A leaked initial public offering prospectus reviewed by Reuters shows Anthropic lost nearly $42 billion while planning about $518 billion in future computing costs, a scale that challenges basic ideas of financial prudence and market risk.
Story Snapshot
- Reuters says Anthropic’s leaked prospectus shows a $42 billion 2025 net loss.
- The document outlines about $518 billion in future cloud and infrastructure commitments.
- Revenue reportedly surged to about $4.6 billion in 2025, yet losses widened.
- The company warns investors about severe artificial intelligence risks.
What The Leaked Prospectus Says About Money Going Out
Reuters reported that Anthropic’s prospectus, seen by the outlet, disclosed a net loss of nearly $42 billion for 2025 and plans to take on about $518 billion in cloud, computing, and infrastructure obligations over coming years. The size of these commitments suggests long contracts for compute power that Anthropic believes it must secure to train and run large models. Those figures shocked many observers because they dwarf the company’s reported 2025 revenue base.
Several summaries that cite Reuters echo the same headline numbers: roughly $4.6 billion in 2025 revenue, a net loss near $42 billion, and an infrastructure plan sized at about $518 billion. Some secondary reporting adds that part of the net loss reflects non-cash accounting tied to financing instruments, while operating losses still grew past $8 billion, but the primary Reuters-based topline remains the same in public discussion. An official filing for public release has not yet appeared, according to those reports.
How The Growth Story Meets The Risk Warnings
Reuters-linked coverage says revenue jumped more than tenfold in 2025 as Anthropic sold access to its artificial intelligence models and tools. Even with that surge, the company’s spending on computing and related needs scaled faster, creating wider losses. The prospectus also reportedly warns investors that advanced artificial intelligence could pose “catastrophic or existential risks,” an unusual risk factor to place next to a massive scale-up plan. That pairing highlights the strain between bold growth targets and hard-to-measure safety costs.
Market watchers say this fits a familiar pattern in hot initial public offerings. Companies ask investors to back aggressive expansion before the unit economics are proven at scale. Research on past listings shows that many newly public firms run losses for years, and that expensive offerings can underperform later, especially when information gaps are wide. Anthropic’s reported losses, giant future obligations, and extensive risk disclosures place it squarely inside that tension.
Why This Touches Public Trust And Concentrated Power
Investors and citizens see the same theme: big bets made by a small set of leaders and lenders, with huge sums on the line and limited public oversight. The leaked numbers suggest a single artificial intelligence lab is locking in compute deals worth more than what many countries spend in a year, even as it warns its own tools could be dangerous. That mix of scale, speed, and caution fuels worries that elites will capture gains while the public shoulders the risks.
🤖 TECH: ANTHROPIC IS TARGETING A $2 TRILLION VALUATION WHILE PLANNING $518 BILLION IN COMPUTING AND INFRASTRUCTURE SPENDING, ACCORDING TO ITS IPO PROSPECTUS. pic.twitter.com/kqgoUIP2T9
— KALIVARY (@KALIVARYX) September 29, 2026
Skeptics on both the right and left question whether government gatekeepers, market regulators, and boards are strong enough to police this rush. Supporters point to rapid revenue growth and say pre-paying for compute can win a long race. Both views can be true at once. But the core facts from Reuters’ review are simple: losses are huge, obligations are larger, and the company itself flags serious risks. That puts clarity, governance, and accountability at the center of the coming debate.
Sources:
zerohedge.com, reuters.com, startupfortune.com, cnbc.com, x.com, ijr.com
