At Cloud Next in Las Vegas this week, a striking contrast emerged from the keynote stages. While Google announced generative AI features for Google Maps—enhancing visual navigation and data analytics for everyday users—it simultaneously closed the kind of deal that reshapes the AI infrastructure landscape. Mira Murati's newly founded Thinking Machines Lab has signed a multibillion-dollar commitment with Google Cloud, securing compute capacity powered by Nvidia's latest GB300 chips.
The deal, confirmed by TechCrunch, represents something beyond ordinary cloud procurement. Murati left OpenAI with a specific vision: building AI systems that are more configurable and steerable than current frontier models. Her lab's choice of Google Cloud as a foundational partner signals that the search giant's AI loyalty strategy has evolved from passive cloud sales into active ecosystem cultivation. Google is no longer simply renting out GPUs—it is purchasing long-term alignment with the researchers most likely to shape where AI goes next.
The GB300 chips underpinning this arrangement matter technically. Nvidia's latest architecture delivers significant improvements in training throughput for large language models, making the infrastructure underneath Murati's work a genuine competitive advantage. But the real story is the curation layer Google has built around this compute access. The startup showcase at Cloud Next displayed a long list of AI-first companies whose infrastructure runs on Google Cloud—a deliberate strategy to attract the next generation of AI builders into Google's orbit before they scale elsewhere.
This creates a compounding effect. Startups that train on Google's chips deploy on Google's infrastructure. Their engineers become fluent in Google Cloud's tooling. When the next Murati-scale researcher spins up a new lab, the gravitational pull toward Google Cloud will already be established. Maps AI, meanwhile, serves as the consumer-facing proof point that Google's AI investments translate into products people actually use—a narrative anchor that reinforces enterprise confidence.
The competitive implications are stark. Microsoft Azure has OpenAI. Amazon Web Services has cultivated its own roster of AI partners. But Google's approach—pairing marquee research relationships with a curated startup ecosystem and integrated consumer products—creates a three-layer moat that pure infrastructure plays cannot easily replicate. The multibillion-dollar price tag for Thinking Machines Lab is not the cost of cloud compute. It is the cost of ensuring that the next decade of AI innovation runs on Google's terms.