Table of Contents
AI startup Anthropic has released its S-1 filing for an upcoming public offering, featuring multiple blank spaces where key financial figures should be. While companies are permitted to redact certain confidential information under Securities and Exchange Commission rules, the missing data in Anthropic’s prospectus highlights the ongoing uncertainties of investing in artificial intelligence as firms race to secure adequate compute power.
Reasons Behind Missing Financials and Leases
Michael Dambra, an accounting and law professor at the University at Buffalo School of Management, told WIRED that there are practical explanations for the absent third-quarter figures. Because the period ended on September 30, closing the books for a major enterprise takes time. Furthermore, companies face liability for registration misstatements, creating a strong incentive to fully vet numbers before publication.
Dambra noted that SEC rules will require Anthropic to update the filing with third-quarter financials if it goes public later than mid-November. Regarding redacted data center lease agreements, Erica Neuman, an accounting professor at Youngstown State University, told WIRED that omitting confidential details from contract exhibits is not unusual.
However, Neuman emphasized that analysts must scrutinize whether these redactions hinder investors from properly evaluating Anthropic’s financial obligations and potential impacts on company cash flow.
Compute Costs and Infrastructure Risks
Compute represents Anthropic’s most significant operating expense and business risk, encompassing both potential undersupply and overinvestment. The prospectus highlights the danger of overinvestment, noting that the firm must make forward investments in resources that may initially appear irresponsibly large.
Chips are treated as a fungible pool sourced from strategic partners including Amazon, Google, Nvidia, and AMD. Anthropic reported entering contracts with an aggregate committed spend exceeding $____ for compute and infrastructure, with completion dates ranging from before December 2030 through 2045.
Despite mentions of supply uncertainties linked to limited land and power resources, Patrick Moorhead, CEO and chief analyst at Moor Insights & Strategy, told WIRED that investors deserve a more granular characterization of the risks.
“It’s about characterizing the odds and impact individually across each of the areas. Anthropic owes investors more information, such as a probability of these individual issues, characterizing the percentage chance and the degree of the impact,” Moorhead said, adding that broader statements are insufficient.
Future Obligations and Valuation Impacts
Footnotes in the prospectus reveal massive estimated future financial commitments. Since December 31, 2025, the company has initiated or amended compute and infrastructure leases totaling approximately $417 billion in future minimum payment obligations, alongside an additional $245 billion added since June 30.
Anthropic also expanded unfinalized commercial agreements by $125 billion to $150 billion and holds $21 billion in short-term agreements with minimum non-cancelable payments. In total, the company faces at least $808 billion in minimum potential payment obligations if all agreements proceed.
Ben Bajarin, chief executive and principal analyst at Creative Strategies, expects many of these gaps to be filled in subsequent drafts of the prospectus.
While redactions are common, Dambra described them as controversial, noting that skeptics often accuse firms of using them opportunistically to hide sensitive information. Dambra also cautioned that companies with S-1 redactions frequently end up underpricing their shares at their initial public offering.
Because investors treat information gaps with suspicion, companies often lower their starting share price to entice participation. Bajarin noted that the broader artificial intelligence industry currently operates with similar unknowns regarding chip and infrastructure access.
“All of this,” Bajarin said, “just shows how absurdly early we are in enterprise adoption of an AI market that will be more than $3 trillion yearly and likely growing.”
