgres, the most popular developer database system that’s an alternative to Google’s Firebase
I've always taken issue with branding Supabase as an alternative to Firebase. Firebase is a PaaS whereas Supabase is more of a BaaS.
can you explain more?
While Supabase does support many equivalent Firebase features like a database, auth, functions, storage, etc, Supabase doesn't support website hosting whereas Firebase does. Supabase is mainly for backend stuff (BaaS) whereas Firebase is for both frontend and backend (PaaS).
To get the equivalent functionality of Firebase, you'd need to add something like Netlify or Vercel to pair it with your Supabase backend.
I love Supabase, but this seems to be the sign they will not be available in next 2-3 years. Or get acquired by someone.
I don't think this is a good sign.
The only that hasn't changed in this industry is that engineers apparently can't stand new, green beginners getting into the field.
"They ship buggy, insecure messes" "They don't know how to fix what AI gave them" etc etc etc
Right. Like that same thing hasn't been happening literally during the entire existence of programming. I, for one, welcome the vibe coders. I hope it grows their interest in the field and encourages them to go deeper and learn more. Will some be lazy and not even try? Of course! Will some get curious and learn the ins and outs? Absolutely.
The thing I believe least in this article is that the investor is betting their career on this investment. Maybe they think that, but they’re still multi-millionaires making bets with other people’s money. They’ll be fine.
Congrats to Anthony and the team! What an amazing milestone.
Congratulations to anyone who can raise $200m, let alone have the VCs fly around the world to make it happen.
please fix the drizzle- --> supabase integration.
What's even the point of using drizzle with supabase? Isn't the entire point of supabase PostgREST and convenience tooling around it?
I don't quite understand the high valuation. I use Supabase for the database and file storage (legacy choice). But those seem interchangeable. Are people using many many more features, if so how?
I'm from the future again, and I predict that either Replicate or fal.ai will get acquired by one of these so-called 'Vibe-coding' companies. Supabase included.
When I see valuations like this, they are overvalued until they use that money to acquire another company for a total addressable market expansion.
I think FAL is already too big for a "viber" company to buy. $55M ARR and 10%+ EBITDA. They're either going to IPO or get acquihired by a big 4.
FAL team is the real deal, they will absolutely not get acquired by any of these other "vibes" companies.
It's really sad that nobody is actually trying to be a real company these days. When Steve Jobs was asked that if he will ever sell Apple, he physically recoiled and scolded the person asking such things. Nowadays, all of these companies are being set up just to be acquired. Nobody has any ambition anymore to be long lasting, profitable and an actual business.
So basically the nobody wants to work of the business longevity world? What is the actual difference? If their goal is profitability and eventually retirement why not do it on an accelerated timescale and exit with millions in their 30s instead of hundreds of thousands in their 70s?
The new age is back
Another heroku in the making. Also - where do these millions go? All into engineering?
Crazy that devs choose supabase and vercel when Google Cloud is right there.
Google were late to the game but they've built perhaps one of the easiest cloud platforms to work with.
Do you use firebase?
This seems like another one of those VC companies which gets pushed by the VCs to the other companies in their portfolios as a probable integration point.
The whole growth of vibe coding really did help them because I don't think actual developers use it because putting things like functions in the database and authorization in the database is something that we learnt a few decades ago is a bad idea.
So I would guess they are used by massive amounts of developers who are new to coding or do not fully know how to code, but are becoming developers and who love the free databases Supabase provides.
Would love to know what is their actual revenu.
> The whole growth of vibe coding really did help them because I don't think actual developers use it because putting things like functions in the database and authorization in the database is something that we learnt a few decades ago is a bad idea.
Why are those things a bad ideas? You could be right but if you insist on making value judgements without explanation or elaboration, you're going to sound like a whiny old crank who is scared of becoming obsolete.
What I mean is that, you know, we tried all of these things back in the day. Like putting your business logic in a database was something that was tried. It's not that the first time that these ideas have shown up. And they were basically tried and tested and found faulty. But now there's this whole new generation of sub-bar developers. And they are being fooled into thinking that all of this works because they just want to get their hands dirty. I guess when normies flood the scene the wolves make a killing.
A recent project on HN declared Supabase dead: https://www.isthistechdead.com/supabase
While the funding is impressive, I haven’t come across too many people touting Supabase or using it in production.
We're currently using supabase in production. I was already planning to leave them. I feel like it's close to good, but still really does have a _lot_ of bugs. It really doesn't feel like it's been tested thoroughly, and the documentation, while present, leaves a lot to be desired.
