Anthropic faces $750 billion in compute commitments as revenue surges

Explosive growth meets heavy debt

Anthropic is experiencing a period of extreme financial growth as businesses integrate its Claude models into core operations. As reported by Fortune, the company saw its first-half revenue soar 1,300% to $16.28 billion. This growth is primarily driven by enterprise customers who pay based on how much they actually use the AI services.

Despite this success, the company's S-1 filing reveals a massive looming financial obligation. Anthropic has committed to at least $475 billion in future cloud compute purchases with partners like Amazon, Google, and Microsoft. When the company includes agreements to lease computing chips, those total commitments reach as high as $750 billion.

The company currently makes a return of over $2 for every $1 spent on computing infrastructure. However, the filing warns that this margin could shrink if Anthropic is forced to lower its prices to stay competitive. The company must now ensure that its sales growth can continue to outpace these fixed, multi-billion dollar costs.

The infrastructure mismatch

A primary risk identified in the filing is the mismatch between how Anthropic buys power and how it sells its products. The company typically contracts for computing capacity 12 to 18 months in advance to ensure it has the hardware ready for future demand. This means Anthropic is often paying for data centers and GPUs before the customers have even signed up to use them.

On the other side of the ledger, the majority of Anthropic's revenue comes from consumption-based fees rather than guaranteed long-term contracts. Many of these customers can terminate their services or reduce their usage with very little notice. This creates a situation where the company is locked into spending billions on hardware while its income remains subject to the whims of the market.

Anthropic has admitted that it has underestimated its compute requirements in the past, which led to periods where it was capacity-constrained. The company notes that the timing of new hardware coming online may not always align with when customers actually need it. This can result in periods of underutilized compute where the company is paying for electricity and hardware that isn't producing immediate revenue.

Securing long-term commitments

To mitigate the risk of fluctuating demand, Anthropic is attempting to shift its business model toward more stable revenue streams. The company is increasingly moving enterprise clients into non-cancelable, multi-year deals that include fixed minimum spends. In exchange for these guarantees, Anthropic offers customers discounted rates for their AI tokens.

As of June 2026, the company reported about $10.8 billion in remaining performance obligations under these specific types of contracts. However, the filing notes that 75% of that revenue will be recognized within the next year. This is a relatively short window compared to the long-term timeframe of its massive data center spending commitments.

The company is also navigating its identity as a public benefit corporation, which requires it to balance shareholder interests with a public mission. Because of this structure, Anthropic stated it wants to avoid the pressure to optimize for user engagement through advertising. Instead, it is focused on growing its customer base rapidly enough to generate the cash flow needed to satisfy its infrastructure debts.

The broader industry landscape

Anthropic's situation reflects a wider trend among tech giants who are racing to build out the physical backbone of the AI era. Major players like Amazon, Microsoft, Alphabet, and Meta are collectively expected to spend over $700 billion on capital expenditures in 2026. Much of this massive spending is being financed through debt to fund the construction of new data centers.

Other companies in the space are following similar aggressive expansion paths to keep up with demand. For example, Oracle recently announced it would need to raise between $45 billion and $50 billion this year alone to fund its own infrastructure growth. These figures illustrate the sheer scale of capital required to maintain a lead in the large language model market.

While Anthropic currently holds $112.4 billion in cash and cash equivalents, it warns that its build-out will require significant upfront payments. These payments must be made before the computing capacity actually comes online, which could strain cash reserves. The company's ability to navigate this period will depend on whether the broader economy continues to support high spending on AI services.