How AI startups are fueling Google’s booming cloud business (techcrunch.com)

Google Cloud announced that AI coding startups Lovable and Windsurf have chosen it as their primary cloud provider and are using Google’s Gemini 2.5 Pro models to power their products; Windsurf (recently acquired by Cognition) is also integrating Gemini into Cognition’s AI agent, Devin. The announcements came at Google’s first AI Builder’s Forum, where the company said it added 40+ startups and reiterated that it now works with nine of the 10 leading AI labs and roughly 60% of generative-AI startups. Many of these companies started on Google’s startup programs, which offer up to $350,000 in cloud credits and access to dedicated Nvidia GPU clusters for accelerators like Y Combinator. The deals highlight a broader trend: high training, fine-tuning, and inference costs are fueling demand for cloud infrastructure, and startups are becoming a strategic growth engine for Google Cloud as it competes with AWS and Azure. Google’s cloud business is expanding rapidly—$33.1B in 2023, $43.2B in 2024, a ~$50B annual run rate noted this year, and $58B of new revenue pipeline claimed—while the global cloud market is projected to exceed $400B in 2025. By locking in rising AI firms early and bundling infrastructure with proprietary models (Gemini) and GPU access, Google is positioning itself to capture long-term, higher-margin AI workloads as these startups scale.