My experience of supabase really demonstrates to me that the ideals of all of the postgres layer technologies - postrest, realtime via wal, jwt auth in the db -, just don't make for an easy experience. It all works (mostly) but I find it more annoying than useful and have to work around it more often than I'd like. I suppose I'm old school, but just building the things that one needs is often more robust and less work than trying to plug into what they've provided.
I really don't know what they're going to do with a series D. It seems they now _have_ to go for a high-value exit, but I really don't see which company would provide that exit.
I had similar experience with everything missing that final attention to detail and polish and having to write issues and ask other how they got past certain problems. I ended up switching to Pocketbase, and while it is not a complete or drop in replacement hosted service, it is light weight and approachable to feel comfortable that it can scale and be more stable long term.
You should try Convex. It’s a better, higher level abstraction than Supabase.
We had a pretty bad experience with it on a NextJS app and had to gut it out (I decided not to charge the client for this multi-week refactoring effort to pull out Supabase since it was upon my recommendation that they went with Supabase, so that decision actually cost us thousands of dollars).
It is good to get started and no doubt useful for simple CRUD apps. But once you want to start doing more complicated stuff, a lot of the RLS primitives become very hard to maintain and test, for example. You could say that that's Postgres's fault, but Supabase strongly pushes you in that direction.
The tooling, while looking quite polished, just felt pretty half baked along with docs (at least a year ago when we pulled the plug). Try to implement even a halfway complicated permissions scheme with it and RLS and you are in for a world of hurt and seemingly unmaintainable code.
So we ditched Supabase Auth for AuthJS, and are using vanilla postgres with Prisma. That's worked well for us. All the tooling is relatively mature, it's easy to write tests, etc.
Maybe if AI is writing some of the code, it might get easier, but for now, I'm avoiding Supabase like the plague until I see a project that's relatively complex that's actually easy to maintain.
Agree, also the logging is very limited across the board,, database functions are impossible to debug without good logging, I think a lot of people don't realize its not a direct copy of Postgres, for the cloud version several Postgres functions are disabled by default.
Nice to see others using Prisma in the wild! Prisma is the first ORM I've ever used (after decades of trials) that I actually enjoy using.
Also HN consensus: Word Press with 40%+ of the web is dead, Firefox with less than 5% of desktop and 0% of mobile is thriving and AI is a fad.
Imagine being a Postgres developer and getting 1% of that while being the core technology powering all these startups
Most of the heavy hitters in Postgres development are already sponsored by companies to do their work. Supabase has at least one Postgres committer on staff for instance.
[flagged]
i dont want to listen to uninformed "AI-generated" music any more than i want to support uninformed "AI-generated" code
Sure. But it's not really about you. I wouldn't break into someone's house and tell them they painted their house an uninformed color.
If something sucks or won't scale, it will sort itself out in the market.
> I wouldn't break into someone's house and tell them they painted their house an uninformed color.
but if you, an experienced professional painter, were hired to repaint somebody's house, and that person used bad paint (or something else entirely, like... i dunno, shellac, or white glue), your scope of work changes from "light prep and paint" to a much more involved job of undoing the previous work (or mistakes) and then getting to the original scope of work.
> If something sucks or won't scale, it will sort itself out in the market.
i truly don't mean this in a derogatory way but that sounds incredibly naive.
I don’t think we are going to agree on this.
At the end of the day we are really just disagreeing on how seriously we should take AI-ification of software engineering.
I take a lighter stance that it’s mostly harmless and has a lot of educational positives. I guess you foresee a lot of potential harm.
I do appreciate people who are willing to be the canary in the coal mine. I just find that a lot of the criticism appears to be more gatekeeping than legitimate worries.
I think folks can gatekeep because of legitimate worry :)
A lot of comments seem to take issue with Supabase being aligned to Vibe-coding, which is understandable, but I do think it's related! I think vibe-coding could very well be a real market-force, and as best as I can tell it exists within a specific type of tech circle. No one is vibe coding Elixir, no one is vibe coding Rust, people are vibe coding React + Node/Python, and more specifically I think people are watching streamers and YT videos on hot-new-frameworks-near-you and want to try them out! Supabase is absolutely a darling of this tech-circle, and I think a few other companies could be as well:
1. oxc (oxlint)
2. vercel
3. fly.io
probably more! and more every day
>no one is vibe coding elixir
I did :) I made a browser-based MMO with Phoenix to test out liveview and learn the language: https://shopkeep.gg
And it was pretty annoying. Elixir doesn't really lend itself to vibe coding due to namespacing and aliasing of modules, pattern matching, all without static typing (I know, Dialyzer...). It also struggles to understand the difference between LiveComponents and LiveViews, where to send/handle messages between layers.
