Who replaces Nvidia? That's the 5 trillion dollar question, and nobody has stepped forward to answer it.
The closest the industry came to replacing CUDA was with Khronos and OpenCL, but that ship has sailed. CDNA, Apple Silicon, Intel Inside, none of them are trying to take Nvidia's crown. The niche for HPC will exist, and Nvidia continues to serve the niche while others turn up their nose.
For my money, shorting Nvidia is like betting on Glass Joe to beat Mike Tyson. There's going to be incredible cloud spend on unglamorous stuff like defense and automated computer vision, and that will keep Nvidia's demand afloat even at a wild valuation.
That's a substantive point, the GPUs are reprogrammable to do other tasks so the people who pay for massive CFD computation in Mil and Oil will have competitive offers, but I'm unsure the spectacular future value is there, if the perpetual "buy more" dries up. They could wind up alive, but with a smaller horizon.
I know I harbour resentment because of the knock on effects on ram and SSD pricing, but I'm serious that I think the amount of capital being sunk in hyperscalers does not look to me to be recoverable inside the investors ROI. If my example is poorly chosen, perhaps the people currently buying the kit are the ones destined to have a fall.
With Nvidia no one needs to replace them. Their valuation is build on large demand and huge margins. Demand just going lower or margins dropping following that should lead to lower valuation without anyone replacing them.
I could very well see current market being in state of overinvestment. Meaning future demand is lower which could lead to less units sold thus lower profits thus lower valuation.
Nvidia's profits are built on demand and margins. Their valuation is constrained by their supply, Nvidia sells-out of almost all their datacenter hardware before it hits shelves. Even if the memory shortage never clears up, Nvidia is the best-poised company to buy memory at-cost and sell it for ludicrous prices.
For their margins to go down, that demand would have to be satisfied. And those customers by-and-large do not see any commodified CUDA alternatives on the market, they only want Nvidia hardware. Hence my question - who will rise to the challenge? If the answer is nobody, then you've just found out why people don't short Nvidia.
A related question might be whether anyone is seriously preparing for whatever opportunities present themselves after the bubble bursts.
For example, after the dot com bust, there was plenty of money to be made in aggregating bankrupt or underutilised telecoms assets - endless duct shares, dark fibre, and networking kit which could turn a profit if bought cheaply enough in a fire sale.
Is anyone preparing to bid pennies on the dollar for bankrupt AI datacentres? I can see how it might potentially make sense to do so in places like the EU or UK where increasing data sovereignty concerns might make locally-based small/mid-scale private inference an attractive proposition if the capital costs are low enough.
Agreed. Shorting requires timing the market perfectly, but buying assets after a crash just requires having liquidity and patience. Historically, that's often been the safer play
After the dot-com bust, infrastructure assets turned out to be one of the most undervalued asset classes. Maybe in a few years people will look back on GPU clusters the same way they looked at dark fiber back then
This doesn't make sense. It completely misses that there are tons of other companies that will happily take the Nvidia capacity at the same price if their biggest customers reduce spend.
I absolutely do NOT want to do this. I am asking if any of the people with real money and fintech are doing this. Not because I want to, because I want to UNDERSTAND.
I think a lot of people are mixing two different ideas: AI is overhyped and AI company stocks are about to crash. The first might well be true, but the second might not be. The dot-com bubble burst too, but the internet didn't go away. The question is more about who will be left standing once the era of endless capex growth ends.
Betting against deep learning has been a losing game over the past 5 years.
As long as the models keep getting better, you will keep losing betting against them.
Who replaces Nvidia? That's the 5 trillion dollar question, and nobody has stepped forward to answer it.
The closest the industry came to replacing CUDA was with Khronos and OpenCL, but that ship has sailed. CDNA, Apple Silicon, Intel Inside, none of them are trying to take Nvidia's crown. The niche for HPC will exist, and Nvidia continues to serve the niche while others turn up their nose.
For my money, shorting Nvidia is like betting on Glass Joe to beat Mike Tyson. There's going to be incredible cloud spend on unglamorous stuff like defense and automated computer vision, and that will keep Nvidia's demand afloat even at a wild valuation.
That's a substantive point, the GPUs are reprogrammable to do other tasks so the people who pay for massive CFD computation in Mil and Oil will have competitive offers, but I'm unsure the spectacular future value is there, if the perpetual "buy more" dries up. They could wind up alive, but with a smaller horizon.
I know I harbour resentment because of the knock on effects on ram and SSD pricing, but I'm serious that I think the amount of capital being sunk in hyperscalers does not look to me to be recoverable inside the investors ROI. If my example is poorly chosen, perhaps the people currently buying the kit are the ones destined to have a fall.
With Nvidia no one needs to replace them. Their valuation is build on large demand and huge margins. Demand just going lower or margins dropping following that should lead to lower valuation without anyone replacing them.
I could very well see current market being in state of overinvestment. Meaning future demand is lower which could lead to less units sold thus lower profits thus lower valuation.
Nvidia's profits are built on demand and margins. Their valuation is constrained by their supply, Nvidia sells-out of almost all their datacenter hardware before it hits shelves. Even if the memory shortage never clears up, Nvidia is the best-poised company to buy memory at-cost and sell it for ludicrous prices.
For their margins to go down, that demand would have to be satisfied. And those customers by-and-large do not see any commodified CUDA alternatives on the market, they only want Nvidia hardware. Hence my question - who will rise to the challenge? If the answer is nobody, then you've just found out why people don't short Nvidia.
A related question might be whether anyone is seriously preparing for whatever opportunities present themselves after the bubble bursts.
For example, after the dot com bust, there was plenty of money to be made in aggregating bankrupt or underutilised telecoms assets - endless duct shares, dark fibre, and networking kit which could turn a profit if bought cheaply enough in a fire sale.
Is anyone preparing to bid pennies on the dollar for bankrupt AI datacentres? I can see how it might potentially make sense to do so in places like the EU or UK where increasing data sovereignty concerns might make locally-based small/mid-scale private inference an attractive proposition if the capital costs are low enough.
Agreed. Shorting requires timing the market perfectly, but buying assets after a crash just requires having liquidity and patience. Historically, that's often been the safer play
After the dot-com bust, infrastructure assets turned out to be one of the most undervalued asset classes. Maybe in a few years people will look back on GPU clusters the same way they looked at dark fiber back then
Did this research last week on the same thing:
NVIDIA (NVDA) — The Short / Underperformance Thesis - https://nvidia-stock-analysis.pagey.site/
This is pump and dump on the largest scale we have ever seen. It's effectively privatize the profits and socialize the losses.
This doesn't make sense. It completely misses that there are tons of other companies that will happily take the Nvidia capacity at the same price if their biggest customers reduce spend.
AFAIK Michael Burry is shorting it via PUT options — can look into what he is doing.
Many have theorized this. But i wouldn't bet money on it, the timing is just too unpredictable IMO.
The market can stay irrational longer than you can stay solvent.
I absolutely do NOT want to do this. I am asking if any of the people with real money and fintech are doing this. Not because I want to, because I want to UNDERSTAND.
I think a lot of people are mixing two different ideas: AI is overhyped and AI company stocks are about to crash. The first might well be true, but the second might not be. The dot-com bubble burst too, but the internet didn't go away. The question is more about who will be left standing once the era of endless capex growth ends.