The company said Tuesday after the bell that it has $455 billion in remaining performance obligations, up 359% from a year earlier.
Oracle now sees $18 billion in cloud infrastructure revenue in fiscal 2026, with the company calling for the annual sum to reach $32 billion, $73 billion, $114 billion and $144 billion over the subsequent four years.
Ok. Who is paying for these? I know OpenAI's revenue 3x'ed this year to $12b. Let's say they get to $15b by end of the year. Let's say they 3x next year to $45b/year.
$114b? $144b for Oracle? per year?
Who? Can someone make a serious attempt at speculating who and how revenue will get to 144b/year?
I know Oracle as a Stargate partner. Maybe this is OpenAI being extremely bullish on its growth figures?
Edit: Ok. I have a theory. Oracle is the only big cloud company NOT developing their own AI silicon chips. AWS, Google, Meta, Apple, Microsoft are all making internal AI chips that compete against Nvidia. Nvidia prioritizes Oracle as a partner so the first and the bulk of Nvidia's chips go to Oracle. Therefore, AI labs are going to Oracle for Nvidia access. Video on Ellison having dinner with Jensen Huang "begging" him for GPUs [0].
How do promises of $300B in future revenue somehow result in a $100B increase in Ellison's net worth today?
- The $300B from OpenAI doesn't even start until 2027 and runs through 2031
- We don't know anything about OpenAI's options to cancel, delay or modify the deal
- There's huge execution risk for Oracle, they have to scale up their operation by more 20x or 30x, obtain gear and staff and gigawatts of power
- Even if everything goes perfectly, and both Oracle and OpenAI perform, that $300B is revenue, not profit -- 70% or 80% of that goes suppliers for capex and operating costs
- That's an uncertain promise of a future stream of profits, and deserves a sizable discount to get back to present value
- Ellison owns about 40% of Oracle, so how does the share attributed to him personally get to $100B in today's dollars?
Years ago I worked at a $BIGCORP which already had a major presence in Azure, AWS and private datacenters, and there was chatter with the execs about putting a footprint in Oracle. Wouldn't be surprised if that's happening now.
Partnering with the non-Amazon cloud providers meant spending several millions in surplus infrastructure costs to hire more engineers to un-fuck the alternate cloud shitshows, but unlocked billions in partner deals (think Microsoft promoting our products running on Azure to Azure customers). And then every 3-5 years when we negotiated new cloud resource SKU discount rates we had leverage on common resources like VMs, storage buckets and SQL databases. "If you can't beat this rate we'll move part of our capacity to the other clouds" kinda thing.
Yeah, I saw the news and I was, WTF? I don't think any serious IT person likes Oracle. So how is it possible that they make so much money? Who is hiring them?
Someone explain to me how Oracle reporting residual performance obligations as though they were fact is different from Enron's mark-to-market accounting, because I don't really see why investors would do anything but run screaming from this. Oracle has been promised money by customers who don't currently have that money to deliver compute resources that Oracle doesn't currently have, and that adds $244 billion to their market cap in one day? That's the nearly the entire market cap of Cisco appearing out of thin air.
Who are Oracle's cloud computing customers? In my circle I hear mostly about AWS, with a little Azure and GCP, but I don't know people running anything on Oracle Cloud. Presumably there are circles where Oracle Cloud is gaining rapidly, given these numbers. Are these Oracle customers shifting from other cloud providers or using cloud resources for the first time?
EDIT: I missed the news from yesterday that OpenAI signed a $300 billion (!) contract with Oracle Cloud:
Oracle Cloud's strategy early on was to go after huge customers who they thought would like to use the cloud, but otherwise couldn't use AWS for technical reasons. They were the first to do bare metal hosts (no hypervisor) and Layer 2 networking, for example (layer 2 networking not necessarily being anything amazing, but they recognized that lots of potential corporate customers had built their own infrastructure that relied on it, so it was a big barrier to cloud adoption). They also will bend over backwards to gain the customers with deep pockets like big banks, auto manufacturers, etc. This is how they ended up with TikTok. And part of this is that they are very aggressive about building regions, because some of the big customers have hard requirements about where physically their data resides (think small governments that need/want their data on their own soil). I don't know where they stand now from a technical competition perspective (those examples I gave are no longer exotic in 2025), but I suspect they have continued this "court the huge customers" strategy.
I’m not at that level in my career, but I have had very negative issues with Google, their product graveyard, and their lacking of human support for serious issues for paid offerings.
