So, about 2 years worth of operations based on alleged $14 billion burn rate projected for 2026.
What an absurd amount of money - if only this was invested in energy sector scientific research and development, or healthcare or anything else practical.
I really hoped to see compact molten salt nuclear reactors in operation before 2030.
I don't understand how this is the top comment. LLMs have unlocked a lot of value for me personally, and arguably for the society as a whole. They are also one of the coolest technologies I've tried in years. As a technologist, I'm really glad that money is pouring in and allowing us to find its limits.
A major cost in building and operating datacenters is energy, so it does mean that significant share of those money would go into energy development. Large demand is one of the best things for stimulating technology development, and in this case we'd definitely see investment in solar and compact/safer nuclear of which MSRs are a part of.
xAI already brings gas turbines on-site, and i think the trend of on-site energy generation will grow, which will open opportunity (by providing well finaned demand) for compact/mobile/safer nuclear, and BigTech companies are among the best for any new tech development. I expect nuclear engineer positions get opened with Google and the likes :)
This .. doesn't seem like such a terrible deal? At the purported growth rates, you'd expect OpenAI to reach 60-100 billion revenue by 2028. This is more or less the equivalent of building a new AWS.
Provided they keep cost growth slower than revenue and don't get disrupted by another model provider/commodification etc.
2020s ai could be the first systemic stall I see. Let's assume agentic could really be a force for improvement but the cost model is unsustainable and will choke.
All of these things can be simultaneously true (and I would say, are true):
1) We are in a huge investment bubble right now and it's going to burst.
2) LLMs are extremely useful right now for certain niche tasks, especially software engineering.
3) LLMs have the potential to transform our world long-term (~10 yr horizon), on the order of the transformations wrought by the internet and mobile.
4) LLM's don't lead directly to AGI (no continuous learning), and we're not getting AGI any time soon.
This is an extremely obvious point, but bears repeating. I feel the assumption of an implicit link (in both truth or falsehood) between these fairly independent assertions can cause people to talk past each other about the really important questions in play here.
Regarding The Great Bubble, I am very very bearish about OpenAI in particular. They've had a good run for three years with consumer mindshare due to their first-mover advantage, but they have no moat, trouble monetizing most of their users, not much luck building out products that stick among consumers that aren't chatbots, and their models are no better than Anthropic's, Google's, or even the best Chinese open weight models 6 months later.
My bet would be on Google and Apple together (with Gemini powering Siri, for now) destroying OpenAI in the consumer AI market over the next 2-3 years. Google has first-rate models... but more than that, both Google and Apple have the enormous advantage of owning underlying platforms that they can use to put their own AI chat in front of consumers. Google has a mobile OS, the leading browser, and search. Apple has the premium hardware and the other, premium, mobile OS. They also have the advantage of the current regulatory climate being less antitrust than it was. And they don't have to monetize their AI offerings (no ads in gemini; ChatGPT is adding them) and can run them at a loss for as long as it takes to eat up OpenAI's market share. If they partner up, as they seem to be doing, OpenAI should very very afraid.
>2) LLMs are extremely useful right now for certain niche tasks, especially software engineering.
Don't get lost in the tech scene sauce, programming is a small sliver of what people are using LLMs for. OpenAI's report in September pegged it at ~4% of tokens being for software generation. Sure Anthropic is probably 80% or something, but only a small sliver of LLM users are using Anthropic. The reality is probably even less if you count Google's AI overviews. We hate it, but I have never seen a regular person skip over it.
The question is if regular people will pay cell phone level subscription costs ($70-$100/mo) for LLMs. If so, then we are probably not in a bubble, and the ROI will have a 5-10 yr horizon, which is totally tenable.
500,000,000 people paying $75/mo is $450B/yr. Inference is cheap too, it's training that is ludicrously expensive. Don't be fooled by the introductory pricing we have today either, that's just to get you dependent.
And yeah, chinese models, but look at what they did to tiktok. No way they are going to let the Chinese government be peoples confidants and no way is more than 0.01% of people gonna home lab.
You forgot to mention the case where the US economy craters because of political missteps leading to Google and Apple being distracted leaving it open for Chinese or other models to surge ahead.
