So $448 billion committed, with unclear cancellation terms, through the next ~48 months.
Clearly this will be rising over time, but even a naive division results in $10bn per month of compute spending.
That honestly doesn't seem so bad relative to some of the claims I've seen? Their Q3 revenue is about $5.6bn per month and it's rising. They have so much money in the bank waiting to be deployed that they get a billion dollars of interest per year. Investors are desperate to put more money into the company.
They spend a lot on compute, but rolling it up into cumulative spend numbers makes it sound scarier.
That’s wild. They have committed to absolutely insane growth with no margin for error.
It just seems like, given what we know about how things have worked so far in the AI space, that paying for the entirety of these agreements is not likely.
Their 2026 revenue looks like it will be ~$67bn so clearly they're anticipating continued significant revenue growth for this to make sense.
Or as Anthropic put it on page 42 of their S-1: https://www.sec.gov/Archives/edgar/data/2130357/000162828026...
> If we are unable to differentiate our models, platform, and applications from competing offerings through enhanced functionality, superior performance, better safety features and reliability, lower cost, or other factors, demand for our models, platform, and applications may decline. The rapid pace of technological change in AI means that models, platforms, and applications can quickly become outdated. We must continuously invest in research and development to maintain our competitive position, but there is no guarantee that our investments will result in models, platforms, and applications that customers prefer over competing alternatives.
My thought is that any competition to Anthropic is going to need to actually have the compute to serve all of Anthropics customers. Chinese labs certainly do not. It seems to be the crux of all these compute agreements.
Even then, Anthropic has the luxury of knowing exactly what hardware (even what datacenter!) will serve their models. This gives them the ability to co-design their model for the exact chips it will run on, something that Open Weights models are completely unable to do - even to the point that Deepseek fully open-sourced their inference engine, revealing that they're still running on only H100s and Huawei Ascend chips (https://github.com/deepseek-ai/DeepInfer/blob/main/README.md).
Anthropic will be running their 2027 Opus and Fable models on Vera Rubin, a chip which produces tokens for about 1/100th the cost of the H200. Seriously. 100x more efficient on GLM 5.3: https://inferencex.semianalysis.com/inference/glm-5-3 compared to what DeepSeek is able to use for inference.
Combined with Kimi K3.1's underperformance I'm honestly not too bullish on Chinese AI being able to catch back up. As an example, according to https://artificialanalysis.ai/, Kimi K 3.1 would need to be run 10x cheaper than its current inference cost in order to hit the Paraeto Frontier (matching 6.1 Sol Medium)
One of Anthropic's most interesting technical moats is that they can serve their models on NVIDIA hardware or Google TPUs or AWS Trainium - they talked about that in an incident postmortem last September: https://www.anthropic.com/research/a-postmortem-of-three-rec...
At this point I think the goal is:
Step 1: Go public
Step 2: Allow exit for early investors
Step 3: file bankruptcy
Step 4: Amazon or Google “rescues” the company at pennies to the dollar on the already heavily diluted valuation.
If they grow like no startup in history they just eek by and manage to barely pay their bills. Impressive, but not worth anything close to $1T plus valuation.
If they achieve AGI money becomes worthless.
So where’s the upside for investors here?
I mean, I can't reveal that because I'm just some schmuck. But I'd like to.
If you make it through the devastation, you'll be one of the richest and most important people in the world. If not, throw a random frontman in front of the wolves and disappear into the background.