They have a very long deprecation policy for anything under the actual “Cloud” banner.
Their solutions are more technically sound, their API and UX significantly more consistent; they make the billing system available through the same interfaces that you use for other data processing on the platform.
I wrote a more detailed diatribe on the failings of other providers[0]; but largely the “correctness”, “consistency” are the main reasons. For example their IAM actually works for everything, including GKE pods; and they don’t oversubscribe and try to align VMs to NUMA zones (AWS didn’t do that when I last checked, Azure still doesn’t even try).
Also, as a previous customer; I got much better support from Google than I ever got from AWS; but this is a very variable thing depending on your size, where you are and who your TAM is. In the games industry in Sweden its seriously 10/10- I even talked to the hypervisor, storage and SDN/Network teams directly.
The only aspects that Azure doesn’t have is the NUMA alignment and good support. Azure also has consistent RBAC model that integrates with Kubernetes, PaaS, and extends out to desktop apps like Office and Visual Studio. Unlike AWS, Azure also has a very consistent UI with a single pane of glass and a single API (ARM).
> their lacking of human support for serious issues for paid offerings.
Turns out this problem disappears if you spend enough money.
GCP does a _lot_ of things as well or better than AWS. I particularly mostly like their IAM model (the entire stack including their nice managed identity implementation and use of public key cryptography instead of API keys) and how "projects" (cloud accounts) can inherit properties in a hierarchy.
Are there any tech people on here choosing any oracle products? With the exception of the DB, I have never met anyone who used anything from oracle willingly. It was always because someone high up who got taken to lunch and came back with a signed contract.
Not for anything real, but Oracle Cloud's free tier is atypically nice.
Oracle Cloud's free tier includes a 4 CPU / 24 GB RAM Ampere A1 ARM VM. On AWS / GCP, an equivalent VPS would be $100+/month, but on Oracle cloud it's free.
Ellison called cloud computing "complete gibberish" and "insane," dismissing it as merely a marketing term. He argued that a "cloud" was simply a network of computers.
He just underestimated how much a combined marketing spend from three of the largest companies can convince people that an API is enough to justify extremely slow hardware at a 5-10x markup.
"Oracle is clearly leveraging a number of advantages in its cloud software/hardware businesses to attract the largest of the AI enterprises, including OpenAI, xAI, Meta, NVIDIA and AMD,"
It’s insanity to me that a large company was able to jump 36% in a day. I don’t care what the reason is. I’ve seen this is based on their cloud's AI demand… but really, 36%? Was there some major major contract signed worth $100B? Has anyone used their cloud? It’s terrible.
A lot of commission plans, even those which advertise "uncapped" commission, will still have the ability to set a cap for commission due to a single whale. Not that hitting your cap is an unenviable position in itself or anything... just perhaps a lot less exciting than it may sound (or not, just depends on the rules in the plan and what the leadership decides on the given deal). That is, if one sales rep can even claim such a large deal as their own work.
E.g. a few years back I worked with a guy (at an IT sector company) who had a quota of 3 million in the region, but won a single deal worth 30 million + had a great quarter in general anyways. He ended up not getting too much more than if he just had the great quarter.
And vaulting Ellison ahead of Musk as the world's richest
> Oracle’s founder, Larry Ellison, added $100 billion to his net worth on Wednesday. Bloomberg reported that he had topped Tesla
CEO Elon Musk as the world’s richest person.
$114b? $144b for Oracle? per year?
Who? Can someone make a serious attempt at speculating who and how revenue will get to 144b/year?
I know Oracle as a Stargate partner. Maybe this is OpenAI being extremely bullish on its growth figures?
Edit: Ok. I have a theory. Oracle is the only big cloud company NOT developing their own AI silicon chips. AWS, Google, Meta, Apple, Microsoft are all making internal AI chips that compete against Nvidia. Nvidia prioritizes Oracle as a partner so the first and the bulk of Nvidia's chips go to Oracle. Therefore, AI labs are going to Oracle for Nvidia access. Video on Ellison having dinner with Jensen Huang "begging" him for GPUs [0].
[0]https://www.youtube.com/watch?v=zPQnPciuxhU
How do promises of $300B in future revenue somehow result in a $100B increase in Ellison's net worth today?