Without references to filenames, the agent perpetually decides "this doesn't exist, so I'll write it :)". I found it to be pretty challenging before figuring out I could force the agent to run `mix xref callers <Some.Module>` when trying to do cross-module refs.
(caveat: this was all with claude 3.5 sonnet)
Rust is an excellent vibe target because it won't compile unless it works. It may not do exactly what you think but that's what vibe testing is for.
I think the idea that "if Rust compiles then it works" is very untrue, especially for non-trivial software.
New frameworks don't align with vibe coding at all, because they are poorly represented in OpenAI, Anthropic & Google's datasets.
But you can just give them a prompt to teach them. Case in point, Transact has a prompt that you can put into an LLM that enables it to one-shot some simple programs:
https://github.com/dbos-inc/dbos-docs/blob/main/docs/python/...
I vibe coded parts of https://gisformat.com/ backend in Rust in order to learn about Rust. So I think you are wrong there.
Oof this bit is rough!
> The startup supports Postgres, the most popular developer database system that’s an alternative to Google’s Firebase. Supabase’s goal: To be a one-stop backend for developers and "vibe coders."
Enabling more people to make cool things themselves isn't necessarily a bad thing.
It's all fun and games until you realize the cool thing you made is locked into an overpriced cloud stack that will bleed you dry. "Vibe coders" seem to invariably end up offloading as much heavy lifting as possible to ultra-expensive PaaS/SaaS vendors, and those vendors are encouraging it (e.g. Vercel v0).
It's the right trade-off, IMO. Most projects won't succeed and the ones that do can be refactored - by real engineers if needed.
You make it sound like it’s a binary situation where in one case it doesn’t matter and in the other it’s No Problem. I think both are wrong.
An unsuccessful project might be unsuccessful because it got eaten by costs before it became successful.
A wildly successful project is risky to migrate.
The YC crowd seems to think the only worthwhile businesses are worth $1 billion+ otherwise why bother
I strongly disagree with this.
Most startups fail. Optimizing for getting revenue is more important than optimizing cost in the beginning.
If you get revenue you can solve the cost problem. If you don’t, it doesn’t matter.
Anything that gives you more shots at the goal is a win in a startup.
If you are trying to commercialize something, a popular project with bad margins is a better spot to be in than an unsuccessful project with good margins. If it's a personal learning project, that might not be the case.
Counterexample:
https://x.com/levelsio/status/1730659933232730443
I don’t think that’s a counter example. If hood maps shows a lot of potential then the $11k is something to figure out.
If not, then it’s poor price controls.
IIUC Pieter Levels talks a lot about not prematurely optimizing engineering solutions because most ideas will flop.
These cloud back and stacks are very cheap at low volume and honestly I expect them to remain so or even go down in price.
I've seen many colleagues bootstrap something - even if they're not themselves very technical - because they've leveraged these well integrated low cost platforms.
I do think it’s binary. The project either shows potential to meet your goal or it doesn’t.
I think it’s rare that fails to show potential because of the underlying technology that’s chosen.
Sure, Vercel is relatively expensive. But I just don’t see how you’d throw in the towel because the costs are too high without first evaluating how to lower them.
If you’re saying that the evaluation is likely to show that you’re stuck - I have never seen that be the case personally.
I’m a “real engineer” and I don’t want to do all that until I have to.
As a "real engineer" you'd likely use LLMs differently. I save my conversations, have chats and codebase exegesis summarized into .txt files, and constantly refactor LLM output. I have an increasingly reliable sense of when to dip in and write things myself and when to let the LLM rip. My LLM-assisted code is better than my hand-written code; how could it not be? I'd have to be committing raw LLM output without even reading it to end up somewhere worse. If I did that, how much of a "real engineer" would I be?
All this is to say: even if all progress on AI halted today, it would remain the case that, after the Internet, LLMs are the most impactful thing to happen to software development in my career. It would be weird if companies like Supabase weren't thinking about them in their product plans.
Have you found any good resources on how to get a good process going? That would be an interesting read.
I have two main issues, first the tooling is changing so rapidly that as I start to hone in on a process it changes out from under me. The second is verifying the output. I’m at like 90% success rate on getting code generated correctly (if not always faster than I could do it) but boy does that final 10% bite when I don’t notice.
An aside, I think the cloud ought to make your (perhaps especially your) list. At least for me that changed the whole economy of building new software enterprises.
