Google Cloud Next 2026: Eighth-Gen TPUs, Gemini Enterprise Agent Platform, and Virgo Network Unveiled

Google Cloud unveiled a sweeping set of AI infrastructure advancements at Cloud Next 2026, centering on its vision of the "agentic enterprise" where AI agents act on business data and context at scale.

The company introduced its eighth generation of Tensor Processing Units (TPUs) with a dual-chip approach tailored for the agentic era. The TPU 8t, optimized for training, scales up to 9,600 TPUs and 2 petabytes of shared memory in a single superpod, delivering nearly three times higher compute performance than previous generations and achieving 121 exaflops of compute. The TPU 8i, engineered for inference and reinforcement learning, triples on-chip SRAM to 384 MB and offers 80% better performance per dollar for inference compared to the prior generation.

Google also launched the Gemini Enterprise Agent Platform, described as "mission control" for the agentic enterprise, enabling organizations to build, scale, govern, and optimize AI agents. The platform addresses the complexity of managing thousands of agents across enterprise environments.

The Virgo Network, Google's new scale-out AI data center fabric, can connect 134,000 TPUs in a single data center and over one million TPUs across multiple sites into a training cluster. Its collapsed fabric architecture offers four times the bandwidth of previous generations.

Nearly 75% of Google Cloud customers are now using its AI products, with 330 customers processing over a trillion tokens each in the past 12 months. Google's first-party models process over 16 billion tokens per minute via direct API use, up from 10 billion last quarter.

Concurrently, Microsoft committed to doubling its AI infrastructure capacity within two years, raising its calendar 2026 capital expenditure to $190 billion. Microsoft's AI business achieved an annual run rate of $37 billion in Q3 FY2026, a 123% year-on-year increase. Oracle Cloud Infrastructure also launched OCI Enterprise AI, making models including xAI Grok 4.3 and NVIDIA Nemotron 3 Nano Omni available for enterprise deployment.