- The $300B from OpenAI doesn't even start until 2027 and runs through 2031
- We don't know anything about OpenAI's options to cancel, delay or modify the deal
- There's huge execution risk for Oracle, they have to scale up their operation by more 20x or 30x, obtain gear and staff and gigawatts of power
- Even if everything goes perfectly, and both Oracle and OpenAI perform, that $300B is revenue, not profit -- 70% or 80% of that goes suppliers for capex and operating costs
- That's an uncertain promise of a future stream of profits, and deserves a sizable discount to get back to present value
- Ellison owns about 40% of Oracle, so how does the share attributed to him personally get to $100B in today's dollars?
I agree. I think Wall Street overreacted a little. $300b seems like absolutely best case scenario.
It's a bubble.
The bubble can stay irrational longer than you can stay solvent.
Years ago I worked at a $BIGCORP which already had a major presence in Azure, AWS and private datacenters, and there was chatter with the execs about putting a footprint in Oracle. Wouldn't be surprised if that's happening now.
Partnering with the non-Amazon cloud providers meant spending several millions in surplus infrastructure costs to hire more engineers to un-fuck the alternate cloud shitshows, but unlocked billions in partner deals (think Microsoft promoting our products running on Azure to Azure customers). And then every 3-5 years when we negotiated new cloud resource SKU discount rates we had leverage on common resources like VMs, storage buckets and SQL databases. "If you can't beat this rate we'll move part of our capacity to the other clouds" kinda thing.
Yeah, I saw the news and I was, WTF? I don't think any serious IT person likes Oracle. So how is it possible that they make so much money? Who is hiring them?
Someone explain to me how Oracle reporting residual performance obligations as though they were fact is different from Enron's mark-to-market accounting, because I don't really see why investors would do anything but run screaming from this. Oracle has been promised money by customers who don't currently have that money to deliver compute resources that Oracle doesn't currently have, and that adds $244 billion to their market cap in one day? That's the nearly the entire market cap of Cisco appearing out of thin air.
Who are Oracle's cloud computing customers? In my circle I hear mostly about AWS, with a little Azure and GCP, but I don't know people running anything on Oracle Cloud. Presumably there are circles where Oracle Cloud is gaining rapidly, given these numbers. Are these Oracle customers shifting from other cloud providers or using cloud resources for the first time?
EDIT: I missed the news from yesterday that OpenAI signed a $300 billion (!) contract with Oracle Cloud:
https://techcrunch.com/2025/09/10/openai-and-oracle-reported...
So this big cloud revenue jump may be from a single giant customer.
Oracle Cloud's strategy early on was to go after huge customers who they thought would like to use the cloud, but otherwise couldn't use AWS for technical reasons. They were the first to do bare metal hosts (no hypervisor) and Layer 2 networking, for example (layer 2 networking not necessarily being anything amazing, but they recognized that lots of potential corporate customers had built their own infrastructure that relied on it, so it was a big barrier to cloud adoption). They also will bend over backwards to gain the customers with deep pockets like big banks, auto manufacturers, etc. This is how they ended up with TikTok. And part of this is that they are very aggressive about building regions, because some of the big customers have hard requirements about where physically their data resides (think small governments that need/want their data on their own soil). I don't know where they stand now from a technical competition perspective (those examples I gave are no longer exotic in 2025), but I suspect they have continued this "court the huge customers" strategy.
Cloud providers are chosen from the top sometimes.
if you’re using Azure, your CFO probably made the decision.
if you’re using AWS, your CEO likely made the decision.
if you’re using Google cloud, likely your CTO made the decision.
If your company relies on government contracts; they dictate the cloud provider; and Oracle knows this.
Why the Google Cloud / CTO association?
I’m not at that level in my career, but I have had very negative issues with Google, their product graveyard, and their lacking of human support for serious issues for paid offerings.
I would not put my business in their hands, ever.
They have a very long deprecation policy for anything under the actual “Cloud” banner.
Their solutions are more technically sound, their API and UX significantly more consistent; they make the billing system available through the same interfaces that you use for other data processing on the platform.
I wrote a more detailed diatribe on the failings of other providers[0]; but largely the “correctness”, “consistency” are the main reasons. For example their IAM actually works for everything, including GKE pods; and they don’t oversubscribe and try to align VMs to NUMA zones (AWS didn’t do that when I last checked, Azure still doesn’t even try).
Also, as a previous customer; I got much better support from Google than I ever got from AWS; but this is a very variable thing depending on your size, where you are and who your TAM is. In the games industry in Sweden its seriously 10/10- I even talked to the hypervisor, storage and SDN/Network teams directly.