Yeah I think it depends on what I’m doing.
For “real work” done by a “real engineer”, I approach it almost exactly as you say.
For side projects/personal software that I most likely would have never started pre-llms? I’ll just go full vibe code and see how far I get. Sometimes I have to just delete it, but sometimes it works. That’s cool.
Regarding refactor: My understanding is that vibes codebases are effectively write-only due to their structural incoherence.
That’s my understanding also. I think a project that’s vibe coded could easily be recreated by a programmer using coding assistance.
Agreed. And you can take a lot of the supabase code and try and host yourself if you get there.
I don’t think it happens because the “vibe coders” are necessarily clueless, it can very much be a calculated risk or tradeoff.
Based on the “vibe coders” crowd I see on X, they are a superset of indie hackers with lower barrier to entry when it comes to coding skills and less patience for mediocre success. They seem to have the “go big or go home” mindset.
As long as they have a popular product, they don’t mind forking over some of their profit to OpenAI or a hosting provider. None of the Ghibli generator app creators complained about paying OpenAI… If the product is not popular, no outrageous costs, and the product will be abandoned anyway very fast.
PostgREST is open source, and so is the rest of Supabase.
Migrating from it is not that hard so far. I did it on an afternoon for a customer.
Also a couple friends are running the open source version in their own containers.
Maybe there are (or will be) cloud only features, but for the basic service there isn’t as much lock-in as something like AWS.
Heroku is the OG "vibe coding" platform and, honestly, it was awesome. The first platform where you could deploy with one CLI command, sure it was expensive, but for years it was my favorite place to prototype or MVP stuff.
Okay? Let's say it's $200/mo or something for a moderately popular app. An engineer starts at $200/hr, so you're still saving a ton.
With Vercel/Netlify, you're paying for ease of use. For a lot of people, that tradeoff is worth it. Not everything can be free.
Is that 200 a made up number?
https://money.usnews.com/careers/best-jobs/computer-programm...
Do you have a better source for your number.
As far as cost, 200/month is nothing, but those are not the numbers we hear about when things spiral out of controll due to a ddos or sudden surge in popularity.
We serve thousands of customers off Supabase for $250/month including PITR. It’s a STEAL. Technical support is also responsive and knowledgeable. Outside of the auth SSR upgrade, it’s technically also been great to work with. Supabase is some of the best value we get from any vendor.
Starts? With 40h/week, that's basically a $400k annual salary. This is certainly not the start price of an engineer.
If you hire someone full time, you can do the math that way.
But the market rate for a freelance midlevel US-based engineer would be about double per hour what you'd pay a full-time employee of the same level, to account for taxes/PTO/health care/etc.
Would be happy to find something at half that, but there’s no work right now. Lots of ghost jobs and even upwork has twenty+ applicants fighting over $30/hr scraps.
>An engineer starts at $200/hr
Starts?!
I’m assuming that’s the fully burdened rate, i.e., salary, benefits, taxes, overhead, profit, etc that employees never see, even though they should.
I remember getting a sheet from an employer early in my career that fully broke down the cost of benefits and taxes and showed me the full cost of just my employment, not including overhead, profit, etc. it was rather eye opening because although I kid of knew it from accounting and finance, it never really impacted me quite as much before seeing the numbers.
If the system didn't structure it that way, everyone would know the numbers and protest. As it is people pay up for the most part, they even defend the concept of taxing, supposedly going to public services.
None of these are expensive if they help you ship products that solve problems that customers are willing to pay for.
I think the real "vibe coders" in the end will be people like me. Making neat little personal projects that don't use many resources and can be relatively insecure because they don't matter much. I made wannawatchsomething.com with no knowledge of how it works and full knowledge that it's insecure and dumb. It still does what I want and I couldn't have done it a year ago so it's a net win.
> the cool thing you made is locked into an overpriced cloud stack that will bleed you dry
Not necessarily applicable to vibing with Supabase specifically, right?
There are several ways to host Supabase on your own computer, server, or cloud.
https://supabase.com/docs/guides/self-hosting
Honestly I love that vercel and Superbase and these companies are making enormous amounts of money by only targeting JavaScript developers who have such a phobia of doing actual development work that they're willing to pay 10 times, 20 times the price for infrastructure to actually host their apps. They are not real developers and it's great to see them being squeezed and I'm glad that companies are making them pay through the nose and locking them in. They should suffer and face the consequences of their lack of skill.
