For the last two years, OpenAI's strategy was essentially: build everything. You have a video generator, a browser, hardware devices, e-commerce tools, shopping features, and whatnot inside ChatGPT.
Sam Altman called it "a series of bets on internal startups." The logic made sense at the time. OpenAI was so far ahead of everyone else that it could afford to explore in every direction at once.
And then the numbers came in.
Anthropic, a much smaller company with far fewer products, has captured 73% of all new enterprise AI spending. 73% of companies buying AI tools for the first time are choosing Anthropic over other options.
One year ago, one in twenty-five businesses paid for Anthropic. Today, it's one in four. But it recently recorded its biggest single-month drop since anyone started tracking it.
Why did this happen?
It's not because Claude is dramatically smarter than GPT-5.4. On many benchmarks, they're comparable. It's because Anthropic chose focus. They have two products: Claude Code for developers, and Cowork for enterprise. That's it.
Developers started using Claude Code every day. The person on every team who evaluates AI tools started recommending Claude. And enterprises followed.
Meanwhile, OpenAI's engineers didn't always know what they were working on. The Sora video team was reporting to the research division, even though Sora was a consumer product. Computing resources were moved between teams at the last minute. Internally, people described it as a lack of clarity.
OpenAI's head of applications, Fidji Simo, walked into an all-hands and said something that rarely comes out of a company that still leads the market: "We are very much acting as if it's a code red."
She called Anthropic's rise a "wake-up call." She told staff: "We cannot miss this moment because we are distracted by side quests."
And so, for the first time in its history, OpenAI is cutting everything that isn't core. Hardware is being deprioritized, browser development is slowing, and the focus is now exclusively on coding and enterprise. Consultancies like BCG and McKinsey are being brought in specifically to push OpenAI's tools into large companies.
Here's the thing worth sitting with:
OpenAI is still bigger than Anthropic, still has more revenue and users. It still has the most recognized brand in AI. And yet the company declared an internal emergency. That's how fast the enterprise market can move. And that's how consequential the decision to focus really is.
For every founder reading this: the company that's beating the world's most powerful AI lab isn't doing more. It's doing less and doing it better.
A mystery trillion-parameter AI appeared out of nowhere. Nobody knows who made it.
On March 11, a model called "Hunter Alpha" silently appeared on OpenRouter with no company name, no press release, nothing. It described itself as "a Chinese AI model" with 1 trillion parameters, a 1 million token context window, and it's completely free.
Within a week: 160 billion tokens processed. Every spec matches what's been leaked about DeepSeek V4, expected in April.
ChatGPT is getting more expensive. OpenAI said so out loud.
OpenAI's CFO publicly admitted unlimited AI at current prices "just doesn't make sense." The company is actively losing money subsidizing every ChatGPT user. Price hikes are coming.
Meanwhile Claude just doubled limits for free. The timing for OpenAI couldn't be worse.
Nvidia is restarting chip sales to China.
After US export controls blocked Nvidia from selling advanced chips to China, they built a downgraded version that stays within the legal limits. Jensen Huang just announced they're restarting production of that chip for the Chinese market.
China is still the world's biggest AI hardware buyer. Nvidia isn't walking away from that regardless of what Washington thinks.
FAST BREAK
Anthropic now wins 73% of new enterprise deals against OpenAI. A year ago, it was 10%.
That's not model quality or pricing. That's what happens when one company chooses focus, and the other chooses everything.
Ramp tracks real credit card spending across 50,000+ businesses. Their economist put it plainly: choosing Claude has become less like a procurement decision and more like a statement of identity. It’s the way some people choose a blue bubble over a green one.
In every market, focus beats scale. AI is no different.