[0]: https://blog.dijit.sh/gcp-the-only-good-cloud/
The only aspects that Azure doesn’t have is the NUMA alignment and good support. Azure also has consistent RBAC model that integrates with Kubernetes, PaaS, and extends out to desktop apps like Office and Visual Studio. Unlike AWS, Azure also has a very consistent UI with a single pane of glass and a single API (ARM).
> their lacking of human support for serious issues for paid offerings.
Turns out this problem disappears if you spend enough money.
GCP does a _lot_ of things as well or better than AWS. I particularly mostly like their IAM model (the entire stack including their nice managed identity implementation and use of public key cryptography instead of API keys) and how "projects" (cloud accounts) can inherit properties in a hierarchy.
You forgot Hetzner :(
As I said in another comment; unfortunately the C-levels think that marketing is safety.
They’re not choosing Hetzner because they think its the unsafe option.
Some small startup called OpenAI signed a $300m deal with them.
Are there any tech people on here choosing any oracle products? With the exception of the DB, I have never met anyone who used anything from oracle willingly. It was always because someone high up who got taken to lunch and came back with a signed contract.
Not for anything real, but Oracle Cloud's free tier is atypically nice.
Oracle Cloud's free tier includes a 4 CPU / 24 GB RAM Ampere A1 ARM VM. On AWS / GCP, an equivalent VPS would be $100+/month, but on Oracle cloud it's free.
> on Oracle cloud it's free
And technically can go poof at any time, I guess?
Yep, but great for stuff where that's okay. I use mine as a CI server and for fuzz testing.
Sounds good but I note there's a limit of 3000 OCPU hours and 18000 GB hours a month. Any idea how easy that is to exceed? Could get expensive.
edit: and 200GB of free block volume storage WITH backups? That seems like a good replacement for standard cloud storage.
Ellison called cloud computing "complete gibberish" and "insane," dismissing it as merely a marketing term. He argued that a "cloud" was simply a network of computers.
in 2008
Clearly he's willing to change his mind. I guess you aren't.
(And AWS was pretty overhyped in 2008.)
I mean, he’s not entirely wrong.
He just underestimated how much a combined marketing spend from three of the largest companies can convince people that an API is enough to justify extremely slow hardware at a 5-10x markup.
Definitely still true
cloud compute business has turned into real state. the market seems to think he was correct at this moment
He's not wrong about what it is, "cloud" literally is just marketing jargon for someone else's network of computers...
I'm struggling to understand the narrative around how Oracle is being positioned as a rival to Nvidia in the context of AI, or AI in general: https://finance.yahoo.com/news/nvidia-may-have-a-rival-on-th...
"Oracle is clearly leveraging a number of advantages in its cloud software/hardware businesses to attract the largest of the AI enterprises, including OpenAI, xAI, Meta, NVIDIA and AMD,"
What exactly are these advantages?
If you're wondering who uses Oracle it's TikTok.
It’s insanity to me that a large company was able to jump 36% in a day. I don’t care what the reason is. I’ve seen this is based on their cloud's AI demand… but really, 36%? Was there some major major contract signed worth $100B? Has anyone used their cloud? It’s terrible.
OpenAI contract, "Sept 10 (Reuters) - OpenAI has signed a contract to purchase $300 billion in computing power over roughly five years from Oracle".
Yeah OpenAI signed a huge deal with them - $300 Billion
Wild - I’d love to be the sales guy who got that commission
A lot of commission plans, even those which advertise "uncapped" commission, will still have the ability to set a cap for commission due to a single whale. Not that hitting your cap is an unenviable position in itself or anything... just perhaps a lot less exciting than it may sound (or not, just depends on the rules in the plan and what the leadership decides on the given deal). That is, if one sales rep can even claim such a large deal as their own work.
E.g. a few years back I worked with a guy (at an IT sector company) who had a quota of 3 million in the region, but won a single deal worth 30 million + had a great quarter in general anyways. He ended up not getting too much more than if he just had the great quarter.
> Wild - I’d love to be the sales guy who got that commission
Sorry to share - there's no sales guy here. It all happened at the top, hint - Stargate.
Yes, contract with OpenAI
Glad I followed my rule of never shorting a stock.
Today might be the day to start
And vaulting Ellison ahead of Musk as the world's richest
> Oracle’s founder, Larry Ellison, added $100 billion to his net worth on Wednesday. Bloomberg reported that he had topped Tesla CEO Elon Musk as the world’s richest person.
Yesterday's news;
Related:
OpenAI, Oracle Sign $300B Computing Deal, Among Biggest in History
https://news.ycombinator.com/item?id=45201595