OpenAI burning $14 billion a year is racing to go public before 'Anthropic' goes first

Two companies, the same technology, opposite bank balances. The gap between them is a lesson in who actually pays for AI.

Sam Altman wants OpenAI to list at a valuation above $1 trillion, and has reportedly told advisers that going lower is a "non-starter." OpenAI sits at a private valuation near $852 billion after a March 2026 round that raised $122 billion, has confidentially filed IPO paperwork, and is working with Goldman Sachs, Morgan Stanley and JPMorgan on a listing that could come as early as this fall. The company's revenue is real and growing fast, roughly $2 billion a month, about $40 billion annualized in 2026, up from $20 billion at the end of 2025. Its projected loss for 2026 is also real: around $14 billion.

Anthropic, meanwhile, just posted its first positive adjusted operating income. Its annualized revenue hit $65 billion by the end of July 2026, up from $47 billion in May and just $9 billion at the end of 2025, a sevenfold jump in seven months. Anthropic is reportedly targeting an IPO as soon as this fall, ahead of OpenAI, at a valuation of $2 trillion or more, which would make it larger than OpenAI's own target despite Anthropic having a fraction of OpenAI's revenue and brand recognition just twelve months ago.

The insight most coverage treats as a personality difference is actually a business model difference.

OpenAI built its reach on ChatGPT, a free consumer product used by hundreds of millions of people. Individual subscriptions make up roughly 40% of OpenAI's revenue, and the free tier, while it drives OpenAI's massive user numbers, costs money to run for every user who never pays. Anthropic took the opposite path. The large majority of its revenue, by some estimates 75 to 85%, comes from usage-based API access sold to businesses and developers who build products on top of Claude, paying for exactly what they use, at margins that scale far better than a free consumer app ever can. Anthropic's own coding tool alone reportedly touches a meaningful share of software commits happening on GitHub every day. One company is selling a habit to the public. The other is selling infrastructure to businesses that have no free alternative once they've built on top of it.

A useful parallel.

Picture two coffee shops on the same street. One gives away a small coffee to anyone who walks in, hoping enough of them eventually pay for a larger size, and spends heavily on rent, staff and beans to keep the free line moving. The other only serves offices with standing weekly orders, charging every cup at full price from day one. The first shop will always look busier. The second shop's books will always look better.

Both are legitimate businesses. They are not the same business.

There is also a clock neither company fully controls. SoftBank, which helped fund OpenAI's latest valuation, is carrying a $40 billion bridge loan that matures in March 2027, creating real pressure for OpenAI to list sooner rather than wait indefinitely for its trillion-dollar number. Microsoft owns roughly 27% of OpenAI and has committed nearly $10 billion more on top of the $13 billion already invested. Nvidia has put roughly $30 billion into OpenAI directly and more than $40 billion across the sector including Anthropic, and OpenAI's own compute roadmap calls for more than 10 gigawatts of Nvidia systems. Every one of those partners has a stake in which valuation actually prices, and when.

Sam Altman has reportedly told advisers that any reduction in OpenAI's trillion-dollar IPO valuation target is a "non-starter," even as SoftBank's $40 billion bridge loan comes due in March 2027.

The honest catch.

Anthropic's own $2 trillion target reportedly leans heavily on projected 2028 revenue rather than what it has actually earned this year, which is a common but aggressive way to justify a valuation that size, and recent mega-IPOs elsewhere have traded well below their debut price once the hype settled. OpenAI's $14 billion loss, meanwhile, doesn't necessarily mean a broken business, heavy losses while scaling infrastructure are normal for capital-intensive tech companies early on, and OpenAI's revenue growth is itself genuinely strong. Both companies could be right about their long-term value and wrong about the exact number the market assigns them on day one.

The race to a trillion dollars was supposed to be about whose AI wins. It is turning out to be about whose customers pay the bill.