Anthropic’s Cloud Capacity Challenge Is Becoming More Complicated

Preface

Anthropic’s rapid growth is creating a difficult infrastructure problem: the company needs vast amounts of computing capacity, but its largest potential suppliers are also increasingly viewed as competitors. Details in a leaked S-1 filing, reported by The Wall Street Journal, add context to figures previously reported by Reuters and the Financial Times in late September. The filing describes Anthropic’s potential cloud spending, its efforts to secure capacity beyond major cloud providers, and concerns about the reliability of some alternatives. It also highlights financial concentration and potential liability associated with harmful AI behavior. Together, these disclosures sketch a company balancing ambitious growth with significant operational and financial exposure. The central issue is not simply how much Anthropic can spend, but whether it can secure dependable infrastructure while managing the risks that come with its expansion.

Lazy bag

Anthropic is reportedly struggling to obtain enough cloud capacity from AWS, Google, and Microsoft as those providers increasingly regard the Claude maker as a competitor. Its S-1 filing suggests cloud contracts could exceed $750 billion, including agreements still under negotiation. The company is also leasing data centers and working with third-party operators, though it questions their limited operating histories. Financial disclosures include an $80.7 billion net loss and $16.3 billion in revenue in the first half of the year, while two customers generated about a fifth of sales. The filing also warns of potential AI-related damages beyond available insurance coverage.

Main Body

Anthropic’s demand for computing resources is placing its relationships with major cloud providers under growing strain. The company relies on large-scale infrastructure to develop and operate its AI systems, including Claude. Yet according to an S-1 filing leaked to The Wall Street Journal, Anthropic is having trouble securing enough capacity from AWS, Google, and Microsoft. These providers have historically served as cloud suppliers to AI companies, but the filing suggests they increasingly see Anthropic as a competitor rather than simply a customer.

This shift creates a practical challenge. Anthropic needs access to substantial and dependable computing capacity, while its suppliers may also be investing in their own AI products and services. The filing does not establish that any provider has ended its relationship with Anthropic. Instead, it describes difficulty obtaining enough capacity, a problem that may become more consequential as the company grows. When a supplier also competes in the same market, negotiations over availability and long-term commitments can become more complex.

The leaked filing adds detail to financial figures that Reuters and the Financial Times reported in late September. It says Anthropic believes it could spend more than $750 billion on cloud contracts. Some of the potential commitments are still being negotiated, so the amount should not be read as a completed or finalized set of contracts. Even so, the figure conveys the scale of infrastructure spending the company may be considering to support its operations and future needs.

That possible spending does not appear to resolve the capacity problem on its own. Anthropic is also leasing data centers and working with third-party data center operators. This approach could provide additional infrastructure beyond what it can obtain directly from the major cloud providers. It also signals that the company is exploring a broader mix of ways to secure computing resources rather than relying exclusively on hyperscalers.

However, Anthropic’s filing expresses concern about some of these alternatives. The company describes third-party operators as having a “limited operating history,” a factor that could make them less reliable than established hyperscalers. That distinction matters because AI operations depend not only on the amount of computing equipment available, but also on whether infrastructure can be operated consistently. The filing presents this as a concern, not proof that third-party operators have failed to meet their obligations. Still, it illustrates the trade-off Anthropic faces: broaden its sources of capacity while assessing the dependability of less-established providers.

The infrastructure challenge sits alongside substantial financial figures disclosed in the filing. Anthropic recorded an $80.7 billion net loss in the first half of the year, while reporting $16.3 billion in revenue during the same period, according to The Wall Street Journal’s account. These numbers show a large gap between reported revenue and net loss. They also underline the financial scale of the company’s expansion and the costs associated with building and operating AI systems. The filing’s discussion of potential cloud commitments adds another dimension to the question of how Anthropic will fund future infrastructure needs.

Revenue concentration is another point of exposure. According to the report, only two customers accounted for about a fifth of Anthropic’s sales during the first half of the year. The figure does not identify those customers in the provided account, and it does not mean the remaining revenue came from only a small number of buyers. It does, however, indicate that a meaningful portion of sales depended on two customers. For a rapidly growing company, customer concentration can make revenue more sensitive to the purchasing decisions of a limited number of accounts.

Anthropic has also shared a more recent revenue figure with prospective investors. Last week, the company said third-quarter revenue alone had reached $16.8 billion, up 45% from the previous quarter. This reported quarterly revenue provides a separate indicator of growth from the first-half figures. It does not, by itself, resolve questions about profitability, infrastructure availability, or the stability of customer demand. Read together, the figures show strong reported revenue growth alongside substantial losses and significant prospective infrastructure spending.

The filing also addresses risks associated with the behavior of AI systems. Building on previously reported catastrophic risk disclosures, it warns that Anthropic could be held liable for damages caused by harmful AI behavior. The potential damages, according to the filing, could exceed the company’s available insurance coverage. This is a risk disclosure rather than a prediction that a particular harmful incident will occur or that a specific claim will be made. Its significance lies in the possibility that legal and financial exposure could extend beyond the protection available through insurance.

These issues are relevant to a reported potential public offering. Anthropic is reportedly targeting a November IPO. The filing’s discussion of cloud capacity, financial results, customer concentration, and liability risks offers prospective investors a view of both the company’s growth prospects and the challenges attached to them. An IPO target does not confirm that an offering will occur on that schedule, and the provided information does not describe final offering terms. It does indicate that these operational and financial disclosures are emerging as the company considers a public-market debut.

Overall, Anthropic’s situation illustrates how growth in AI can turn infrastructure access into a strategic constraint. Major cloud providers may be essential suppliers while also pursuing competing businesses; alternative data center arrangements may add capacity but raise reliability questions; and large financial commitments must be considered alongside losses, revenue concentration, and potential legal exposure. The filing does not suggest that any one challenge has a simple solution. Instead, it presents a company seeking to scale quickly while managing dependencies and risks across several parts of its business. Anthropic did not immediately respond to TechCrunch’s request for comment.

Key Insights Table