Google introduced a new agentic AI platform and expanded its infrastructure spending plan during its Cloud Next event in Las Vegas. The company said it is building systems designed to support autonomous AI agents across cloud computing, software development, and cybersecurity operations.
Chief executive Sundar Pichai stated that planned capital expenditure may reach between 175 billion and 185 billion dollars in 2026. The investment aligns with the rising demand for AI systems that can perform tasks with limited human input. It also expanded partnerships and tools for developers and enterprises.
Google said it is moving from chatbot-based tools toward agentic AI systems that can complete multi-step tasks. The company introduced the Gemini Enterprise Agent Platform, designed for organisations to build and manage AI agents at scale.
Pichai stated, “As we move into the agentic era, we are taking this to the next level.” The platform targets cloud, coding, and enterprise workflows across multiple sectors.
The company linked its infrastructure expansion to its capital expenditure plan of between 175 billion and 185 billion dollars. Google said this spending will support data centers, computing systems, and AI development tools. The platform supports both proprietary and open models, including Gemini and Gemma.
Google also announced a $750 million fund for its Google Cloud partner ecosystem. Firms such as Accenture, Deloitte, and McKinsey & Company will receive access to engineering support and early tools for deployment.
Google reported increased use of AI agents within its internal engineering operations. Pichai stated, “Today, nearly 75% of all new code at Google is AI-generated and approved by engineers.”
The company said this figure increased from 50% in the previous year. Human engineers continue to review all AI-generated code before release. Google said internal workflows now rely more on agent-based systems across software teams.
The company also applied AI systems in cybersecurity operations. Google said its security teams handle large volumes of threat reports each month. Pichai noted, “Our security operation center agents automatically triage tens of thousands of unstructured threat reports each month.”
Google reported that response times for threat mitigation have dropped by over 90%. AI tools now filter and organise data to support faster analysis of security risks across cloud environments.
Google said its partnership with NVIDIA continues to support its AI infrastructure development. Both companies collaborate on hardware, software, and cloud systems for large-scale computing needs. New infrastructure includes NVIDIA Blackwell and Blackwell Ultra GPUs.
Google also introduced A5X instances powered by Vera Rubin systems. The platform supports training, inference, and simulation workloads used in enterprise and research settings. It also includes confidential computing features.
Google said enterprises and developers are using its systems to deploy AI applications in production environments. Companies such as Salesforce, Snap Inc., and Schrödinger are among the adopters.
The platform also supports robotics and digital twin applications across industries. Google said more than 90,000 developers joined its joint community with NVIDIA within a year. The company also stated that OpenAI uses its infrastructure for large-scale inference workloads across AI models.
Also Read: NVIDIA Stock Slips as Meta Weighs Multi-Billion Google TPU Partnership