AI Market Movers: The Real Shifts Behind the Hype

Anthropic just closed a sixty-five billion dollar funding round at a nine hundred sixty-five billion dollar valuation. Yesterday we talked about bio-hybrid brains playing Doom. Today, we're talking about a valuation that just eclipsed OpenAI's. The ground is shifting. Fast. Here’s what else moved today, June first, 2026.

SoftBank Group is dropping seventy-five billion euros to build five gigawatts of AI data center capacity in France. This is not a plan. This is a deployment order. Masayoshi Son says the countries that build the infrastructure will shape the future, and he’s putting his money where his mouth is.

Meanwhile, a Chinese lab called MiniMax just launched M3. It’s a model with a one-million-token context window that claims to beat GPT-5.5 on coding benchmarks. They’re promising to open-source the weights within ten days. That’s a direct challenge to the US-based developer stack. And while the giants throw billions around, the ground floor is getting crowded.

Google’s Gemini 3.5 Flash is now generally available, offering frontier-level intelligence for just ten cents per million tokens. And almost unnoticed, OpenAI's GPT-5.5 Instant quietly became the most-used AI model in history. The battle for the top and the battle for the bottom are happening at the same time.

Let’s go back to Anthropic. That nine hundred sixty-five billion dollar number isn't a typo. It makes them the most valuable AI company on the planet, pushing past OpenAI's eight hundred fifty-two billion dollar mark from March. And this isn't just about their new model, Claude Mythos, which was just cleared for broad release.

It’s about why that model was held back in the first place. The initial reports were not hype. Claude Mythos was originally gated because it demonstrated, and I'm quoting the internal memo, a "step change in autonomous cybersecurity capabilities." Specifically, it could find zero-day vulnerabilities in every major operating system and browser.

It was a digital skeleton key. So they spent months building safety guardrails before letting it out. That capability, even neutered, is what drove a two-point-five-X revenue growth in three months. Their run-rate revenue went from fourteen billion dollars in February to forty-seven billion in May.

Investors aren't buying a research lab. They're buying a money printer with a strategic weapon attached. But a model like Mythos doesn't run on air. Which brings us to SoftBank's seventy-five billion euro check to France. Five gigawatts. That’s not a number you hear outside of national energy grids.

This is the physical manifestation of the AI arms race. It’s a sovereign infrastructure play. Schneider Electric is building the data center enclosures. EDF, the state power company, is guaranteeing low-carbon electricity. This isn’t just a company building a server farm. This is a country deciding it will not be a digital colony of the United States.

It's an industrial strategy. So what changed today? The scale. We just went from talking about models in the abstract to talking about valuations that rival a small country's GDP. We went from benchmarks on a leaderboard to gigawatts on a national power grid. The prompt is no longer just how to ask a question.

It’s how to secure a trillion-dollar valuation and the energy supply to defend it. The frontier is no longer a research paper. It's a balance sheet. And today, that balance sheet is measured in trillions of dollars and terawatts of power.