Elena Voss
Nine each. On April 17, 2026, OpenAI and Anthropic each appeared in nine articles — the first clean tie. Both z-scores: +16.0.
One of OpenAI's nine, reported the same afternoon, was this: a four-month-old startup founded by ex-OpenAI and ex-DeepMind engineers had just closed a $500M round at a $4B valuation. The backers were Google Ventures and Nvidia — OpenAI's largest competitor for frontier-model attention, and the chip supplier every frontier lab depends on. Another of the nine, filed hours earlier: Bill Peebles, the researcher who built Sora, and Srinivas Narayanan, CTO of enterprise applications, both departing OpenAI.
The two stories don't need to share a destination to share a price signal. OpenAI's most important rival and its most critical supplier just put $4 billion on the idea of building OpenAI's successor — four months in, before the startup shipped anything. The nine-to-nine coverage tie is the surface. That valuation is the substance.
The Ratio
For most of the last two years, OpenAI received three to six times more coverage than Anthropic. In 2024, the ratio was 6.5. In January 2026, it was 1.8. For Q1 2026, it was 1.22 — OpenAI with 431 articles to Anthropic's 353. For Q2 2026 to date, it is 0.92. Anthropic: 103 articles. OpenAI: 95.
Not the first signal of convergence. The first flip.
The Two Stacks
Anthropic's nine: Claude Opus 4.7, a flagship model release the company called "a notable improvement." Claude Design, a new product turning text prompts into visual layouts. An 800-seat London office, signed days after OpenAI began leasing its first London address — a 4x expansion of headcount in a single lease. Dario Amodei walking into the West Wing to meet Chief of Staff Susie Wiles, ending a Pentagon blacklist dispute. Treasury Secretary Scott Bessent joined the meeting in person. An FT interview with Amodei reframing the industry narrative — a posture available to companies on the ascending curve.
OpenAI's nine, beyond the three departures and the Recursive Superintelligence disclosure: Kevin Weil, former CPO and VP of OpenAI for Science, also departing; Prism, the product he launched in January, shuttered into Codex. Shareholders questioning whether Sam Altman should lead the company through its IPO, floating Bret Taylor as successor. Ronan Farrow on his New Yorker Altman profile, 18 months of reporting, subtitled by interviewers as Altman "changing." Ad buyers reporting ChatGPT ad CPMs falling from $60 to $25 and minimum spend from $250K to $50K — a 58% price cut five months after the product launched.
OpenAI's largest competitor and its most critical supplier just jointly priced its ex-employees at $4 billion.
The Reversal
For most of OpenAI's history, the flows ran inward. Microsoft's $135 billion produced a dependency, not a competing Microsoft model. Nvidia's flagship chip allocations went to Stargate — OpenAI's data-center buildout — before they went anywhere else. When senior researchers left, they started companies that competed with OpenAI on ideology (Anthropic) or approach (Thinking Machines), not companies funded by OpenAI's own capital and supply chain.
April 17 is the first day the flow visibly went around. Google Ventures and Nvidia didn't co-invest in OpenAI's next round. They co-invested in its alumni at a $4B price — the cost of owning a piece of the next OpenAI without having to fund OpenAI. The same two weeks, three Stargate leaders left for Meta. COO Brad Lightcap moved to special projects. Fidji Simo went on medical leave. Peebles, Narayanan, Weil. The pattern is not individual attrition. It is capital and talent routing around the center instead of through it.
The Counter-Flow
Anthropic's day is the other side of the same transaction. Its new hires are not coming from its own alumni network — its pipeline is being fed from OpenAI's. Its government access is expanding while OpenAI's Stargate permit regime loses the people who negotiated it. Its 2026 revenue forecast stands at $18B; its London footprint tripled in a single signing; its CEO is the one explaining to the press what the industry narrative has gotten wrong — a posture that requires being the story, not explaining it.
The obvious counter is that OpenAI is still larger by every financial measure that matters. $2B per month in revenue, 800 million weekly ChatGPT users, an IPO on deck. True. But the IPO prices yesterday's position. The $4B valuation on Recursive Superintelligence prices tomorrow's. And Google Ventures and Nvidia just committed to the second number, four months into a company with no product.
Nine, Recontextualized
The opening number is small. Nine articles apiece is not a volume story. What the tie describes is a structural one: on a single day, three reporters independently chose to write about a company's flagship model, its geographic expansion, and its White House access — and three other reporters independently chose to write about another company's departing executives, its shareholder revolt, and the valuation its competitor just put on its alumni.
The ratio was 6.5 when OpenAI was defining the industry and Anthropic was the footnote. It was 1.8 when Anthropic was the co-author. It is 0.92 on the day Google Ventures and Nvidia priced OpenAI's ex-engineers at $4 billion.
Parity is not where the two companies met. It is where the arrangement that built OpenAI — a competitor's money, a supplier's chips, and its own talent all routed through one address — started routing somewhere else.