Tech • AI • Robotics • Game
$ANT, Yasmin 🐐, Sneaky Qwen, Samsung Controversy, IGTV
9.5/10TL;DR
Anthropic has filed for an IPO with surging revenue, large cash reserves, improving margins and an unusually strong founder-control structure, reinforcing investor confidence in the AI sector.
KEY POINTS
Massive scale before listing
Anthropic filed to go public without disclosing final deal size or valuation, though market estimates put the raise near $100 billion and the company’s value between $1.5 trillion and $2 trillion. If priced near the top of that range, it would rank among the largest IPOs ever attempted. The shares are expected to trade under the ticker ANT.
Revenue growth exceeds expectations
The filing showed $67.2 billion in annualized revenue, up 45% quarter over quarter, confirming earlier investor-day figures. That pace is exceptional at this scale and helped calm concerns that growth might be slowing. Investors broadly viewed the numbers as solid rather than speculative, with no major negative surprise emerging from the filing.
Huge cash position and limited burn
As of mid-2026, Anthropic reported $131.7 billion in cash and short-term investments. The company had raised about $141 billion in total capital, implying cumulative burn of roughly $9.8 billion, a remarkably low figure relative to its size. That has fueled claims that it may be one of the most capital-efficient businesses ever built.
Compute strategy relies on renting, not owning
Despite the scale of its AI operations, the company spent only $482 million on capital expenditures in the first half of the year. Instead of building out massive owned infrastructure, it has leaned heavily on rented compute. That approach lowers upfront spending while preserving flexibility as demand and model requirements shift.
Long-term cloud commitments are large but spread out
The filing detailed roughly $637 billion in spending commitments across infrastructure and supplier contracts, mostly extending into the early-to-mid 2030s. Minimum commitments include $114.5 billion with Google Cloud through 2033, $110 billion with AWS through 2036, and $45 billion with Azure through 2033. Additional commitments include $161 billion with Broadcom, $45 billion with Nscale, and $42 billion with Lambda.
Margins improved sharply
Compute costs fell from 72% of revenue to 48% quarter over quarter, a major sign of improving efficiency. Adjusted operating margin improved by 32 percentage points, moving from negative 13% to positive 19%. The company also approached $1 billion in GAAP profit, suggesting the business is scaling faster than many expected.
Enterprise adoption is broad and deepening
More than 10,000 customers spend over $100,000 annually, and over 1,000 spend more than $1 million. Net revenue retention for the $100,000-plus cohort was about 900%, indicating that many customers expanded usage nearly tenfold as AI tools spread beyond engineering teams into operations, marketing and other functions. More than half of the Fortune 500 spend at least $100,000 a year with Anthropic.
Some revenue comes from policy-violating use
A previously discussed 5% to 7% slice of revenue appears tied not specifically to model distillation, but more broadly to terms-of-service violations. That category could include unauthorized distillation, cyber abuse, adult content or other banned use cases. Even at the lower estimate, the figure points to a sizable gray-market demand for high-end model access.
Founder control is unusually strong
The filing creates a special Class F share held by a founder-controlled LLC that will command 50.1% of total voting power while carrying no economic rights. That gives the seven co-founders lasting control over the company even if their economic ownership falls sharply over time. The structure is more durable than standard dual-class shares and may become a model for future AI companies seeking insulation from public-market pressure.
Marketing spend has surged
The company spent just $2 million on marketing in 2023, but that rose to $568 million in 2025 and $3.17 billion in the first half of 2026 alone. The scale of that increase highlights how aggressively Anthropic is pursuing enterprise distribution. At the same time, the filing did not clearly break out spending on AI safety, despite catastrophic-risk language in the risk section.
One investor stands out
Spark Capital emerged as one of the biggest winners in the filing. After leading the Series C and continuing to invest through later rounds, the firm is estimated to have put in $777 million for a stake that could be worth about $18 billion at a $2 trillion valuation. Yasmin Razavi, the firm’s board representative, is the only venture capitalist on the board.
CONCLUSION
Anthropic’s IPO filing portrays an AI company with rare scale, improving economics and unusual governance power concentrated in its founders. The offering now stands as a test of how far public markets are willing to back the economics and structure of the generative AI boom.
Ask a question