Anthropic’s planned $518 billion AI infrastructure buildout is emerging as one of the largest technology spending commitments on record, but the structure of those obligations may matter as much as the headline figure. According to a confidential IPO prospectus reviewed by Reuters, roughly 80% of the commitments are non-cancelable or require payment regardless of actual usage, placing significant financial obligations behind the company’s expectations for future AI demand.

Long-Term Contracts Create Significant Fixed Commitments

Anthropic expects to spend at least $111.1 billion with Alphabet’s Google, $110 billion with Amazon and $31.4 billion with Microsoft under infrastructure agreements extending between seven and 10 years. The contracts with Google and Amazon require Anthropic to pay committed amounts even when actual usage falls short, while the Microsoft agreement is non-cancelable except in specified circumstances involving an uncured material breach.

The company also disclosed approximately $161.2 billion in Broadcom-related equipment lease obligations that are largely non-cancelable. Together, these arrangements mean a substantial portion of Anthropic’s future infrastructure expenditure is effectively locked in, reducing the flexibility to scale spending down if computing demand, pricing or AI economics develop differently from current expectations.

Anthropic Is Betting That Compute Will Remain Scarce

Anthropic’s rationale is based on the expectation that access to computing capacity will become a fundamental bottleneck for increasingly sophisticated AI systems. The company told prospective investors that future demand for advanced AI could exceed available supply and that development will be constrained principally by the availability of compute. This represents a strategic shift from securing computing resources when needed toward reserving capacity years in advance.

The company is also moving beyond a cloud-only model toward dedicated data centers and directly leased computing equipment. Separate agreements with xAI could generate as much as $84.5 billion in Nvidia-based computing spending through 2029, although those arrangements are largely cancellable with 90 days’ notice. Anthropic has also deepened its relationship with AMD, which committed to purchase up to $5 billion of Anthropic stock while providing more than $20 billion in expected computing capacity.

The Financial Risk Extends Beyond Anthropic

The scale of these commitments has implications across the broader AI infrastructure ecosystem. Google, Amazon and Microsoft are simultaneously infrastructure providers, investors, distributors and competitors to Anthropic, creating a complex commercial structure that the company itself identified as a potential risk. Anthropic warned that if third-party compute were curtailed, repriced or terminated, its business and financial results could be adversely affected.

The broader financing environment also matters. The rapid expansion of AI infrastructure is increasing demand for semiconductors, data centers, networking equipment and power capacity, while simultaneously raising questions about the amount of capital required before AI investments generate sufficient economic returns. For infrastructure providers, Anthropic’s commitments provide visibility into future demand; for Anthropic, however, they create substantial fixed obligations that must ultimately be supported by revenue growth and sustained utilization.

For investors in Israel and global markets, the next phase of Anthropic’s story will therefore be measured not simply by model development or IPO valuation, but by capacity utilization, revenue growth, infrastructure economics and contractual flexibility. How quickly AI demand translates into sustainable cash flows will be particularly important because a large portion of the company’s infrastructure spending cannot easily be reduced if market conditions change.