Tech • AI • Robotics • Game
Anthropic Files for IPO, Samsung Spyware Concerns | The TBPN Diet
9.4/10TL;DR
Anthropic has filed for an IPO that could value the AI company at $1.5 trillion to $2 trillion, with investors focusing on its rapid revenue growth, improving margins, large compute commitments and unusual founder-control structure.
KEY POINTS
Scale of the offering
Anthropic filed for what could become one of the largest IPOs ever, though the exact fundraising target has not been disclosed. Market talk has centered on roughly $100 billion of new capital, with valuation estimates ranging from $1.5 trillion to $2 trillion. The company is expected to list under the ticker ANT, with timing discussed for the week of November 16.
Revenue growth and cash position
The filing shows annualized revenue of $67.2 billion, up 45% quarter over quarter, underscoring how quickly enterprise AI demand is expanding. Anthropic also reported $131.7 billion in cash on hand, giving it one of the strongest balance sheets in the sector. As of June 30, 2026, it had raised $141 billion in total capital and had burned only about $9.8 billion, a level some investors described as exceptionally capital efficient for a company operating at this scale.
Improving economics
Compute costs fell sharply as a share of revenue, declining from 72% to 48%. Adjusted operating margin improved by 32 percentage points quarter over quarter, moving from negative 13% to 19%, and the company was close to $1 billion in GAAP profit. Those figures helped calm concerns that the business might still be structurally unprofitable despite its size.
Enterprise adoption
Customer metrics point to broad corporate uptake. More than 10,000 customers spend over $100,000 annually, and over 1,000 spend more than $1 million. Net revenue retention on the $100,000-plus cohort was reported at about 900%, indicating many customers expanded usage by roughly 10 times as AI tools spread beyond technical teams into operations, marketing and other functions.
Fortune 500 penetration
More than half of the Fortune 500 spend over $100,000 a year with Anthropic. That is a strong foothold, though it also suggests large enterprises are still splitting AI budgets across several providers, including OpenAI, Google and internal or open-source systems. The data implies corporate AI adoption is broad but not yet consolidated around a single platform.
Renting compute rather than owning it
Despite its size, Anthropic spent only $482 million on capex in the first half, signaling a strategy centered on renting infrastructure rather than building most of it directly. The company has minimum cloud commitments of $114.5 billion with Google Cloud through 2033, $110 billion with AWS through 2036, and $45 billion with Azure through 2033. Additional agreements include $161 billion with Broadcom, $45 billion with Nscale, $42 billion with Lambda, and a $120 billion maximum arrangement with SpaceX AI, for total commitments of about $637 billion spread over many years.
Question marks around questionable usage
A previously discussed 5% to 7% slice of revenue tied to terms-of-service violations appears smaller than some critics had suggested. The filing does not explicitly label that revenue as model distillation, and it may also include other prohibited uses such as cybercrime or adult content. Even so, the figure indicates that abusive or gray-market activity could still amount to billions of dollars at current scale.
Seasonality and marketplace dependence
The filing indicates that usage drops during the year-end holiday period, with API consumption and revenue in the final two weeks of December expected to be down about 20%. It also shows that 48% of revenue comes through cloud marketplaces, a sign of how dependent the business remains on hyperscaler distribution channels even as its direct brand grows.
Marketing spend and security scrutiny
Marketing costs have climbed rapidly, from $2 million in 2023 to $568 million in 2025, and $3.17 billion in the first half of 2026 alone. The company also faces scrutiny over cybersecurity after researchers claimed they were able to access internal systems, including Slack channels tied to sensitive model discussions. While the incident did not appear to derail investor enthusiasm, it reinforced concerns about the risks attached to frontier AI labs.
Founder control structure
The filing outlines a highly unusual governance model. A single Class F share will be issued to a founder-controlled LLC whose voting power equals 50.1% of the company’s total votes on key matters, despite carrying no economic rights. That structure allows the seven co-founders to retain effective control indefinitely, even if their economic ownership falls sharply over time.
CONCLUSION
Anthropic’s IPO filing presents a company with enormous scale, fast-improving economics and deep enterprise demand, while also exposing the governance, infrastructure and security risks that come with the AI race. The central question for investors is whether those risks are outweighed by a business that already looks like one of the fastest-growing large companies in modern market history.
Ask a question