Anthropic, a leading AI lab, is set to spend at least $518bn over the next decade building AI infrastructure with six major partners. This massive investment, revealed in a confidential IPO prospectus, ranks among the largest AI buildout commitments on record. The company has outlined plans to work with Alphabet's Google, Amazon, Microsoft, and others to secure the necessary computing power for its AI development. This move comes as Anthropic faces increasing demand for advanced AI systems.
According to the prospectus, about 80% of the $518bn investment is non-cancellable, requiring payment regardless of usage. Anthropic has committed to spend $111.1bn with Google, $110bn with Amazon, and $31.4bn with Microsoft over the next seven to 10 years. These long-term infrastructure service obligations will ensure Anthropic's access to critical computing power. The company also has $161.2bn in Broadcom-related equipment lease obligations that are largely non-cancellable.
Anthropic's massive investment in AI infrastructure is driven by the company's expectation that future demand for advanced AI systems will exceed available supply. The company believes that access to computing power is becoming the main constraint on AI development. To mitigate this risk, Anthropic is securing long-term commitments from major tech giants. This strategy allows Anthropic to focus on developing its AI capabilities while ensuring a stable supply of computing power.
The prospectus also reveals that Anthropic has agreements with Elon Musk's SpaceXAI that could result in up to $84.5bn of spending for Nvidia-based computing capacity through 2029. However, this commitment is largely cancellable with a 90-day notice. Additionally, Anthropic has deepened its relationship with Advanced Micro Devices, which has committed to buying up to $5bn of Anthropic stock and supplying AI computing capacity expected to exceed $20bn.
Anthropic's shift towards building its own AI infrastructure marks a significant change in its strategy. The company is moving away from a cloud-only model and towards dedicated data centres and directly leased chips. This approach allows Anthropic to have more control over its AI infrastructure and reduce its dependence on third-party providers. However, the company still relies heavily on Amazon, Google, and Microsoft, which provide computing infrastructure, distribution, and develop rival AI models.
The company's dependence on these tech giants poses a risk to its business, as their incentives "may not be fully aligned" with Anthropic's interests. If Anthropic's access to computing power from third-party providers is curtailed, repriced, or terminated, its business, financial condition, and results of operations could be adversely affected. Anthropic's confidential IPO filing with the Securities and Exchange Commission in June has not been publicly disclosed.
The scale of Anthropic's AI infrastructure plan is comparable to OpenAI's Stargate project, a $500bn AI infrastructure plan whose costs are expected to be shared among OpenAI, SoftBank, Oracle, and MGX. As Anthropic continues to develop its AI capabilities, its massive investment in infrastructure will be crucial in meeting the growing demand for advanced AI systems. The company's progress will be closely watched by industry stakeholders and investors.
Key points
- Anthropic plans to spend $518bn over a decade on AI infrastructure with six partners.
- 80% of the investment is non-cancellable, requiring payment regardless of usage.
- The company's dependence on major tech giants poses a risk to its business.
