I see two factors converging to cause a collapse of this house of cards:
1. People are realizing that what they need isn't more general intelligence, it's more specialization. A small but well tuned coding model, a small but well tuned customer service model, a small but well tuned document explorer.
2. Specialized hardware - TPUs and NPUs - especially coming out of china. The latest GLM model was trained and runs on Huawei hardware. Nvidia is only worth so much because they are the biggest and best provider of the kind of compute needed to run llms, but the export bans mean china has a lot of incentive to topple that monopoly.
The amount of compute we need to do the things llms do is falling rapidly, the number of people who can provide that compute is rising.
It’s not quite as simple as that. Several studies have shown the opposite: models trained on more diverse knowledge tend to cross-pollinate across domains. So a more generalized model can actually perform better than a specialized one.
That’s why you’re not seeing tons of tiny models (one for Python, one for Pascal, one for Rust, etc).
But it doesn't match my experience. Qwen3.8 27b is clearly smarter at coding than MANY bigger models. gpt-oss-120b for example, is almost 4x the size, and performs way worse at coding tasks.
It's clear to me that you can build small models that work well at specific tasks.
Python vs Rust is probably too fine grained a way to build a model. Coding in general seems like a better target.
There will always be a place for large generalist models, no doubt. But I think that place is much smaller than the big ai companies are counting on.
It practically became a joke about how a huge amount of the training data for GPT-4 was bottom of the barrel reddit vomit and obvious bot spam. Leading to many bizarre edge cases.
Shows qwen3.8-27b along side seven larger models of ~similar vintage. Only one scores above 27b.
Many of those are closed models so idk their exact parameter count / active param count, but it hardly matters - i’m sure all of them are far above 100b params
My point is not that bigger is pointless. It’s just clearly not the only road to take to make a model better, which is obvious just from seeing how models of the same size have gotten better over the past few years
First off, I'd include Qwen flash-next and GLM 5.3 to show some of the other strong open weight models, and they predictably dominate it, but they're much larger. But, it shows up right next to DSv4 Flash 0731 on the overall index, and that's much larger. It's a great model! But then scroll down and hit Time Per Task, and you'll see that DSv4 Flash takes 3.6 seconds per task to Qwen's 21.1. That's what I meant when I said this:
>speed due to excessive thinking maybe to make up for the smaller amount of world knowledge baked in (qwen 27b's main issue iirc), etc - they're tuned for different things.
It can make up for its shortcomings by iterating a lot longer, and using way more thinking tokens. And that's a great trade if you don't have the vram to run the bigger models, but speed is pretty important for getting things done... And that's why DSv4Flash is great, too, despite being much larger, and scoring similarly on the intelligence index.
I make heavy use of smaller local models on a daily basis (Qwen3-VL for auto-captioning images, Gemma3:27b for some translation work, etc.). Gemma3:27b is a good example of a very capable general purpose multimodal model and has handled almost everything I've thrown at it from sentiment analysis to documentation writing.
I suppose I was drawing a distinction between specialized and general intelligence versus small and large. I don’t think those are necessarily mutually exclusive.
And Qwen3.8-27b is still better at coding than opus 4.1.
Yes, if you list off models 27b is better than it’s all older models. But that’s my point - newer models are better than older models at the same AND much smaller size. That’s because model size matters less than they say. Training data and model architecture matter more.
Yeah, the cross domain transfer learning from RL is overstated by a lot.
In 2026, the default outlook should be suspicion for any big private organisations with profit motive.
You don’t need to think about climate change studies. Instead you can read the allegedly tainted studies we’re actually talking about and profess to all of us what is wrong with them. You can’t point to exactly where they’ve fudged them.
And even if compute demand were perfectly elastic it’s only a good thing insofar as it drives demand for new Nvidia hardware. If tokens can be served from Apple hardware or Google hardware or Huawei hardware that doesn’t help Nvidia.
I mean… some homes definitely do. You must have seen those houses that are all lit up front the outside by lawn mounted spotlights.
