AI Market Shifts: Anthropic Overtakes OpenAI & China’s Moonshot Debuts Kimi

Anthropic just raised sixty-five billion dollars in a single funding round. In our last episode, we talked about cutting through the hype. Today, that hype is being priced at a nine-hundred-and-sixty-five-billion-dollar valuation, making Anthropic the most valuable AI company on the planet. This isn't just another big number. It's a structural shift. OpenAI is no longer the undisputed king. Here’s what else is moving. China's Moonshot AI just dropped its new frontier model, Kimi K2.7. They're now chasing a thirty-billion-dollar valuation—a sevenfold increase since December. The performance gap between U.S. and Chinese models is now just two-point-seven percent.

Google is finally making its move. Gemini 3.5 Flash is now the default engine for Google Search's AI Mode. This is the biggest overhaul to Search in twenty-five years, and it changes everything for businesses that rely on search traffic. Microsoft, not to be outdone, unveiled a suite of seven new in-house models under the "MAI" designation. The flagship is MAI-Thinking-1, and it's built for one thing: delivering high-end logical reasoning without the high-end cost. And the latest Stanford AI Index just put a number on the cost of all this progress. Training a single model, Grok 4, emitted over seventy-two thousand tons of CO2. That's the equivalent of seventeen thousand cars on the road for a year.

The same report confirms entry-level software developer jobs are down nearly twenty percent, automated away by boilerplate code generation. Meanwhile, Yann LeCun's new startup, AMI Labs, just pulled in over a billion dollars for its seed round in Europe. They're betting against the large language model approach, focusing instead on "world models" that learn from physical dynamics for robotics and healthcare. Finally, a deadline is looming. The EU AI Act goes into full enforcement on August second. For any company operating in or with Europe, AI governance just got very, very real. Okay, let's go back to Anthropic. Forget the sixty-five billion for a second.

The real story is how they're spending it. Kersai Research confirms they are paying SpaceX one-point-two-five billion dollars a month. Per month. Why? For over three hundred megawatts of compute capacity. This isn't about building a better chatbot. This is about building the power grid for an AI-native economy. While everyone else was focused on model leaderboards, Anthropic was securing the energy supply. They're not just training models; they are building a utility. This scale of infrastructure investment has no precedent. It reframes the competition entirely. It's no longer about who has the smartest model today, but who has the power to run a million autonomous agent workflows tomorrow.

Then there's Google. Their move is quieter, but maybe even more profound. Making Gemini 3.5 Flash the engine of Search changes the internet's center of gravity. For two decades, your relationship with Google was a list of blue links. That's over. Now, you're getting custom-built pages, generated by AI, on the fly. You can use text, images, even video in your queries. And behind the scenes, Google is deploying what it calls "information agents" to do the work for you. The strategy is clear, and it’s brilliant. They aren't trying to sell you a new AI stack. They are making the software you already use more intelligent. This is the quiet integration that wins markets.

It's not a flashy demo. It's a fundamental change to a daily habit for billions of people. So you have two tectonic plates shifting at once. On one side, you have the brute-force scaling of Anthropic, building centralized AI power on an almost national scale. On the other, you have Google's ambient, pervasive intelligence, embedding itself into the fabric of the web. These aren't competing visions. They are two different answers to the same question: what happens when AI stops being a feature and starts being the foundation? One builds the power plant. The other rewires the city. The era of AI demos is officially over. The era of AI infrastructure has just begun.