Its not clear that the infrastructure will ever be paid back. Its also not clear that once models are trained and distilled that the infrastructure will be useful. Ignoring 1.9 trillion dollars of debt is probably not the greatest financial analysis.
>However, there's no reason for why this level of spending has to be maintained in the first place.
Exactly. So Anthropic and OpenAI might be holding the bags for latecomers who distill their models and get ahead. Sort of the Commodore64 and Atari of the AI space.
>Once all the data centres are built, AI spending is going to plummet.
Well data centers need maintenance, and GPUs and similar do have a very short lifespan. But even if spending plummets repayments on existing debt needs to be accounted for.
>In a pinch, AI companies can also cut costs by no longer allowing people to use their products for free and temporarily suspending expensive AI research.
Research keeps them in the headlines, see above about the latecomers.
>For context, Alphabet (parent company of Google and Youtube) reported earning almost 162 billion dollars of revenue just in 2019. Adjusted for inflation, that's an even more impressive 213 billion dollars.
Yep, so all Open AI or Anthropic need to do is get acquired by a much more profitable business to pay back their debt.
>at some point, the rate of increase has to level off, but that doesn't have to happen anytime soon.
The rate of increase of revenue in the AI sector also doesnt need to be apportioned to OpenAI or Anthropic. If it wasnt for their generous loss leading subscriptions or their rnd, theres not much there to hold on to people.
>it's hard for me to see AI consumption not being 10x what it is now.
Right AI consumption will increase very likely. But will it increase at the cost of an already subsidised subscription, or increase the use of cheap Deepseek 4.1 tokens.
>For starters, the average person isn't going to use a locally hosted model, ever.
Isnt this expressly where Apple Intelligence is heading? Building small low cost models that run on Apple hardware? If I was nVidia I would have both plans being run down at the same time, looking towards selling GPUs to hyperscalers on one side, and working with OEMs to deploy AI coprocessors in laptops on the other side.
>For example, I pretty much only use ChatGPT, not because I have any really reason to, but because it was the first LLM that I ever used.
N = 1. OpenAI is already what halfway to their trust thermocline anyway?
>A potential way for a bubble to pop is for the companies involved to run out of capital to spend while they grow, causing them to fail.
Isnt Anthropic already down to BB+?
> NVIDIA are all AI companies that have seen their market capitalization explode over the past few years
NVIDIA is a hardware company that can has so much demand on their products they will likely survive 3 AI market collapses.
>Alphabet, Meta, and NVIDIA
Alphabet, Meta and NVidia all have other sources of revenue. They are companies building AI, not AI companies.
>When the dot-com bubble happened, it was largely driven by companies that couldn't turn a profit or had P/E ratios in the hundreds.
Anthropic and OpenAI dont have "E" (as in earnings)
Anthropics Debt to Equity is 363%.
>A potential way that this could be a bubble is if AI proves to be a fad.
A potential way this could be a bubble is if the makers of consumer devices point customers at cheaper or local models as part of their onboarding experience. OpenAI and Anthropic get thoroughly zuckerberged by the people holding the primary relationship with the customer base.
Or perhaps, the funding to the top 1% of token purchasers also turns out to be loans, and that dries up too.
https://www.pymnts.com/news/artificial-intelligence/2026/ope...
>As one final note, I don't think the rhetoric about the AI bubble is really about finances.
The people bringing it up every 10 seconds are doing it to spread FUD about AI. That doesnt really change the financial picture.