#Anthropic与SpaceX签署845亿美元算力协议
It is not just the $84.5 billion headline figure that is striking; what that number reveals about the leading edge of the AI ​​economy is also significant.

Anthropic’s filing indicates that its potential computing commitment—involving SpaceX—could reach up to $84.5 billion through 2029, compared to the previously disclosed figure of approximately $45 billion. However, a 90-day termination clause is key: the amount cited represents potential spending under the agreement, not necessarily a guaranteed $84.5 billion expenditure.

Several points stand out:

* Computing is becoming a strategic constraint. Anthropic explicitly states that future growth could be primarily limited by the availability of computing resources.

* The scale extends beyond SpaceX. Anthropic’s disclosed infrastructure commitments now run into the hundreds of billions of dollars, with at least $518 billion projected over the next decade across six infrastructure partners.

* The business model is becoming increasingly capital-intensive. Anthropic’s 2025 revenue was approximately $4.6 billion, yet its infrastructure commitments are many times larger—creating a massive imperative for future AI demand and commercialization.

* There is a risk of circularity: while AI labs purchase vast amounts of computing power from major tech companies, some of those same companies also invest in or distribute the labs' products. Anthropic itself highlights potential conflicts and a reliance on a small group of suppliers.

Therefore, my objective view is this: the AI ​​arms race is evolving less into a race of models and more into a race for infrastructure financing. The fundamental question is not whether companies can acquire more GPUs or TPUs, but whether future AI revenue and usage can economically justify the massive commitments to computing power being made today.
$SPCX

‌
$TSLA

‌$XAGUSD

‌$SUGAR