News
Google Cloud reports doubling customer velocity and high utilization of TPUv7 and NVIDIA infrastructure
Wednesday, February 4, 2026 at 09:59 PM
Google Cloud Platform reports significant growth in AI infrastructure utilization, including a 400 percent year-over-year revenue increase from generative AI products. The hardware strategy involves a dual-track approach using both NVIDIA GPUs and custom TPUv7 chips for training and inference. Customer velocity for AI platforms has doubled, and existing clients are exceeding their initial infrastructure commitments by 30 percent.
Context
Google Cloud ended the year with significant momentum, reporting a 2x increase in new customer velocity. This growth is driven by a hybrid infrastructure strategy utilizing both NVIDIA GPUs and proprietary TPUv7 chips for training and inference. A vertical integration approach—spanning chips, models, and platforms—is now utilized by 70% of customers, who consume 1.8x more products than non-AI clients. This high utilization demonstrates that Google’s custom silicon is successfully capturing enterprise demand alongside merchant hardware.
Financial metrics signal a transition to large-scale production, with generative AI model revenue jumping 400% year-over-year. By December 2025, over 350 customers were processing more than 100 billion tokens each, while Gemini Enterprise surpassed 8 million paid seats. With 14 product lines generating over $1 billion annually and customers outpacing initial spending commitments by 30%, Google is showing rapid monetization of its AI supply chain and cloud ecosystem.
Related Companies
Google
GOOGL