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.
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.
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.
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.
No question that there are some players who need a good weekend in Vegas to make it across the chasm of sorrow.
It could happen, I hope it does.
> 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.
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.