Three things happened this week that, taken together, mark a turning point for the AI industry. OpenAI is filing for its IPO. Anthropic is projecting its first profitable quarter. And the White House pulled an executive order that would have required AI companies to share models with the government before launch. The era of AI as a money-losing moonshot backed by patient venture capital is over. What comes next looks a lot more like Big Tech.
OpenAI Goes Public
OpenAI plans to confidentially file its IPO prospectus with the SEC as early as Friday, with Goldman Sachs and Morgan Stanley running the deal. CEO Sam Altman is targeting a September debut.
The numbers behind the filing are staggering. OpenAI says it generates $2 billion in revenue per month, with an annualized run rate above $30 billion. Its last private valuation hit $852 billion after closing a record $122 billion funding round in March. A successful IPO could push the company past $1 trillion.
The timing is not accidental. On Monday, a federal jury threw out Elon Musk’s lawsuit against OpenAI, removing the most visible legal cloud hanging over the company. Musk plans to appeal, but the jury took less than two hours to reject every one of his claims. For OpenAI’s IPO lawyers, that was the green light.
The company still is not profitable, which means the S-1 will need to tell a convincing growth story. At $30 billion annualized revenue with losses still mounting, OpenAI is essentially asking public markets to fund the next phase of the AI arms race.
Anthropic Turns Profitable
While OpenAI races to go public, its closest rival just hit a different milestone. Anthropic told investors it expects $10.9 billion in Q2 revenue — up 130% from $4.8 billion in Q1 — and projects a $559 million operating profit. That would be its first profitable quarter ever.
The driver is Claude Code, Anthropic’s coding assistant, which has seen explosive enterprise adoption. In Q1, Anthropic spent 71 cents on compute for every dollar of revenue. That ratio is expected to fall to 56 cents this quarter, meaning the company is scaling faster than its infrastructure costs.
But Anthropic itself warns this might not last. The company cautioned investors that it may not sustain profitability for the full year, because it plans to ramp compute spending significantly in the second half of 2026. Which brings us to the most revealing deal of the week.
The $45 Billion Compute Deal Nobody Expected
Buried in SpaceX’s IPO filing is a contract that says more about the state of AI infrastructure than any earnings report: Anthropic will pay xAI $1.25 billion per month through May 2029 for access to the Colossus 1 data center near Memphis. The total contract value: roughly $45 billion.
Read that again. Anthropic — one of xAI’s direct competitors — is renting Elon Musk’s GPUs because it cannot get enough compute anywhere else. The deal gives Anthropic 300 megawatts of capacity and more than 220,000 Nvidia GPUs.
Why would xAI sell capacity to a rival? Because Grok usage has dropped significantly in recent months, leaving Colossus 1 underutilized. Musk built the world’s largest AI data center and his own product cannot fill it. Anthropic’s money keeps the lights on while SpaceX heads toward its own IPO.
The contract includes a 90-day exit clause for either side, which means this is a marriage of convenience, not commitment. But $15 billion a year in compute spending from a single customer tells you exactly how capital-intensive frontier AI has become — and why these companies need public market money.
Washington Takes a Pass
Against this backdrop, the White House on Wednesday pulled an executive order that would have created a pre-launch review process for frontier AI models. The order would have required companies to share advanced models with the government 14 to 90 days before public release, with the NSA involved in classified security testing.
President Trump said he delayed it “because I didn’t like certain aspects of it” and added: “We’re leading China, we’re leading everybody, and I don’t want to do anything that’s going to get in the way of that leading.”
The invitations had already gone out for the signing ceremony. Reports suggest the real reasons were a combination of Trump’s instinctive resistance to regulation and the fact that not enough tech CEOs could fly to Washington on short notice for the photo op.
Whatever the reason, the signal is clear: the two largest AI companies are heading to public markets with no new federal oversight framework in place. Europe has the AI Act. The U.S. has a postponed executive order and a president who does not want to slow things down.
What This Means
The financial facts are now unavoidable. OpenAI and Anthropic are no longer scrappy research labs — they are infrastructure companies with combined annualized revenue approaching $60 billion, burning through compute at rates that make cloud computing look cheap, and preparing to sell shares to your retirement fund.
The Anthropic-xAI compute deal reveals the absurdity of the current moment. The demand for AI training and inference compute is so extreme that direct competitors are renting each other’s data centers. GPU access has become the oil of the 2020s, and there are not enough refineries.
For anyone watching the AI industry’s trajectory, the next six months will be decisive. OpenAI’s S-1 will reveal, for the first time, exactly how much money it takes to stay at the frontier. Anthropic’s profitability will be tested as it ramps spending. And the regulatory vacuum means both companies will enter public markets with minimal government guardrails on what they build next.
The question is no longer whether AI companies can make money. It is whether the money they make can keep pace with what they need to spend.