Google Cloud Bets on AI Startups as Rivals Cement Billion-Dollar Infrastructure Deals

The AI infrastructure race is heating up — and Google Cloud is taking a very different path from its rivals.

While Nvidia and OpenAI made headlines this week with a $100 billion mega-deal involving massive chip purchases and enough computing power to power over 5 million U.S. homes, Google Cloud is quietly locking in the next generation of AI startups before they become industry giants.

Francis deSouza, Google Cloud’s COO since January, is leading this charge. A former CEO of genomics giant Illumina and co-founder of AI alignment startup Synth Labs, deSouza knows the AI landscape from multiple perspectives — from drug discovery to AI safety challenges.

Now at Google Cloud, he’s focused on one goal: making Google the go-to infrastructure partner for emerging AI companies.

Google Cloud’s Numbers Tell the Story

DeSouza highlights that nine of the top 10 AI labs use Google’s infrastructure. Nearly all generative AI unicorns run on Google Cloud. More than 60% of all AI startups worldwide have chosen Google as their cloud provider.

And there’s more: $58 billion in new revenue commitments for the next two years, more than double Google Cloud’s current annual run rate.

“AI is resetting the cloud market,” deSouza says. “Google Cloud is leading the way — especially with startups.”

Rivals Are Going Bigger — and Spending Billions

The Nvidia-OpenAI deal isn’t the only massive investment shaping AI infrastructure:

Microsoft has pumped nearly $14 billion into OpenAI.
Amazon invested $8 billion in Anthropic, customizing AI hardware for its cloud.
Oracle secured a $30 billion OpenAI cloud deal, plus a staggering $300 billion commitment starting in 2027.
Meta signed a $10 billion Google Cloud deal, despite building much of its own infrastructure.

Even the U.S. government is in the mix, with the Trump administration’s $500 billion “Stargate” project involving OpenAI, Oracle, and SoftBank.

By contrast, Google Cloud’s approach is different: Instead of chasing multi-hundred-billion-dollar deals, it’s signing up AI startups early, offering cloud credits, technical support, and a “no-compromise” AI stack — from chips to models to applications.

Why This Strategy Might Pay Off

Google gives AI startups $350,000 in cloud credits, access to enterprise-grade infrastructure, and a marketplace to bring their products to market.

This strategy could pay off big if today’s small AI companies become tomorrow’s billion-dollar giants. As deSouza puts it:

“Companies can go from startup to multi-billion-dollar company very quickly. Capturing them early could be worth more than fighting over today’s giants.”

Open, Yet Competitive

Google Cloud plays a complex game:

It provides TPU chips to OpenAI.
Hosts Anthropic’s Claude on Vertex AI.
Competes directly with both through its Gemini models.

And yet, Google maintains what deSouza calls an “open ethos”, offering technology even to potential rivals.

“We’ve always been open at every layer,” deSouza says. “Companies can take our tech and build competitors on top. That’s fine with us.”

The Big Picture

Regulators are watching closely. A recent federal ruling in Google’s long-running search monopoly case raised concerns about whether its search dominance could unfairly extend to AI.

But Google Cloud’s open platform approach — empowering startups instead of locking them in — might help counter accusations of monopolistic behavior while securing relationships with the next wave of AI innovators.

As deSouza sees it, the goal goes beyond market share:

“We want to pioneer the technologies that help solve diseases like Alzheimer’s and Parkinson’s and drive breakthroughs in climate science. That’s the real opportunity.”