Remember "worse is better". The model doesn't have to be the best; it just has to be mostly good enough, and used by everyone -- i.e., where switching costs would be higher than any increase in quality. Enterprises would still be on Java if the operating costs of native containers weren't so much cheaper.
So it can make sense to be ok with losing money with each training generation initially, particularly when they are being driven by specific use-cases (like coding). To the extent they are specific, there will be more switching costs.
This says absolutely nothing.
Extremely simplified example: let's say Sonnet 4.5 really costs $17/1M output for AWS to run yet it's priced at $15. Anthropic will simply have a contract with AWS that compensates them. That, or AWS is happy to take the loss. You said "they won't be doing that at a loss" but in this case it's not at all out of the question.
Whatever the case, that it costs the same on AWS as directly from Anthropic is not an indicator of unit economics.
Is the bottleneck primarily capex, long lead times on power and GPUs, or the strategic risk of locking into fixed infrastructure in such a fast-moving space?
But the max 20x usage plans I am more skeptical of. When we're getting used to $200 or $400 costs per developer to do aggressive AI-assisted coding, what happens when those costs go up 20x? what is now $5k/yr to keep a Codex and a Claude super busy and do efficient engineering suddenly becomes $100k/yr... will the costs come down before then? Is the current "vibe-coding renaissance" sustainable in that regime?
See people get confused. They think you can charge __less__ for software because it's automation. The truth is you can charge MORE, because it's high quality and consistent, once the output is good. Software is worth MORE than a corresponding human, not less.
And why is that? Should they not be interested in sharing the numbers to shut up their critics, esp. now that AI detractors seem to be growing mindshare among investors?
The interesting question is if they are subsidizing the $200/mo plan. That's what is supporting the whole vibecoding/agentic coding thing atm. I don't believe Claude Code would have taken off if it were token-by-token from day 1.
(My baseless bet is that they're, but not by much and the price will eventually rise by perhaps 2x but not 10x.)
Can you provide some numbers/sources please? Any reporting I’ve seen shows that frontier labs are spending ~2x on inference than they are making.
Also making the same query on a smaller provider (aka mistral) will cost the same amount as on a larger provider (aka gpt-5-mini) despite the query taking 10-100x longer on OpenAI.
I can only imagine that is OpenAI subsidizing the spend. GPUs cost by the second for inference. Either that or OpenAI hasn’t figured out how to scale but I find that much less likely
First: marginal inference cost vs total business profitability. It’s very plausible (and increasingly likely) that OpenAI/Anthropic are profitable on a per-token marginal basis, especially given how cheap equivalent open-weight inference has become. Third-party providers are effectively price-discovering the floor for inference.
Second: model lifecycle economics. Training costs are lumpy, front-loaded, and hard to amortize cleanly. Even if inference margins are positive today, the question is whether those margins are sufficient to pay off the training run before the model is obsoleted by the next release. That’s a very different problem than “are they losing money per request”.
Both sides here can be right at the same time: inference can be profitable, while the overall model program is still underwater. Benchmarks and pricing debates don’t really settle that, because they ignore cadence and depreciation.
IMO the interesting question isn’t “are they subsidizing inference?” but “how long does a frontier model need to stay competitive for the economics to close?”