Google used Cloud Next 2026 to make one message clear: the company wants to be seen not only as an AI model developer, but as a full-stack infrastructure partner for enterprises moving AI into daily operations. In a post published on April 22, CEO Sundar Pichai said Google Cloud is entering a new phase of momentum, with customer demand rising across models, chips, and enterprise AI tools.
At the center of the announcement was Google’s push toward what it calls the “agentic” era. According to Pichai, Google’s first-party models are now processing more than 16 billion tokens per minute through direct customer API use, up from 10 billion in the previous quarter. Google also said nearly 75% of Google Cloud customers are already using its AI products, while 330 customers processed more than one trillion tokens each over the last 12 months.
Google expands its enterprise AI platform
A major focus of this year’s Cloud Next was Gemini Enterprise. Google is positioning it as an end-to-end platform that connects enterprise data, employees, and workflows with AI agents. Pichai said paid monthly active users of Gemini Enterprise grew 40% quarter over quarter in the first quarter, signaling stronger commercial traction for the product. Reuters also reported that Google is rebranding and expanding parts of Vertex AI under the Gemini Enterprise banner as it sharpens its focus on enterprise deployments.
This matters because Google is trying to move beyond experimental AI use cases and into broader enterprise adoption. At the event, executives emphasized governance, scalability, and production-readiness, suggesting Google wants to compete not just on model quality, but on how easily businesses can build, manage, and secure AI systems at scale.
New TPU chips support training and inference
Google also used the event to introduce its eighth-generation Tensor Processing Units, TPU 8t and TPU 8i. The company says TPU 8t is designed for large-scale model training, while TPU 8i is optimized for low-latency inference, which is especially important for AI agents expected to respond quickly and handle complex tasks. In its chip announcement, Google said both processors were custom-engineered for the next phase of AI computing and will become available later this year.
Reuters reported that TPU 8i delivers 80% better performance for fast inference workloads than the previous generation, while TPU 8t can scale to large training clusters. The hardware rollout reinforces Google’s strategy of combining proprietary chips, models, cloud services, and security tools into one enterprise AI stack.
Another notable signal came from capital spending. Pichai reaffirmed Alphabet’s plan to spend $175 billion to $185 billion in 2026, with just over half of the company’s machine learning compute investment expected to support the Cloud business. That level of investment shows Google is willing to keep spending heavily to strengthen its position against Amazon, Microsoft, and emerging AI infrastructure rivals.
Overall, Cloud Next 2026 showed Google taking a more aggressive enterprise stance. Instead of focusing only on headline AI breakthroughs, the company is trying to prove it can provide the infrastructure, chips, software, and governance enterprises need to operationalize AI at scale. For cloud customers, that makes Google’s latest push less about experimentation and more about long-term adoption.