At ~$100B in AI funding, the question isn’t capital—it’s physical constraints.
Data centers need power (H100s are ~700W each), and recent capacity additions were mostly pre-allocated. Chip supply is also constrained by CoWoS packaging, not fab capacity, and expansions take years.
If power, packaging, and GPUs are fixed in the near term, does $100B mostly drive inflation in AI infrastructure prices rather than materially more deployed compute? Are we seeing the real cost of a usable GPU cluster rise faster than actual capacity?
Has anyone modeled what $100B actually buys in deployable compute over the next 2–3 years given these constraints—and whether that figure is shrinking as more capital piles in?
Oh it is driving inflation alright, inflation for everyone else because OpenAI are buying things like a years supply of DRAM with money they don't even have yet (surely that should be an investigation?).
> as part of the startup’s [OpenAI's] efforts to raise up to $100 billion
At this stage, why not go public? Yes, they would need to manage quarterly financial reports and answer to shareholders, but they have reached a size where they are in the top 20 range on the NASDAQ. These public companies doing well, so it seems like a logical next step.
I actually think the public markets have a lot less faith in openai than softbank does. They need these crazy investors. Public markets would not value openai at $1.4trn. So they can't go public. It would reveal how bad things are.
I think it should be 70 billion, scratch that, 125 billion, scratch that - 180 trillion. I'd rather have pictures of crocodile Trump on the Moon than a warm house, scratch that, rather than kids, scratch that, rather than any decent future at all. Great move, guys, can't get enough of it.
It can be also just a trade part of a deal; like in YC companies, where investor Y buys company A from investor Z, and in exchange investor Z buys company B from investor Y, so the choo-choo train keeps running.
Not even close. The 2008 financial crisis is better comparison. And even then I think most negative effect will come from investors pulling money away from everything non-AI, than OpenAI/Anthropic/Oracle crashing and burning.
I'm pretty sure I could become a billionaire just by meeting Masayoshi Son and having a chat for 15 minutes. I've never seen a better case for a fool and his money being parted.
Sure, let's throw more wild amounts of money at a wildly unprofitable company with no clear roadmap to profitability any time in the near-term, and in the long-term it's very probable that any arguably-useful use cases for LLM-based technology will be a commodity anyone can run anywhere.
My gosh, this bubble can't burst soon enough. It's a form of torture to keep waiting on the pain we all know is coming…
[+] [-] redrix|1 month ago|reply
[+] [-] Ronsenshi|1 month ago|reply
What an absurd amount of money - if only this was invested in energy sector scientific research and development, or healthcare or anything else practical.
I really hoped to see compact molten salt nuclear reactors in operation before 2030.
[+] [-] ActionHank|1 month ago|reply
Love this for them.
[+] [-] bix6|1 month ago|reply
Haven’t you heard? AGI is going to solve every problem for us!!!
[+] [-] josu|1 month ago|reply
I don't understand how this is the top comment. LLMs have unlocked a lot of value for me personally, and arguably for the society as a whole. They are also one of the coolest technologies I've tried in years. As a technologist, I'm really glad that money is pouring in and allowing us to find its limits.
[+] [-] oefrha|1 month ago|reply
Incidentally that’s how much SoftBank lost on WeWork.
[+] [-] trhway|1 month ago|reply
xAI already brings gas turbines on-site, and i think the trend of on-site energy generation will grow, which will open opportunity (by providing well finaned demand) for compact/mobile/safer nuclear, and BigTech companies are among the best for any new tech development. I expect nuclear engineer positions get opened with Google and the likes :)
[+] [-] lumost|1 month ago|reply
Provided they keep cost growth slower than revenue and don't get disrupted by another model provider/commodification etc.
[+] [-] agumonkey|1 month ago|reply
[+] [-] amelius|1 month ago|reply
[+] [-] android521|1 month ago|reply
[+] [-] libraryofbabel|1 month ago|reply
1) We are in a huge investment bubble right now and it's going to burst.
2) LLMs are extremely useful right now for certain niche tasks, especially software engineering.
3) LLMs have the potential to transform our world long-term (~10 yr horizon), on the order of the transformations wrought by the internet and mobile.
