The $100 billion Nvidia–OpenAI deal may dominate headlines, but Google Cloud is taking a different route. Instead of chasing massive, high-profile partnerships, it’s focusing on the next wave of AI startups — the companies that could become tomorrow’s unicorns.
Francis deSouza, Google Cloud’s new COO, believes this is where the real opportunity lies. He points out that nine of the top 10 AI labs already use Google’s infrastructure, and 60% of generative AI startups run on its cloud. The company has also secured $58 billion in new revenue commitments over the next two years, more than doubling its current run rate.
To win over early-stage players, Google Cloud is offering serious perks: up to $350,000 in cloud credits, access to its engineers, and a “no-compromise” AI stack — spanning chips, models, and applications. Startups like Loveable and Windsurf are already signing on as primary partners.
Behind the scenes, Google is doubling down on infrastructure too. It’s placing its custom tensor processing units (TPUs) in other providers’ data centers and hosting rivals’ models, like Anthropic’s Claude, on its Vertex AI platform. It’s a delicate balance — competing with others while still powering their tools.
DeSouza frames this approach as openness by design, pointing to Google’s history with open-source breakthroughs like Kubernetes and the transformer architecture. While regulators remain wary of Google’s dominance, positioning itself as the startup-friendly, innovation-first cloud could help the company build goodwill — and future market share.
For Google Cloud, the bet is clear: nurturing today’s scrappy startups may prove more valuable than fighting over the giants.
About the Author
