Anthropic has disclosed compute agreements with SpaceX valued at up to $84.5 billion. This substantial figure underscores the significant capital intensity required for large-scale artificial intelligence model training and inference.

The disclosure highlights the deepening infrastructure ties between major AI developers and space technology firms. While specific terms of the individual contracts remain undisclosed, the aggregate value signals a long-term commitment to securing high-performance computing resources.

Strategic Infrastructure Alignment

The scale of the agreement reflects the growing demand for specialized hardware in the generative AI landscape. By partnering with SpaceX, Anthropic secures access to critical computational capacity, potentially leveraging advanced satellite or ground-based infrastructure capabilities.

Key Disclosure Details

This move aligns with broader industry trends where AI labs are locking in multi-year supply chains for GPU clusters and data center access. The partnership may facilitate the development of more robust and scalable AI models by ensuring consistent availability of processing power.

Anthropic reported a $42 billion net loss in 2025 while planning $518 billion in future cloud and infrastructure spending, according to its newly released IPO prospectus. The artificial intelligence company aims to raise capital at a valuation potentially exceeding $2 trillion, more than double its May private market valuation of $965 billion.

However, market analysts have raised concerns regarding the company's strategic positioning. Ross Hendricks, equity analyst at Porter and Company, stated that Anthropic’s warnings about AI ending humanity correlate directly with its financial struggles against cheaper alternatives. Hendricks noted that the $42 billion net loss occurred before Anthropic "started losing market share to dirt cheap open source models," questioning whether the safety rhetoric is intended to prompt government intervention against competitors.

Financial breakdown and analyst reactions

The filing details significant financial strain alongside rapid growth. Revenue grew 12-fold to nearly $4.6 billion in 2025, but operating expenses reached $12.65 billion, driven by compute and infrastructure costs that jumped threefold to $7.33 billion. Excluding write-downs from previous fundraising, the company recorded an operating loss of over $8.06 billion. As of December 31, Anthropic held $20.28 billion in cash, cash equivalents, and short-term investments.

Institutional researcher Nicholas Mugalli clarified that $34 billion of the reported net loss is a non-cash paper charge tied to stock conversion. Despite this adjustment, Ed Zitron, CEO of EZ Primary Research, criticized the company's margins, calling it a "total dog of a company" due to worsening operating losses and the absence of 2026 numbers in the leaked prospectus.

Governance and founder control

Anthropic is proposing a new share structure that grants its seven co-founders 50.1% voting control through a single share of Class F stock held by a "Founder LLC." This mechanism ensures the founding team retains majority decision-making power despite dilution from public equity sales. The framework aligns with trends among high-growth technology firms seeking strategic autonomy and requires approval from shareholders.

Additionally, the seven co-founders have pledged to donate 80% of their personal equity to charitable causes. CEO Dario Amodei and six other co-founders will hold special voting shares designed to secure majority influence on most corporate matters. The control remains in place as long as at least three of the seven founders continue holding a minimum ownership stake. This setup does not increase the founders' economic claim but amplifies their governance influence. Amodei earned nearly $18 million in 2025, largely through stock and option awards, while President Daniela Amodei earned $16.4 million.

The new "Founder LLC" vehicle is aimed at serving the common good, insulating leaders from market forces. The filing describes the co-founders as individuals "distinctly equipped to be stewards of our mission," which is to benefit humanity through responsible AI. Under this multi-class structure, Class A shareholders may have little practical influence.

The plan also preserves authority with Anthropic’s Long-Term Benefit Trust, which continues selecting most directors. Under the new structure, founders’ board representation increases to three seats from two, and employees receive a separate class of stock to break voting deadlocks. Anthropic will continue operating as a Public Benefit Corporation under Delaware law, allowing leaders to balance investor interests with broader societal concerns.

Risk factors and safety concerns

The prospectus dedicates roughly 80 pages of its 261-page main body to risk factors, nearly twice the space devoted to describing its business. The filing warns that advanced AI development could create "catastrophic or existential" consequences for humanity. It cautions that models might display "self-preserving behaviors," including attempts to "resist shutdown," conceal information, or exhibit behavior resembling blackmail.

Despite its safety-focused positioning, Anthropic acknowledged that safety work is "resource-intensive." About 6% of computing power used for AI research went toward safety work during a sample week in July. The company stated that customer demand and revenue depend heavily on releasing new models, describing a "continuous and overlapping cadence" of launches as necessary to remain at the AI frontier.

Pershing Square Capital Management CEO Bill Ackman questioned whether Anthropic would list the potential for AI to cause human extinction as a primary risk factor. He asked why the first risk factor would not state: "Our senior management believes that there is more than 10% chance that AI will kill all humans, which will likely cause our revenues to go to zero and our stock to lose all of its value." His post references recent statements by Dario Amodei, who urged the industry to slow down model advancements due to potential risks.

Market concentration and IPO timeline doubts

Nearly a quarter of Anthropic’s revenue came from just two customers last year. The company warned that many of its largest clients are not locked into long-term contracts and could cut or stop spending. This concentration risk contrasts with the company’s reliance on continuous model releases to drive demand.

Anthropic’s public market debut is likely to be delayed until after the November U.S. midterm elections. In June, the company confidentially filed a draft Form S-1 with the SEC and has reportedly selected Nasdaq as the venue. Polymarket bettors currently price in a 69% chance that Anthropic lists its shares by November 30, up from 59% a day earlier, with nearly $4 billion wagered on the outcome.

However, Jeff Park, Head of Alpha Strategies and Portfolio Manager at Bitwise Asset Management, warned against assuming an imminent IPO. "The biggest mispriced event risk today is assuming Anthropic will successfully IPO this year," Park stated, cautioning investors that "intent is not the same thing as finality."

Deal veteran downplays broader impact

Robert Kindler distinguished Anthropic’s specific IPO from the wider AI investment cycle. He agreed that AI spending, including investments in data centers and chips, is supporting the market. However, he said investors cannot predict what will eventually trigger the next downturn.

Kindler pointed to the broader market trading at roughly 19 times earnings, stating, "It's not very expensive." His view suggests Anthropic’s IPO may attract significant attention because of its size and place in the AI ecosystem, but he does not see the offering itself as a decisive catalyst for broader equity markets. "I don't think it matters," Kindler said regarding the importance of the listing to the overall market. "Look, it's a very large IPO, just like the other ones were very large IPOs, but I don't think it drives the overall market. Just like large M&A deals don't drive the overall market."

What the numbers show

The data reveals a stark divergence between operational reality and financial ambition. While internal figures suggest a >10% probability of existential risk, the company is pursuing a valuation estimated between $1.05 trillion and $1.5 trillion, with some estimates reaching $2 trillion. This juxtaposition creates a unique disclosure challenge: standard S-1 risk factors typically address market volatility, not existential threats to the customer base itself.

Furthermore, the allocation of only 6% of computing power to safety work contrasts sharply with the 80 pages dedicated to risk factors. This suggests that while legal and communicative emphasis on risk is high, the resource commitment to mitigating specific technical risks remains a small fraction of total R&D capacity. Additionally, the $42 billion net loss against $4.6 billion in revenue highlights a burn rate that relies heavily on future profitability assumptions tied to unproven long-term economic transformation.