4) LLM's don't lead directly to AGI (no continuous learning), and we're not getting AGI any time soon.
This is an extremely obvious point, but bears repeating. I feel the assumption of an implicit link (in both truth or falsehood) between these fairly independent assertions can cause people to talk past each other about the really important questions in play here.
Regarding The Great Bubble, I am very very bearish about OpenAI in particular. They've had a good run for three years with consumer mindshare due to their first-mover advantage, but they have no moat, trouble monetizing most of their users, not much luck building out products that stick among consumers that aren't chatbots, and their models are no better than Anthropic's, Google's, or even the best Chinese open weight models 6 months later.
My bet would be on Google and Apple together (with Gemini powering Siri, for now) destroying OpenAI in the consumer AI market over the next 2-3 years. Google has first-rate models... but more than that, both Google and Apple have the enormous advantage of owning underlying platforms that they can use to put their own AI chat in front of consumers. Google has a mobile OS, the leading browser, and search. Apple has the premium hardware and the other, premium, mobile OS. They also have the advantage of the current regulatory climate being less antitrust than it was. And they don't have to monetize their AI offerings (no ads in gemini; ChatGPT is adding them) and can run them at a loss for as long as it takes to eat up OpenAI's market share. If they partner up, as they seem to be doing, OpenAI should very very afraid.
[+] [-] WarmWash|1 month ago|reply
Don't get lost in the tech scene sauce, programming is a small sliver of what people are using LLMs for. OpenAI's report in September pegged it at ~4% of tokens being for software generation. Sure Anthropic is probably 80% or something, but only a small sliver of LLM users are using Anthropic. The reality is probably even less if you count Google's AI overviews. We hate it, but I have never seen a regular person skip over it.
The question is if regular people will pay cell phone level subscription costs ($70-$100/mo) for LLMs. If so, then we are probably not in a bubble, and the ROI will have a 5-10 yr horizon, which is totally tenable.
500,000,000 people paying $75/mo is $450B/yr. Inference is cheap too, it's training that is ludicrously expensive. Don't be fooled by the introductory pricing we have today either, that's just to get you dependent.
And yeah, chinese models, but look at what they did to tiktok. No way they are going to let the Chinese government be peoples confidants and no way is more than 0.01% of people gonna home lab.
[+] [-] ActionHank|1 month ago|reply
[+] [-] cong-or|1 month ago|reply
Data centers need power (H100s are ~700W each), and recent capacity additions were mostly pre-allocated. Chip supply is also constrained by CoWoS packaging, not fab capacity, and expansions take years.
If power, packaging, and GPUs are fixed in the near term, does $100B mostly drive inflation in AI infrastructure prices rather than materially more deployed compute? Are we seeing the real cost of a usable GPU cluster rise faster than actual capacity?
Has anyone modeled what $100B actually buys in deployable compute over the next 2–3 years given these constraints—and whether that figure is shrinking as more capital piles in?
[+] [-] stefan_|1 month ago|reply
[+] [-] randomtoast|1 month ago|reply
At this stage, why not go public? Yes, they would need to manage quarterly financial reports and answer to shareholders, but they have reached a size where they are in the top 20 range on the NASDAQ. These public companies doing well, so it seems like a logical next step.
[+] [-] edoceo|1 month ago|reply
[+] [-] dgrin91|1 month ago|reply
On the other hand they would probably set things up in a way where they still control all the voting power
[+] [-] zpeti|1 month ago|reply
[+] [-] Ekaros|1 month ago|reply
[+] [-] trolleski|1 month ago|reply
[+] [-] rvnx|1 month ago|reply
[+] [-] captain_coffee|1 month ago|reply
[+] [-] CodingJeebus|1 month ago|reply
[+] [-] zvqcMMV6Zcr|1 month ago|reply
[+] [-] rvz|1 month ago|reply
[+] [-] hnthrow0287345|1 month ago|reply
[+] [-] randomtoast|1 month ago|reply
[+] [-] jaredcwhite|1 month ago|reply
My gosh, this bubble can't burst soon enough. It's a form of torture to keep waiting on the pain we all know is coming…
[+] [-] heyaco|1 month ago|reply
[+] [-] lifetimerubyist|1 month ago|reply