Google CEO Sundar Pichai unveiled the Gemini Enterprise Agent Platform, a comprehensive system for building, deploying, governing, and optimizing AI agents at enterprise scale. The platform evolves from Vertex AI, offering access to 200+ models (including Gemini 3.1 Pro and third-party options like Claude), a no-code Workspace agent builder, managed MCP servers, and the production-grade Agent2Agent protocol for cross-platform agent communication. This is Google's clearest bet yet that the agentic era — not chatbots — is the real enterprise AI battleground.
Mira Murati's 14-month-old startup Thinking Machines Lab signed a multibillion-dollar deal with Google Cloud for AI infrastructure powered by Nvidia's latest GB300 chips. The startup will run training and inference on A4X Max virtual machines delivering 2x speedup over prior GPU generations, making it the third frontier AI developer — behind Anthropic and Meta — to lock in Google's Blackwell and TPU capacity this month. The deal is non-exclusive, and supports Thinking Machines' Tinker product for automated custom frontier model creation.
Google announced its eighth-generation TPUs split into two specialized chips: the TPU 8t for model training and the TPU 8i for inference. This marks a strategic departure from one-size-fits-all silicon, directly targeting Nvidia's dominance in both training and inference workloads. The chips will be available through Google Cloud alongside continued support for Nvidia processors.
Google Cloud announced a $750 million fund to support its 120,000-member partner ecosystem in building and deploying agentic AI solutions. The fund covers AI value assessments, prototyping, deployment, upskilling, and embedded forward-deployed engineers at major consultancies including Accenture, Deloitte, and McKinsey. Partners get early access to Gemini models, signaling Google's intent to win the enterprise agentic market through channel partners rather than direct sales alone.
A major PwC study of 1,217 senior executives across 25 sectors found that three-quarters of AI's economic gains are concentrated in one-fifth of organizations. The gap separating leaders from laggards is not about deploying more tools — leaders are 1.8x more likely to run AI autonomously within guardrails and are increasing decisions made without human intervention at 2.8x the rate of peers. The study underscores that AI's value requires organizational transformation, not just technology adoption.
The Stanford AI Index Report 2026, covered by Nature, finds that despite surging adoption — 6-9% of natural-science publications now mention AI — AI agents remain unreliable on complex, multi-step scientific workflows. Human-AI combinations often performed worse than either humans or AI alone on decision-making tasks, though content creation saw significant gains. The report is a useful corrective to the hype around autonomous AI research agents.
Google announced "auto browse" capabilities for Chrome enterprise users, allowing Gemini to understand live context across open browser tabs and autonomously handle tasks like research, data entry, and workflow automation. This effectively turns the browser itself into an agentic platform, blurring the line between browser and AI assistant. The feature was announced at Cloud Next and targets workplace productivity use cases.
OpenAI and Indian IT giant Infosys announced a partnership to integrate OpenAI's AI tools into Infosys's Topaz AI platform, targeting software modernization, workflow automation, and large-scale AI deployment for Infosys's global enterprise clients. The deal gives OpenAI a major distribution channel into legacy enterprise environments via Infosys's massive consulting footprint. Initial focus areas include software engineering, legacy modernization, and DevOps.
Horizon Robotics announced Stellar, China's first AI chip that consolidates intelligent cockpit and autonomous driving functions onto a single chip with a unified central domain controller. The chip reduces hardware complexity and cost for automakers adopting AI-driven vehicle intelligence. It signals China's continued push toward domestic AI silicon self-sufficiency, particularly in automotive applications.
Google unveiled generative AI features for its mapping and geospatial platforms, targeting enterprise users with enhanced visual and data analytics capabilities. Announced at Cloud Next, the features extend Google's AI integration beyond its core cloud and productivity products into geospatial intelligence. The move positions Google Maps as an AI-augmented enterprise data layer, not just a consumer navigation tool.
