How Anthropic Is Quietly Winning the AI Race

In March 2026, Fidji Simo told her team at OpenAI that they couldn't afford "side quests" anymore. As CEO of Applications, she was overseeing the shutdown of Sora (which burned an estimated $15 million per day in inference costs while earning just $2.1 million in total lifetime revenue), the consolidation of a scattered product portfolio, and a hard pivot to enterprise. Her words to the team were blunt: "We cannot miss this moment because we are distracted by side projects."

As of March 2026, Anthropic's ARR (annualized run rate, the most recent month's revenue multiplied by twelve) has surged to $19 billion, up from $1 billion just fifteen months earlier. ChatGPT's consumer market share has dropped from 86.7% to 60.4% in a little over a year, according to First Page Sage's blended estimate, with all major trackers confirming the directional decline.

Anthropic did not win by having the most users, the most compute, or the loudest marketing. They won by being the most focused, the most disciplined, and the most strategic about where to compete. While OpenAI chased video generation, browser plugins, and hardware, and Google tried to integrate AI into everything at once, Anthropic placed a single bet on coding AI and enterprise. That bet created a flywheel that is now spinning faster than anyone can match.