Look, the main thing for you and me — the actual humans writing code — is this: Google’s top AI evangelist, Richard Seroter, basically told the world last week that developer loyalty to Google is currently sitting at absolute zero. Not ‘low,’ not ‘declining.’ Zero. And before you start thinking this is some kind of mea culpa, it’s not. It’s a declaration of war, repackaged as pragmatic business strategy.
Seroter’s not sweating it, though. His pitch? “Our job is that we should make it easy. If you want to build with AI or if you build AI apps, we should be excellent at both and be the best at both. We are not ceding anything. We will be the best at AI dev tools. We’re gonna get there — always.” Translation: Even if you’re using Cursor, or some other shiny new AI pair programmer they don’t offer, they’re betting you’ll still run your code on Google Cloud because, well, they’re going to make it the fastest, most efficient place to do so.
Why Does This Matter for Developers?
This isn’t just corporate navel-gazing. It’s a signal that Google’s shifting its focus from trying to be your favorite tool provider to being the indispensable platform. They’re saying, “We don’t care if you use Copilot or Claude Code. We do care that when that code finally needs to run, it’s doing so on our infrastructure, and doing it better than anywhere else.” It’s a classic platform play, amplified by the AI gold rush. If they can make running any AI model, built with any tool, superior on their cloud, then they win. The developers might still grumble about loyalty, but the bill will still land in their inbox.
Seroter even drops this gem, a clear shot across the bow to Azure and AWS:
“If you look at the latency, the best performing way to use Anthropic is on Vertex over Azure, AWS, and Anthropic itself. And so even if you’re using Claude Code, you should be using Vertex — otherwise you’re just getting worse performance.”
That’s the core of their strategy: performance. They’re not trying to win your heart with a charming UI or a beloved open-source library (though they dabble). They’re trying to win your wallet by promising you speed. And who doesn’t want speed when you’re churning out AI models at breakneck pace? It’s a very Silicon Valley way of thinking – brute force through infrastructure superiority. They’re essentially arguing that their cloud offers the equivalent of a sports car engine for your AI applications, regardless of whether you bought your car (the dev tool) from Ford or Ferrari.
The ‘One Platform’ Philosophy
Curiously, Seroter pushes back against the idea of Google Cloud being a sprawling portfolio of 300-plus services. “I don’t think Google sells 300 services. I think we sell one platform,” he states. This is a crucial distinction. It suggests a strategic tightening of the belt, a focus on areas where they must excel – especially AI – and a willingness to let other services be merely ‘good enough.’ It’s a necessary triage for a company that, frankly, has a notorious reputation for spreading itself too thin. The AI push isn’t just another initiative; it’s presented as the core of their ‘one platform’ strategy, a laser focus that’s supposedly been the biggest change in the last three years. Whether this singular focus is a genuine strategic pivot or just more polished PR spin remains to be seen, but the implication is clear: if it’s not directly serving their AI ambitions, it might be on the chopping block.
Seroter also addresses the whispers of Google moving too slowly, a criticism amplified by former Google engineer Steve Yegge. Yegge’s point? Google’s internal AI adoption is allegedly lagging, and engineers are restricted from using top-tier external tools like Claude Code because, well, it’s the ‘enemy.’ Seroter’s rebuttal is… predictable. Big companies, he argues, have to balance speed with stability. “We can’t break Maps,” he says, drawing a sharp contrast between building a production-critical platform and shipping a consumer chatbot. It’s a valid point, but it also conveniently sidesteps the core of Yegge’s critique: if your own engineers aren’t empowered to use the best available tools, how can you truly claim to be a leader in the field? It smacks of “do as I say, not as I do,
