BOSTON, May 4, 2026, 18:03 (EDT)
IBM on Monday announced that its Db2 Genius Hub adds support for Google Vertex AI and Intel Gaudi AI accelerators, opening up more options for customers to run database AI inference both on-premises and in the cloud. Inference refers to the phase when an AI model generates answers, recommendations, or other outputs from data.
The shift comes as major firms keep working to move generative AI out of pilot territory and into actual production—where issues like latency, infrastructure spend, and data location can make or break scaling plans. IBM pitched this Db2 update as a flexibility play, aiming to give customers options around cloud, hardware, and data-sovereignty requirements.
IBM’s Ashok Kumar, Satya Krishnaswamy, Bryan Tang, and Miran Badzak say the move opens up “freedom to build and scale AI inferencing exactly how they want” for Db2 users. Google Vertex AI serves as Google Cloud’s hub for building and running machine-learning models and AI tools. Intel, for its part, positions Gaudi 3 as targeting big tasks like large language models, multimodal AI, and enterprise retrieval-augmented generation—which lets AI tap into a company’s own data. IBM
Db2 Genius Hub previously worked with Amazon Bedrock, AMD Instinct accelerators, and Nvidia H100 GPUs. Now, Google Cloud and Intel are stepping further into IBM’s database AI stack, but Amazon, Nvidia, and AMD remain in play as rivals for enterprise AI infrastructure.
Kumar, Krishnaswamy, Tang, and Intel’s Murali Madhanagopal put it this way in a separate IBM blog: “AI must work reliably where enterprise data lives.” According to IBM, Gaudi served as the inference server for Db2 Genius Hub agents. The team ran tests on concurrent requests, contextual search, and what they call tool-calling workflows—AI tapping software tools to get tasks done. IBM
IBM’s new announcement, part of the larger Db2 Genius Hub rollout linked to IBM Think 2026, marks a shift: the system will now let AI agents not just suggest, but actually carry out database tasks—with user approval. Starting in June, users can expect to see wider feature access, including MCP integration, host-level diagnostics, and natural-language scheduling. MCP stands for Model Context Protocol, an open standard designed to link AI apps with tools and systems.
IDC research director Devin Pratt sees the market turning in that direction. In a recent post, Pratt said IBM is centering Db2 Genius Hub on routine governed operations—tuning, performance management, incident response. Autonomous database operations, he noted, are shifting from something that sets products apart to simply “table stakes.” LinkedIn
IBM is spotlighting that message at Think 2026. On Tuesday in Boston, Chief Executive Arvind Krishna is set to kick off the company’s flagship conference, with IBM planning to roll out enterprise AI news to more than 5,000 business and tech leaders from upward of 80 countries.
Still, plenty of questions remain. IBM hasn’t shared pricing, listed any customers by name, or released performance benchmarks for the Google Vertex AI and Intel Gaudi integrations. Whether clients benefit will hinge on how these new inference routes perform under real production workloads—ideally, without bumping up costs or governance headaches.
IBM stock ended the session at $229.48 on the New York Stock Exchange, slipping 1.17%. By 18:00 EDT, MarketScreener data had it quoted at $230.50 in after-hours action.
