Artificial intelligence startup Anthropic filed its initial public offering prospectus on Monday, unveiling a corporate blueprint marked by hundreds of billions of dollars in infrastructure obligations, an unconventional governance structure, and an extensive catalog of existential risks.
The developer of the Claude chatbot is preparing to debut on public markets between Nov. 9 and Thanksgiving, Fortune reported, in a listing analysts expect could value the company at more than $1 trillion. The 335-page regulatory filing offers the first broad window into Anthropic’s financial performance, hardware architecture, and corporate hierarchy.
Multi-chip strategy and ballooning compute
In the first six months of 2026, Anthropic generated $3.9 billion in free cash flow and spent $5.6 billion on computing and infrastructure, Fortune reported. The company recorded an adjusted operating margin of 19 percent in its most recent reported quarter, though that figure excluded stock-based compensation, charitable donations, and author copyright litigation settlements.
Anthropic disclosed that it manages compute capacity across four distinct hardware families: Amazon Trainium, Google TPUs, Advanced Micro Devices GPUs, and Nvidia GPUs, according to Business Insider. Treating these processors as a single interchangeable pool gives the company flexibility that analysts say could challenge Nvidia’s pricing power over time. Patrick Moorhead, chief executive of Moor Insights & Strategy, told Business Insider that customers capable of shifting workloads across multiple platforms negotiate far differently than single-vendor buyers.
Securing that capacity has required massive commitments. Fortune reported the company has lined up commitments that could reach $808 billion, including $110 billion with Amazon, $115 billion with Google, $121 billion with SpaceX, $45 billion with Microsoft, and a $161 billion leasing agreement with Broadcom.
Baroque board structure shuts out public
While Anthropic seeks public capital, public investors will wield virtually no influence over corporate direction, TechCrunch reported. Organized as a public benefit company, Anthropic has split oversight among its seven founders, an independent trust, and common stockholders.
The company’s Long-Term Benefit Trust, an independent body created to prioritize public benefit, holds the authority to elect and remove a majority of corporate directors, known as Class T seats, TechCrunch reported. The trust is led by global health specialist Neil Shah and includes former Federal Reserve Chairman Ben Bernanke, national security analyst Richard Fontaine, and Harvard University physicist Christopher Stubbs. Meanwhile, Anthropic’s co-founders retain 50.1 percent of shareholder voting power to pick three board seats. Strategic backers Amazon and Google hold non-voting Class N shares, while common Class A stock sold on the market holds minimal influence unless founders split or special classes sunset.
Sprawling catalog of hazards
Nearly a third of the prospectus—93 pages—outlines operational and societal dangers, CBS News reported. Beyond standard business risks, Anthropic warned of extreme outcomes associated with advanced frontier artificial intelligence, including systems resisting human shutdown commands, attempting blackmail to maintain operation, or causing catastrophic harm.
The company also cited vulnerabilities stemming from recursive self-improvement, where automated systems iterate on their own code faster than human operators can oversee them. In addition, Anthropic noted external hurdles, such as community moratoriums on data centers over electrical and water strains, potential nationalization by governments, and liabilities so immense that the insurance industry currently lacks the capacity to underwrite them.
Whether Anthropic can maintain its rapid sales trajectory while funding unprecedented hardware pledges remains the central question for prospective investors before trading commences next month.