If that dies because a lot of people’s needs turn out to be met by a system at home they can run a 30b-150b model on, a lot more of that money goes to apple or intel or amd.
1. Yes, smaller models will become more popular, especially as the tokenmaxxing trend dies down and people start stretching their budgets farther. That is a downward pressure on demand.
But along the same dimension, consider that currently only about 40 - 60% of the world uses AI for only about 5 - 15% of their work hours. That means there is still 2x growth from users and 7x - 20x growth from the rest of the work hours left to capture! That is 14x - 40x more demand. Then consider that agentic tasks require multiples more tokens, and that is the kind of usage that is most likely to be deployed, and also the kind of usage that is the least used right now. That's another huge multiple to be tacked on.
And the entire AI industry has been lamenting the extreme compute crunch they're facing (and also why Claude has 9's comparable to GitHub; whereas OpenAI has been chugging along because Altman was OK being called a "podcasting bro" while desperately scrounging for compute years in advance.)
Nvidia's meteoric rise is entirely due to this kind of exploding demand with extremely limited supply.
2. Competing hardware is definitely a threat, but it has its own hurdles. Because the real bottleneck is not Nvidia, it's TSMC.
Pretty much all demand for all chips in all devices in all the world flow to, like, 3 companies in the world that actually fabricate them, and TSMC is the biggest. And the supply is extremely tight, as the exploding costs of electronics clearly shows.
So now TSMC will of course try to keep all its customers happy, but it will inevitably be forced to choose which ones it will keep happiest. And those will be the customers who can pay it the most. And that would be the one with all the money from its de facto status as a monopoly (and possibly even a monopsony)...
Which would be Nvidia ;-)
So yes, compute per task is falling rapidly... but it's barely a dent in the humongous total addressable demand, and the amount of hardware to support that compute is still very constrained, and most of that supply will likely flow through Nvidia.
Generative video requires significantly more computing power and energy than generative text.
OpenAI is fucked, compute is still needed, it's just them that isn't.
Without a material change in the market (more buyers, vastly cheaper generation), it's unlikely a different company could make that work. More buyers isn't likely to happen, so that leaves vastly cheaper generation - something that would cause nvidia's value to collapse if it happened.
I'd suggest that's only the case given the current quality of output. Media is incredibly expensive to produce. A model capable of sufficiently high quality could charge prices that are absurd by today's standards.
Video generation would only make sense at that scale if it was targeting individual consumers, but then it’d need to cost something that consumers are willing to pay - which practically is probably a few hundred per year at most among US consumers, and much less globally, so again it doesn’t solve for the size of the AI companies.
I don’t see a way that video generation becomes a big industry without making generation much much cheaper.
That's already not the case today. If you sat me in front of an LLM and told me to figure out if I'm working with K3 or Astra, I could probably do it, but it would take some work to be certain.
> it's more specialization
China, constrained by hardware, and talent (not to slight the Chinese, but they are limited to domestic resources - and much of the US effort is very international). They did, what the Chinese do, and optimized the process of production, and drastically lowered the cost of development of their models. Cheeper to build, cheaper to run is just good economics.
Meanwhile in the us, we have open AI doing "experiments" - it looks like the costs around the hugging face hack are going to be about the same as China would spend on building out one of their smaller efforts (several million dollars). (Depending on whos numbers you trust, the fact that I can even make this claim should make you raise an eyebrow).
Go back to the 80s' and "expert systems" - most people will tell you that for their time, they were amazing, and useful. People would have loved to have more of them but they were so cost prohibitive that we all but abandoned them for serious use. The US frontier labs seem to have forgotten this lesson and their calls to "slow down" look like an excuse to "cut the waste so we can move to making money".
The problem with that is that OpenAI can only afford to pay for the compute because they are burning investor money (and so are most of OpenAI's biggest clients). They are losing billions. If they stop burning money, nobody else will be there to pay for that compute at OpenAI's cost.
