Google expands agentic enterprise push

Google used Cloud Next ’26 to position itself as a more integrated enterprise Artificial Intelligence provider, combining models, infrastructure, security, and multicloud data services. The strategy broadens its reach into enterprise software while emphasizing interoperability with rival clouds and platforms.

Google used Cloud Next ’26 in Las Vegas to present a broader push into the agentic enterprise, tying together its cloud, security, and Artificial Intelligence offerings around a more integrated operating model. The centerpiece is the Gemini Enterprise Agent Platform, which incorporates and replaces Vertex Artificial Intelligence and adds services for creating, orchestrating, securing, simulating, tracking, and monitoring agents in Google Cloud. The platform also adds Model Context Protocol support and an Agent Marketplace with integrations for Atlassian, Box, Oracle, ServiceNow, Workday, and other enterprise systems.

Google’s strategy centers on owning more of the full stack than many rivals. It combines proprietary models, cloud infrastructure, custom chips, networking, and security controls into one platform, with the company arguing that this approach can improve enterprise control, isolation, and protection for sensitive workloads. The launch also signals a more direct move into enterprise software territory, increasing pressure on vendors such as Microsoft, Oracle, Salesforce, and ServiceNow as investors weigh how agentic Artificial Intelligence could reshape traditional software categories.

To support the new services, Google expanded its Artificial Intelligence Hypercomputer architecture with new compute and networking components. TPU 8t, optimized for training, which can scale up to 9,600 TPUs and 2 PB of shared, high-bandwidth memory in a single superpod. Google says it achieves 3x the processing power of its Ironwood predecessor and delivers up to 2x more performance/watt. TPU 8i, optimized for inference, directly connects 1,152 TPUs in a single pod. Google says it delivers 80% better performance per dollar for inference than the prior generation and supports millions of concurrent agents. Google Axion N4A instances of Google’s custom Arm-based Axion CPUs, which Google says provide twice the price/performance compared to current-generation x86-based VMs. Google’s Virgo Network links either NVIDIA Vera Rubin NVL72 systems or TPU 8t superpods into supercomputers with hundreds of thousands of accelerators.

Even as Google expands competitively, it is also emphasizing multivendor support. Its Artificial Intelligence Hypercomputer supports NVIDIA chips and Intel processors. In Agentic Data Cloud, Google introduced a Cross-Cloud Lakehouse based on Apache Iceberg that lets customers query data from AWS or Azure without moving data out of those clouds. Google is also providing integrations with Microsoft Office 365, reinforcing a message that customers can adopt its agentic services without giving up existing tools and cloud relationships.

Security is another major layer of the rollout. Google added capabilities from Wiz, which it bought for ? billion recently, including the Wiz Artificial Intelligence Application Protection Platform, Wiz Red, Blue, and Green Agents, and a Wiz Agentic Workflows hub. It also introduced Agentic SecOps tools such as Dark Web Intelligence, a Threat Hunting Agent, and a Detection Engineering Agent, alongside Google Cloud Fraud Defense, which evolved from reCAPTCHA to identify humans, bots, and agents. To speed adoption, Google also created resources and incentives for software partners, channel partners, systems integrators, and consulting firms including Accenture, BCG, Deloitte, and McKinsey.

68

Impact Score

China still blocking Nvidia H200 chip sales

Nvidia has yet to complete H200 sales into China even after the United States reopened exports. Chinese authorities are reportedly limiting imports as Beijing pushes buyers toward domestic semiconductor suppliers.

OpenAI prepares GPT-5.5 launch

OpenAI is reportedly preparing GPT-5.5, its first fully retrained base model since GPT-4.5, as it pushes harder into enterprise software. The model is expected to bring native multimodal capabilities and stronger support for agent-based workflows.

Meta expands AWS Graviton deal for agentic Artificial Intelligence

Meta is expanding its partnership with AWS by deploying Graviton processors at scale for its next generation of Artificial Intelligence systems. The move highlights growing demand for CPU-heavy agentic Artificial Intelligence workloads alongside continued reliance on GPUs for model training.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.