Tech • AI • Robotics • Game
Anthropic: Tech’s Biggest Cash Machine?
9.4/10TL;DR
Anthropic’s IPO filing depicts a company with explosive growth, improving operating profits and vast cash reserves, but its central risk is not rogue AI: it is securing enough computing power to sustain a business that may require hundreds of billions of dollars in future infrastructure commitments.
KEY POINTS
Record scale ahead of listing
Anthropic plans to list on the Nasdaq under ANT, with Goldman Sachs and Morgan Stanley leading the offering. The deal is expected to seek at least $100 billion, with a valuation discussed in a range of $1.5 trillion to $2 trillion, which would make it the largest IPO ever. The filing shows a company already operating at a scale rarely seen in software.
Revenue growth is extraordinary
The company reported $16.3 billion in revenue in the first half of the year, up roughly 14-fold from a year earlier. Second-quarter revenue alone reached $11.6 billion, exceeding the company’s full-year 2025 revenue. Updated third-quarter figures reinforced the picture of hypergrowth, placing Anthropic in a category more comparable to infrastructure giants than to a normal software startup.
Operating profit exists, but with caveats
The filing shows first-half operating profit of about $974 million. Third-quarter operating margin was presented near 15%, but that figure excludes stock-based compensation, estimated around $3.2 billion. That matters because equity awards are a core part of compensation and talent retention in AI, making headline profitability stronger than a fully loaded view of earnings.
Cash reserves are huge
Anthropic has raised about $140 billion and still holds roughly $130 billion in cash and cash equivalents. That suggests it has not burned capital as quickly as many expected, despite the massive cost of training and serving frontier AI models. The balance sheet gives the company unusual flexibility going into public markets.
The real exposure is off-balance-sheet compute
The major risk lies in long-term commitments for chips, data centers and electricity. These obligations are not recognized as expenses until the compute becomes available, but published commitments imply a total future exposure of roughly $850 billion to $1 trillion, with some payments stretching to 2045. Much of that appears to be binding, making liquidity and execution more important than current operating profit.
Compute, memory and power outrank AI safety in stated risks
The filing’s top risks are the ability to stay at the frontier of model performance, secure enough compute and avoid supply bottlenecks in chips, memory and energy. Concerns about misuse or loss of control appear lower down the list. That ordering underlines a blunt reality in generative AI: market leadership may depend less on abstract safety fears than on access to scarce hardware and electrical capacity.
Customer traction is unusually deep
Anthropic reported about 11,900 customers spending more than $100,000 annually, up from 950 a year earlier. It also has around 250 customers spending more than $10 million a year and about 40 million smaller users. That enterprise footprint is exceptionally large for a company of its age and suggests adoption has spread through developers into major organizations without a traditional, heavy sales buildout.
Margins are strong, but durability is still debated
Bulls argue the company’s inference economics imply gross margins near 80% to 85% before research costs, though that may overstate profitability if free users are classified as sales and marketing expense. Another uncertainty is pricing: token costs are falling across AI, but frontier models may preserve premium pricing longer than weaker systems. The long-term test is whether Anthropic can evolve from selling API access into becoming a durable enterprise platform.
The broader market impact could be substantial
A successful IPO could mint thousands of multimillionaires, drive a fresh surge in San Francisco Bay Area real estate and seed a new wave of startups founded by former employees. It also highlights a larger macro shift: big tech groups are borrowing heavily to finance AI infrastructure, competing more directly with government debt markets for investor capital. As a result, the AI boom is beginning to reshape not only software economics but also capital markets and industrial planning.
CONCLUSION
Anthropic is emerging as either one of the greatest businesses ever built or one of the largest infrastructure bets in tech history. Its future now hinges less on demand than on whether it can keep buying the compute required to stay ahead.
Ask a question