BOSTON, May 4, 2026, 18:03 (EDT)
IBM said on Monday its Db2 Genius Hub now supports Google Vertex AI and Intel Gaudi AI accelerators, widening the ways customers can run AI inference for database operations across cloud and on-premises systems. Inference is the stage where an AI model produces answers, recommendations or other output from data.
The move lands as large companies try to push generative AI from pilots into production, where latency, infrastructure cost and where data is stored can decide whether projects scale. IBM framed the update as a choice issue for Db2 customers with different cloud, hardware and data-sovereignty needs.
IBM executives Ashok Kumar, Satya Krishnaswamy, Bryan Tang and Miran Badzak wrote that the expansion gives Db2 customers the “freedom to build and scale AI inferencing exactly how they want.” Google Vertex AI is Google Cloud’s platform for training and deploying machine-learning models and AI applications, while Intel says Gaudi 3 is built for workloads such as large language models, multimodal models and enterprise retrieval-augmented generation, a method that pulls business data into AI responses. IBM
Db2 Genius Hub had already supported Amazon Bedrock, AMD Instinct accelerators and Nvidia H100 GPUs. The latest additions put Google Cloud and Intel deeper into IBM’s database AI stack, while keeping Amazon, Nvidia and AMD in the competitive mix around enterprise AI infrastructure.
In a separate IBM post, Kumar, Krishnaswamy, Tang and Intel software solutions architect Murali Madhanagopal wrote that “AI must work reliably where enterprise data lives.” IBM said Gaudi was used as an inference server for Db2 Genius Hub agents, with tests covering concurrent requests, contextual search and tool-calling workflows, a term for AI systems invoking software tools to complete tasks. IBM
The announcement is part of a broader Db2 Genius Hub release tied to IBM Think 2026. IBM said the system is moving from advice toward supervised action, allowing AI agents to propose and execute database operations after user approval, with wider features including MCP access, host-level diagnostics and natural-language scheduling planned to begin rolling out in June. MCP, or Model Context Protocol, is an open standard for connecting AI applications to tools and systems.
That shift is where the market is heading, at least in the view of IDC research director Devin Pratt. In a recent post, Pratt wrote that IBM is positioning Db2 Genius Hub around governed daily operations such as tuning, performance management and incident response, and said autonomous database operations are moving from a differentiator to “table stakes.” LinkedIn
IBM is using Think 2026 to put that message in front of customers. Chief Executive Arvind Krishna is due to open the company’s annual conference in Boston on Tuesday, where IBM said it will outline enterprise AI announcements for a gathering of more than 5,000 business and technology leaders from over 80 countries.
But the update leaves some details open. IBM did not disclose pricing, named customers or benchmark figures for the new Google Vertex AI and Intel Gaudi support, and the payoff for clients will depend on whether the new inference paths meet production demands without raising cost or governance risk.
IBM shares closed at $229.48 on the New York Stock Exchange, down 1.17%, and were quoted at $230.50 in after-hours trading at 18:00 EDT, MarketScreener data showed.