“Real developer” label or not, it is now easier than ever to dream up, build, and ship an app. And at the end of the day, that’s all that matters—-what you ship. Just seems incredibly gatekeepy to devalue someone’s work based on the tools they used to build their product.
Yes, “vibecoding” still has issues (and likely will for the forseeable future). I’m sure the next decade will be an absolute boon for security researchers working with new companies. But you shouldn’t dismiss people based on their use of these tools.
And other commenters are right that these expensive infra tools can be replaced later when the idea has actually been validated.
Let’s not raise ourselves up by saying that other developers are no true Scotsmen based on what language they prefer.
I don't necessarily think this is a bad goal, but the term "vibe coder" is almost certainly considered derogatory now.
The day before yesterday I got a technical assignment from a company I was interviewing with to build a Next.js app. Normally I would build it myself, but that day it just felt so tedious, so I gave Claude Code a try. To my surprise, with little to no guidance (probably cuz it was React), it built the app and it even looked very similar to mock-ups the company gave. I changed some things here and there and submitted the task. The whole thing was done in about half an hour. Yesterday they emailed me to schedule the next interview. I'm hooked.
Sadly it seems the days of take home interview assignments are numbered. I much preferred them to live coding assessments when given the option.
We still do take-homes. You just design them assuming people are going to use LLMs. They're going to do that on the job anyways, so why tie a hand behind their back?
Same. We just ask people to explain what they delivered, and this is 1000x better and more interesting than any quiz or whiteboard.
LLMs have also allowed us to significantly expand the scope of what we ask candidates to look at. Previously, we were constrained by time budgets (the candidate's, not ours) to a relatively small project, from which we had to read tea leaves; minor variations, objective but still small-bore, were determinative of how candidates ranked. Now we can drop a pretty ambitious project, which creates a lot of variation and room to demonstrate approach.
I turn 50 tomorrow and I love vibe coding. In the hands of an expert with decades of experience in all the internal corners of C, Python and Postgres I find AI tools to be miracles of technology. I know how to ask them exactly what I want and I know how to separate the goodness from the bullshit. If Supabase is bringing AI closer to the developer at the database level then that is a great thing.
Vibe coding is excellent if you have the experience to understand what the AI is churning out and then what to do with it.
The problem we have now is we have people who aren't engineers trying to make an app and they end up creating insecure and buggy messes, then struggle with figuring out how to deploy, or they end up destroying all their code with no recovery because they didn't know anything about version control.
All of those things were already happening with normal developers 10, 20, 30 years ago, and will keep happening, with or without AI.
As a professional developer, this seems like a problem I don't need to care about.
Do you have any writings or materials that show your process in depth? I’m interested in learning from those who know how to really squeeze the juice out of these tools.
I think what they are saying is the 'secret sauce' to successfully vibe coding is being an expert with all the languages, frameworks and tools yourself.
Makes sense to me, vibe coding basically shifts your burden to specification and review, which are traditionally things a senior developer should be good at.
I think there is a lot to be said about what tasks you give to it, how you describe the work and prompt it, and the iteration / workflow loop.
I have a limited intuition for this based off my AI usage the past few years, but I want to learn from the pros.
I agree with this - I hear a lot of hate towards vibe coding but my experience with voice dictation and using 20 years experience in the trenches and so being very specific telling the model what to do has been, well, refreshing to say the least.
I used to pride myself of knowing all the little ins and outs of the tech stack, especially when it comes to ops type stuff. This is still required, the difference is you don't need to spend 4 hours writing code - you can use the experience to get to the same result in 4 minutes.
I can see how "ask it for what you want and hope for the best" might not end well but personally - I am very much enjoying the process of distilling what I know we need to do next into a voice dictated prompt and then watching the AI just solve it based on what I said.
Only rough if you're pretentious.
Making it easy for engineers, experienced OR aspiring is huge.
Part of what I think is rough is the framing of Postgres being "an alternative to Google’s Firebase" in the article. I mean, ok yeah they are both databases in a sense, but they are not the same thing at all. Firebase is a service and Postgres a database technology, and certainly not well-described as an alternative to Firebase by anyone competent in the industry. Lol.
I don't mean to demean "vibe coders" exactly either, but rather jumping on the hype train of using that term for your funding pitch. You're using AI to learn to become a software developer? Great! No problem with that.
But also — if you now have a database involved and you're handling people's data, you better learn what you're doing. A database provider pushing "vibe coding" is not a good look imo.
Yeah the heroku model, arguably AI makes the barriers for aspiring coders lower which means more potential customers
Hey man, their money spends like anyone else's.