OpenAI is shutting down consumer experiments, killing products, and realigning its entire company around business customers. The catalyst is not a visionary pivot. It is Anthropic eating its lunch.

CFO Sarah Friar told the Associated Press on Friday that OpenAI will release a new AI model designed for "high-value professional work" in the coming weeks. Internally codenamed Spud, the company describes it as its "smartest model yet," offering "stronger reasoning, better understanding of intent and dependencies, better follow-through and more reliable output in production."

Spud is OpenAI's answer to Anthropic's Claude Mythos, the model so capable that Anthropic limited its access to select customers after it demonstrated the ability to surpass human cybersecurity experts. On the same day Anthropic dropped Opus 4.7 as its most powerful generally available model, OpenAI rushed out GPT-Rosalind, a specialized life sciences model named after Rosalind Franklin. The timing was not subtle.

The Numbers Tell the Story

OpenAI has 900 million weekly ChatGPT users. About 95% of them pay nothing. That is an extraordinary burn rate for a company valued at $852 billion that still loses more money than it makes.

When Friar joined as CFO in 2024, business customers accounted for roughly 20% of revenue. That share has doubled to 40%. She expects it to reach 50% by year end. The message: free ChatGPT users built the brand, but paying enterprises will keep the lights on.

The shift has casualties. Sora, the AI video generator that Sam Altman once paraded alongside a Disney partnership, is dead. "I think it was a little heartbreaking," Friar said. "But we need to make sure that our new model that is coming has enough compute." Translation: Sora was eating GPU cycles that Spud needs more.

Anthropic Made Them Do It

Enterprise AI has been Anthropic's bread and butter from the start. While OpenAI chased consumers with ChatGPT ads, erotica features, and a Disney deal, Anthropic quietly signed up corporate customers. Ramp data from Q1 2026 shows Anthropic now accounts for 37% of trackable enterprise AI spending, versus OpenAI's 33%. First time Anthropic has won on dollars.

The response was structural. Three months ago OpenAI hired Slack CEO Denise Dresser as its first chief revenue officer. She has been laser-focused on corporate meetings and positioning OpenAI as the go-to workplace AI platform. The consumer era at OpenAI is over.

The Real Question

Friar framed the shift as healthy discipline: "Great companies are very good at doing that winnowing down and refocusing." Altman said on the Mostly Human podcast that a "sharper focus" was needed.

But here is the uncomfortable truth. Late last year, Altman was promoting Sora with Disney, launching ChatGPT ads, and floating AI erotica for paid users. Now all three are dead or dying. That is not a pivot. That is a company that tried everything, watched Anthropic win the market that matters, and is now scrambling to catch up.

Both companies are privately held, both lose more money than they make, and both are racing toward IPOs. The difference: Anthropic never had to abandon its strategy to find one. OpenAI is on its third in six months.

First reported by the Associated Press.