Google Cloud Next 2026: $240B Backlog, 750M Gemini Users, Agent Search

The financial scale of Google’s AI and cloud ambitions came into sharp focus as Sundar Pichai opened this year’s Cloud Next conference. Google Cloud now generates over $70 billion in annual revenue, a figure growing at a remarkable 48% year-on-year. More striking is the division’s $240 billion backlog, a sum that has more than doubled in twelve months and underscores the immense enterprise demand for its services. To meet these commitments, the company has earmarked a staggering $175 to $185 billion in capital expenditure for 2026, nearly double its spending last year. These numbers frame a pivotal moment, one Pichai described as a fundamental rewiring of technology.

The central theme of the keynote was the transition to an agentic enterprise. Google is positioning itself not merely as a cloud provider but as the integrated operating system for a future where autonomous AI agents handle business operations. The core argument is that Google’s unique advantage lies in its full-stack integration, controlling every layer from custom silicon and frontier models to the cloud platform and consumer-facing products like Search and Workspace, which collectively reach billions of users.

The metrics presented reveal a platform operating at unprecedented scale. The Gemini app now boasts 750 million monthly active users. Its AI-generated search summaries, known as AI Overviews, reach two billion users monthly and now trigger on nearly half of all tracked queries. Behind the scenes, the Gemini API processed 85 billion requests in January alone. This massive user base and query volume are what Google calls its “surface area,” a critical advantage as AI agents become primary interfaces. The thesis is simple: the platform with the broadest reach and deepest integration will win in an agent-driven world.

A profound shift is underway for Google’s most iconic product. Pichai has articulated a vision where Search evolves into an agent manager. Instead of returning links, it will orchestrate AI agents to complete tasks directly on a user’s behalf. The infrastructure for this agentic future is already being constructed. In January, Google unveiled the Universal Commerce Protocol (UCP), an open-source standard developed with major retailers like Shopify, Target, and Walmart. This protocol allows AI agents to interact programmatically with online stores, handling everything from product discovery to payment. This move signals a potential transformation for the digital economy, shifting the foundation from cost-per-click advertising toward a model where AI agents facilitate transactions directly.

Google’s competitive messaging was notably pointed. Thomas Kurian argued that rivals offer disconnected pieces, leaving enterprises to handle complex integration, while Google provides a complete, vertically-aligned platform. This vertical integration spans from custom Ironwood TPUs and next-generation AI chips to the Gemini model family, the developer-focused Gemini Enterprise Agent Platform, and the no-code Workspace Studio. The strategy takes clear aim at competitors like Microsoft Copilot, which has faced adoption challenges and accuracy issues, highlighting the economic necessity of controlling the underlying silicon layer.

The colossal capital expenditure plan is what makes this ambitious strategy tangible, or concerning, to observers. With roughly 60% allocated to servers and 40% to data centers, this spending, combined with similar investments from other tech giants, is reshaping global infrastructure markets and straining power grids. Pichai acknowledged ongoing constraints in compute capacity, land, and power, expecting them to persist through next year. However, the enormous backlog justifies the spend; it represents over three years of contracted revenue, turning the expenditure from a gamble into an obligation to fulfill existing customer commitments.

While Google Cloud holds an 11% market share, trailing AWS and Azure, its growth rate is the fastest among the three and it has recently achieved sustained profitability. The strategy outlined at Cloud Next is not about incrementally closing that gap. It is an attempt to redefine the cloud itself, transforming it from a repository for data and workloads into an active platform where AI agents perform work and make decisions across organizations. If this transition succeeds, the company that built the complete stack,the agents, models, chips, protocols, and distribution,will capture value that current market share figures cannot capture. That is the substantial bet Google has now placed squarely on the table.