3 ms·

Ed is a smart guy, but you or anyone basing your opinion on what one eloquent journalist says is ultimately a risky bet, no matter how much his reporting hits y

Ed is a smart guy, but you or anyone basing your opinion on what one eloquent journalist says is ultimately a risky bet, no matter how much his reporting hits your particular dopamine receptors. Please don't forget that Ed's entire brand identity is now 1:1 with exposing "AI" as a giant, unmitigated failure. That's a very specific flow chart to hook your caboose to when none of this is even remotely close to endgame.

yogthos 4mo ago
We don't have to take Ed's word for it. Anybody who's capable of doing grade school math can see that the numbers simply don't work. These companies are literally spending orders of magnitude more money than they're actually bringing in. Cursor, who've been renting Claude, estimated just recently that a $200-per-month Claude Code subscription could use up to $2,000 in compute. https://www.forbes.com/sites/annatong/2026/03/05/cursor-goes-to-war-for-ai-coding-dominance/ https://www.forbes.com/sites/annatong/2026/03/05/cursor-goes...
- simonw 4mo ago
  Interesting story. Here's what it says: > According to a person familiar with the company’s internal analysis, Cursor estimated last year that a $200-per-month Claude Code subscription could use up to $2,000 in compute, suggesting significant subsidization by Anthropic. Today, that subsidization appears to be even more aggressive, with that $200 plan able to consume about $5,000 in compute, according to a different person who has seen analyses on the company’s compute spend patterns. The load-bearing detail here is if that means $2,000 of internal server+electricity costs, or $2,000 if they were to charge at their API pricing instead of the subscription cost. The latter is how I understand these things to work right now. If it's the former then yeah, Anthropic are losing a TON of money on those subscriptions.
  - yogthos 4mo ago
    That's the big question, and nobody really knows what the operating costs actually are right now.
    - pama 4mo ago
      Frankly, everyone in the industry knows. When people make these statements without additional clarity they always talk about API prices. You can look at the NVL72 specs and make estimates for electricity and ownership costs rather easily. Inference at data-center scale is dirt cheap, even with public codes using dynamo and sglang. The mystery is why the early misconceptions about inefficient inference persisted even after NVIDIA was very open about everything they did to help reduce costs dramatically in the last two years.
      - yogthos 4mo ago
        I imagine it's the lack of transparency. The costs are obviously coming down as people figure out how to tune both hardware and software. But there are costs other than just electricity as well. For example, chips do burn out, I recall reading that 2 to 3 years is roughly what you can expect under inference loads, so replacing chips is a non trivial operational cost. Also, as the costs of running this stuff come down, the incentive to rent models goes down with them. Running local models has the benefit that you get to keep your data local, you can tune them to do what you like, and you're not subject to model or price changes down the road. This makes self hosting appealing both to individuals and companies. Currently, the barrier is in needing significant resources to run the models, but companies are already increasingly doing that with open models. And local inference that regular people can run is becoming a possibility as well. While I'm sure there's always going to be a market for renting out models as a service, it may shrink significantly as the costs continue to come down.
HerbManic 4mo ago
Pretty much. Ed does a lot of great work in digging through all this stuff but his conclusions always feel far too doomer oriented. OpenAL should have closed 5 times by now if you have been following his assertations from the start. There will be big parts of what he says are true once the rubble settles but it will not be anywhere near what he is predicting. How that will shape out may not be great for the average person, what money shuffling tricks will be used? But it won't be a total wreck.
- peteforde 4mo ago
  > It won't be a total wreck. Honestly, I think it's very short-sighted to assume that all of this will be seen as any kind of wreck in the long term. Normies are still catching up and reacting to chat-based LLMs. HN types are further ahead of the curve, but still catching up and reacting to agentic coding and design workflows. What often gets completely ignored is that entirely new modalities for how the underlying tech can be applied will continue to be demonstrated, and those will once again cause new ripples of excitement and disgust. There are companies building world models and systems for protein discovery. Comparatively speaking, these approaches are barely in the zeitgeist today. Deciding that we already have the data points we need to extrapolate how all of this plays out is like someone in 1974 deciding that microprocessors are just for accounting and inventory. Don't be that someone.
  - HerbManic 4mo ago
    I think the big issue isn't so much a technology thing, I mean that will improve for a long while yet, but it is an economic one. My whole concern over the rapid expansion of LLM's is the massive build out on a technology that hasn't found its feet in a big enough range of markets yet that are willing to pay top dollar. Yes, world models for protein discover is very cool stuff and I kind of hate it gets lumped in with these other companies efforts because it has a very clear path forward that doesn't rely on massive IPO's just to keep the lights on. This stuff is here to stay but I'm not sure how many of the current front runners will be able to stay solvent if they cannot turn these things in to massive money spinners. Revenue is fine-ish but spending is out of control. I see the debt in hundred of billions of dollars and start to wonder "Who is going to pay for this?" and "Will the people be willing to pay that much?". It just all feels forced rather than organic growth. This is why I think Google may end up being one of the leaders in this field. They have their custom TPU's that seem to be fairly efficient at these tasks, they are slowly but surely improving their training and inference tech using their massive data set and most importantly, other parts of the business can subsidize this stuff for a decade if needed until it is genuinely profitable. I am not against the industry but I do worry that many are rushing in with no means of genuine sustainability other than jump out for a golden parachute and let someone else clean up the mess. I do hope I am wrong.

