Google just reshuffled the team that was supposed to win the AI race, and the timing is telling. On the latest episode of Decoder, The Verge AI’s senior AI reporter Hayden Field broke down what last week’s shakeup at Google DeepMind actually means. Chief scientist Jeff Dean, who founded Google Brain, is leaving to start his own lab that will run on Google Cloud. DeepMind cofounder Demis Hassabis is stepping back from day-to-day leadership to focus on longer-term research. Two of the most important names in AI, moving out of the driver’s seat at the same time.
What stands out here is the contradiction. Google should be the runaway leader, and it isn’t.
The company that has everything
Look at the ingredients. Google owns Search, a product billions of people touch daily. It has the data, the distribution, the cloud infrastructure, and the research pedigree. As The Verge AI notes, during the Elon Musk-Sam Altman trial, rivals spent years genuinely afraid of Hassabis. On paper, this race was Google’s to lose.
Field’s read is sharp: Google is in real trouble on the frontier, but writing them off is a mistake. “They have a huge safety net,” she said, comparing the company to someone with a trust fund. That trust fund is Search. Even if Google keeps stumbling, it has a cushion almost no competitor can match.
Still, the admissions from leadership are striking. Sundar Pichai says the company is “committed to being at the frontier” and focused on areas it needs to improve. Field translates that as CEO-speak for “we know we need to get our act together.” Hassabis wrote that Google is “entering a next chapter” with “the ingredients to lead from here.” Read closely, and both men are admitting the same thing: Google isn’t leading right now.
The bet that may have missed
Here’s the deeper question The Verge AI raises. Did Google bet on the wrong thing?
Hassabis has been focused on multimodality and world models, the kind of long-horizon research that could pay off enormously someday. Meanwhile, the money in AI today is in enterprise. Anthropic has jumped out to a lead by focusing on business use cases, especially coding. Consumer AI, the arena where Google’s Search dominance matters most, hasn’t turned into big profits for anyone yet.
So Google may have optimized for the prize everyone assumes is coming while competitors banked the revenue that’s already here.
Field floats a provocative angle: maybe Google doesn’t fully want to be at the frontier. There’s a real business in selling cloud services to companies like Anthropic, and that business is growing. Winning the model race and renting shovels to the winners are two very different strategies.
Why this matters now
This is an org chart story, and org charts reveal priorities. When a company moves its visionary founders into research roles and lets a Brain founder walk out the door, it’s making a statement about where the real work is happening.
A few takeaways for practitioners and businesses watching this:
- Don’t confuse resources with results. Google proves that distribution, data, and talent don’t automatically translate into a frontier lead. Execution and focus decide the race.
- Enterprise is where the revenue lives. Anthropic’s coding-first approach is generating real money now. If you’re betting on an AI vendor, follow the paying customers, not the demos.
- Watch the cloud angle. Google can lose the model race and still profit by hosting competitors. That makes its cloud business, not just Gemini, worth tracking.
- Never fully count Google out. The Search cushion buys time that no startup has.
The honest answer to whether Google is losing is: yes, for now, on the frontier. But Field’s framing is the one to hold onto. A company with this much power that still finds itself behind is either about to make a serious correction or quietly deciding that a different game is worth playing.
The next few quarters will show which. For the full conversation, the Decoder interview at The Verge AI is worth your time.