Google's cloud chief Thomas Kurian is betting on a new generation of custom chips and advances from the DeepMind AI lab to narrow the gap with Amazon and Microsoft in the fiercely competitive cloud market. The company unveiled the eighth generation of its Tensor Processing Units at an event in Las Vegas this week, with one chip optimized for training AI models and another designed to run AI systems faster through enhanced memory for inference tasks.

After joining Alphabet from Oracle eight years ago, Kurian has grown Google Cloud's market share from 7% to 14%, though it remains a distant third in the $418bn cloud-computing market behind AWS and Azure. The cloud unit reported a 48% revenue jump in the final quarter of 2025 and is on track to generate more than $70bn this year, up from $43bn in 2024. Kurian argues that Google's "full-stack" approach—building chips, data centres, foundation models, and products in-house—provides a competitive edge by retaining more revenue per dollar earned.

The search giant's strategy faces a significant test. It recently committed up to $40bn to Anthropic while providing 5GW of computing capacity over five years, a deal valued at more than $200bn. Capital expenditures are forecast to reach $185bn this year. Kurian predicts market consolidation ahead, stating "Over the next year to two you will see some shakeout in the market" as AI startups burning through billions face pressure from private capital markets reaching saturation.