Sure, somebody will probably be able to use these GPUs, they just won't be able to pay nearly as much for them as OpenAI does.
In reality, it's just nowhere near worth as much as OpenAI pays for it. Inflating the cost of compute is part of the problem caused by the circular financing, and if (or maybe when) OpenAI goes, the price of compute will go with them.
But that’s the point. Investors believe investment in AI will pay off.
Failure to take into consideration those kind of correlations ("If my biggest client isn't able to buy it, I would be able to find someone else who will") is one of the principle causes why many risk models turned out to be garbage during the Great Financial Crisis.
But I also doubt Nvidia is on the hook if OpenAI just no longer wants the compute. I bet they are only on the hook if OpenAI cannot pay for it (is insolvent in some way).
I also have to bring up that OpenAI has already spat out an inference chip that beats Nvidia on flops per watt. So they could potentially not need the compute while other ai companies do.
They would have to go insolvent in a way that hits Nvidia revenue. Those are related by distinct factors, a difference that may matter in a crisis.
This is the big point IMO since I have never given $1 to OpenAI but I subscribe to Vidu and Typecast, and have given money to Kling, Hailou, and even Gemini in the form of Google Workspace.
So these other guys have products and use cases, which OpenAI has never been able to crack beyond ChatGPT. And ChatGPT was never worth paying for, IMO.
If OpenAI dies, it's not because there is no market for the technology (which is all NVIDIA cares about), it's more that OpenAI doesn't know how to run a relevant technology company.
They were given everything, not just NVIDIA's billions of dollars and credit backing but all the first-mover advantage, all the respect and credibility early on, so it's really sad to see them unable to develop interesting products and turn a profit in a space they helped pioneer, while so many others are making money with the tech all around them.
NVIDIA is fine. The technology will continue to improve and NVIDIA will stay at the center. OpenAI is fucked - knew it when they retired Sora to focus on text-to-text and coding (a largely solved problem).
All the "frontier" AI companies *are* currently insolvent. They have never been anything other than cash burning machines.
The only way they keep the lights on and the doors open is by borrowing money --- and epic amounts of it. If those operating the cash spigot decide to turn it off, all AI companies will likely be similarly affected --- and so will Nvidia.
OpenAI expects to burn through more cash between 2024 and 2029 than Uber, Tesla, Amazon and Spotify did - combined - before those companies started making money
https://www.morningstar.com/news/marketwatch/20251205243/thi...
Fairly sure data center construction costs are also going up (they require so many resources that everything is constrained at the moment, especially electricity production).
So I don't understand in what world these frontier AI companies can somehow become profitable. The basic tech they're using is basically the same. Yes, around the edges there are a lot of things that can be done, and were done, like caching, batching, mixture of experts, etc, but basically everyone has done all of that by now, and they're still losing money.
So:
Total costs going up a lot - revenues per unit not increasing proportionally, if anything, Chinese models are forcing those down.
How does that math work out to profits? I don't see it.
Or about as bad, after trillions of dollars in investments over multiple years, let's say the entire frontier AI sector has a total profit of $20bn by 2030. In what world does that make sense? Assuming they can scale that total profit to $100bn in 2035 without investing another cent from 2027 to 2035 (utterly ridiculous), the return on investment would happen in roughly 20 years.
China is the one that is really in the driver's seat here. They have the opportunity and the ability to nullify/wipe out our huge investment in AI.
How? The hardware is in OpenAI's datacenters. Does Nvidia have a couple hundred semi trucks, contractors, and IT technicians, to repo the hardware and resell it to someone else before it's lost most of its value? These chips will be replaced approx every 3-4 years. So if OpenAI tanks, after Nvidia pays for and waits for the process to collect the hardware, they then have to sell it for pennies on the dollar. They lose almost all the investment.
Also consider that SpaceXAI already had datacenters full of gear that they basically weren't using because nobody wanted their product, so they now rent it to Anthropic. The demand for hardware isn't really there at the scale of OpenAI.