yogthos 4mo ago

We don't have to take Ed's word for it. Anybody who's capable of doing grade school math can see that the numbers simply don't work. These companies are literally spending orders of magnitude more money than they're actually bringing in. Cursor, who've been renting Claude, estimated just recently that a $200-per-month Claude Code subscription could use up to $2,000 in compute. https://www.forbes.com/sites/annatong/2026/03/05/cursor-goes-to-war-for-ai-coding-dominance/ https://www.forbes.com/sites/annatong/2026/03/05/cursor-goes...

simonw 4mo ago
Interesting story. Here's what it says: > According to a person familiar with the company’s internal analysis, Cursor estimated last year that a $200-per-month Claude Code subscription could use up to $2,000 in compute, suggesting significant subsidization by Anthropic. Today, that subsidization appears to be even more aggressive, with that $200 plan able to consume about $5,000 in compute, according to a different person who has seen analyses on the company’s compute spend patterns. The load-bearing detail here is if that means $2,000 of internal server+electricity costs, or $2,000 if they were to charge at their API pricing instead of the subscription cost. The latter is how I understand these things to work right now. If it's the former then yeah, Anthropic are losing a TON of money on those subscriptions.
- yogthos 4mo ago
  That's the big question, and nobody really knows what the operating costs actually are right now.
  - pama 4mo ago
    Frankly, everyone in the industry knows. When people make these statements without additional clarity they always talk about API prices. You can look at the NVL72 specs and make estimates for electricity and ownership costs rather easily. Inference at data-center scale is dirt cheap, even with public codes using dynamo and sglang. The mystery is why the early misconceptions about inefficient inference persisted even after NVIDIA was very open about everything they did to help reduce costs dramatically in the last two years.
    - yogthos 4mo ago
      I imagine it's the lack of transparency. The costs are obviously coming down as people figure out how to tune both hardware and software. But there are costs other than just electricity as well. For example, chips do burn out, I recall reading that 2 to 3 years is roughly what you can expect under inference loads, so replacing chips is a non trivial operational cost. Also, as the costs of running this stuff come down, the incentive to rent models goes down with them. Running local models has the benefit that you get to keep your data local, you can tune them to do what you like, and you're not subject to model or price changes down the road. This makes self hosting appealing both to individuals and companies. Currently, the barrier is in needing significant resources to run the models, but companies are already increasingly doing that with open models. And local inference that regular people can run is becoming a possibility as well. While I'm sure there's always going to be a market for renting out models as a service, it may shrink significantly as the costs continue to come down.

HerbManic 4mo ago

Pretty much. Ed does a lot of great work in digging through all this stuff but his conclusions always feel far too doomer oriented. OpenAL should have closed 5 times by now if you have been following his assertations from the start. There will be big parts of what he says are true once the rubble settles but it will not be anywhere near what he is predicting. How that will shape out may not be great for the average person, what money shuffling tricks will be used? But it won't be a total wreck.

peteforde 4mo ago
> It won't be a total wreck. Honestly, I think it's very short-sighted to assume that all of this will be seen as any kind of wreck in the long term. Normies are still catching up and reacting to chat-based LLMs. HN types are further ahead of the curve, but still catching up and reacting to agentic coding and design workflows. What often gets completely ignored is that entirely new modalities for how the underlying tech can be applied will continue to be demonstrated, and those will once again cause new ripples of excitement and disgust. There are companies building world models and systems for protein discovery. Comparatively speaking, these approaches are barely in the zeitgeist today. Deciding that we already have the data points we need to extrapolate how all of this plays out is like someone in 1974 deciding that microprocessors are just for accounting and inventory. Don't be that someone.
- HerbManic 4mo ago
  I think the big issue isn't so much a technology thing, I mean that will improve for a long while yet, but it is an economic one. My whole concern over the rapid expansion of LLM's is the massive build out on a technology that hasn't found its feet in a big enough range of markets yet that are willing to pay top dollar. Yes, world models for protein discover is very cool stuff and I kind of hate it gets lumped in with these other companies efforts because it has a very clear path forward that doesn't rely on massive IPO's just to keep the lights on. This stuff is here to stay but I'm not sure how many of the current front runners will be able to stay solvent if they cannot turn these things in to massive money spinners. Revenue is fine-ish but spending is out of control. I see the debt in hundred of billions of dollars and start to wonder "Who is going to pay for this?" and "Will the people be willing to pay that much?". It just all feels forced rather than organic growth. This is why I think Google may end up being one of the leaders in this field. They have their custom TPU's that seem to be fairly efficient at these tasks, they are slowly but surely improving their training and inference tech using their massive data set and most importantly, other parts of the business can subsidize this stuff for a decade if needed until it is genuinely profitable. I am not against the industry but I do worry that many are rushing in with no means of genuine sustainability other than jump out for a golden parachute and let someone else clean up the mess. I do hope I am wrong.