Intel to highlight “silicon to systems” push for AI infrastructure at AI Infra Summit 2026

Intel will use AI Infra Summit 2026 to lay out how it intends to scale AI infrastructure beyond model creation and into real-world deployment, from edge devices and enterprise systems to cloud platforms and physical robotics use cases. The company says the event is designed to showcase progress executing its “vision for AI infrastructure at scale,” emphasizing a systems-level approach that ties together silicon, systems, and software.

A headline item is a fireside chat on September 15 featuring Intel CEO Lip-Bu Tan and Edmund (Ed) Nelson, co-founder and strategy director of AI Infra Summit. Intel frames the conversation around the next phase of AI infrastructure development and what it calls close ecosystem collaboration, arguing that infrastructure is emerging as a differentiator as enterprises implement AI due to performance, efficiency, and cost trade-offs.

Intel also plans multiple panel discussions aimed at the technical and operational barriers it associates with building purpose-built inference infrastructure. One session will feature Huma Abidi, senior director and general manager of AI Frameworks at Intel, addressing the architectural and operational challenges of designing and deploying specialized inference infrastructure for AI workloads.

Another set of discussions is oriented around shifting from experimentation to real-world scale, with Intel pointing to what it calls open infrastructure spanning heterogeneous compute and software frameworks. Intel says Nagesh Puppala, general manager for Physical AI and Robotics within Intel’s client computing and physical AI group, will address how hybrid architectures can combine cloud-scale training and reasoning with edge-based inference and control, supported by developer tools and ecosystem collaboration.

Intel’s agenda further includes a blueprint for agentic AI infrastructure, a term used here to describe AI systems that can take steps toward goals with orchestration and decision-making, rather than only running static responses. Intel says Anil Nanduri, vice president of AI Products and GTM in the data center group, will join Anton McGonnel, vice president of product at SambaNova Systems, to discuss patterns Intel says are emerging as foundational, including heterogeneous computing, disaggregated inference, and intelligent orchestration to scale autonomous AI systems.

Beyond talks, Intel says it will host virtual and on-site hackathons connected to its physical AI technology. The company says developers will be able to build, optimize, and deploy real-world robotics systems, with details directed to the AI Infra Summit hackathon page associated with the event.

Intel’s release also points to commissioned research suggesting robotics adoption is accelerating, while warning that gaps in strategy, skills, safety, and infrastructure could prevent organizations from scaling robotics at the pace they want. The company does not provide the underlying methodology or the specific metrics in the event announcement, but it positions the research as motivation for the infrastructure-focused discussions.

In addition to programming content, Intel says it is leveraging Google Cloud and Google’s Gemini Enterprise as part of efforts it describes as expanding its AI workforce capabilities through scalable agentic workflows across core business functions. Intel does not lay out the specific workflow designs, deployment timeline, or governance measures in the announcement, but it presents the collaboration as part of its broader “edge-to-cloud” narrative.

The company’s wider positioning also references investment intended to create permanent high-tech jobs and engage specialized tradespeople, while expanding leading-edge manufacturing capacity in Ireland. Intel does not tie that manufacturing plan to specific product nodes or timelines in this event release, but it frames the summit as part of a broader push to build the industrial base behind AI infrastructure.