The 2026 OCP Global Summit, October 12–15 in San Jose, brings the OCP Community together across 22 technical tracks addressing the technologies shaping the next generation of open infrastructure.
From compute, networking and silicon to power, cooling, storage, firmware and validation, the program goes deep into the engineering challenges behind increasingly complex AI systems—while also addressing interoperability, deployment, adoption and the open standards needed to move innovation into production. Whether you work at the chip, server, rack, cluster or data center level, here’s a look at what’s ahead.
AI Clusters
Go inside the AI factory with sessions spanning scale-up, scale-out and scale-across networking; 800G/1.6T interconnects; BYO-XPU architectures; cluster telemetry and validation; liquid cooling; and AI factory operations.
AI Computing Continuum
See how AI inference is extending beyond hyperscale data centers through agentic AI, edge server and rack architectures, XPU options, open modular standards, and deployments across neocloud, colocation, enterprise, telco and industrial environments.
AI Open Data Center: Power Distribution
Dive into the electrical infrastructure required for megawatt-scale AI factories, including LVDC architectures up to 1500VDC/800VDC, two- and three-wire DC systems, medium-voltage conversion, energy storage system (ESS) integration, safety and protection, and fault mitigation.
Cooling Environments, Cooling: Cold Plate & Cooling: Immersion
Across three days of cooling tracks, go from the facility to the chip with CDUs, door heat exchangers, heat reuse, coolant chemistry, flush/fill standards, fleet-scale operations, immersion cooling, and single- and two-phase cold plate design.
Data Centric Computing
Learn how compute is moving closer to data through near-data processing, computational memory, direct xPU access, GPU-to-device communication, privacy-preserving AI training and cross-vendor API alignment.
Market Impact & Adoption
See how open infrastructure moves into production through AI factory economics, factory-built infrastructure, standardized platforms, OCP-aligned supply chains, sovereign AI, regional deployments, Open BOM interchange and real-world OCP adoption.
Networking
Dig into the technologies connecting AI at scale, including ESUN 1.0, UALink-SAI, MetaRoCE, optical circuit switching, 102.4T liquid-cooled switching, routing beyond BGP/ECMP and linear pluggable optics.
Open Chiplet Economy
Follow the evolution of custom AI silicon through chiplet and IP reuse, chiplet-based SoC security, FCSA interoperability, HBM integration, AI infrastructure and silicon co-design, BoW Flexi connectivity and open chiplet standards.
Open Platform Firmware (OPF)
See what’s next for open firmware with flash-less streaming boot, UPCI, host-delivered libraries, memory-safe firmware stacks, agentic firmware automation, openSFI, Intel FSP and silicon interoperability.
Photonics
Examine the transition from copper to photonics through XPO, NPO and CPO architectures; large-scale AI cluster interconnects; engineering trade-offs; reliability and resiliency at scale; and open standards for interoperability.
Quantum Technologies
Look ahead to the integration of quantum technologies into general-purpose data centers, including fault-tolerant quantum systems, quantum communications, live testbeds, infrastructure requirements and approaches to scalable, sustainable deployment.
Rack & Power
Get into the engineering behind unprecedented rack density with 800V DC and direct-to-server power, 1MW+ racks, Open Rack Wide validation, high-density GPU power, HVDC grounding, backup technologies and AI power oscillation filtering.
Server
Follow the evolution of server architecture through MHS and M-PNP, OCP NIC 2026 Gen 7 and liquid cooling, 400G-to-800G networking, 448G co-packaged copper, power and thermal co-design, and memory for agentic inference.
Server: Composable Memory Systems (CMS)
See how the industry is decoupling memory from compute through CXL-based memory expansion, pooling and sharing; memory tiering and disaggregation; software and orchestration; benchmarking; reliability; open specifications and reference designs.
Storage
See how storage is adapting for AI and inference with QLC adoption, AI-optimized SSD architectures, KV cache and LLM prompt caching, EDSFF, NVMe, DPU technology, testing and telemetry, and OCP Storage Specifications.
Test & Validation
Follow new approaches to hardware validation at AI scale, including CTAM GPU compliance testing, AI-powered validation and debugging, wind tunnel testing, fan reliability, cable validation, standardized diagnostics and L10–L12 manufacturing testing.
Time Appliances (TAP)
Go from network to workload with time uncertainty, multidomain synchronization, PTP boundary clocks, NIC timing recovery and ePPS, clock-aligned telemetry, resilient A-PNT, reference clock phase noise and precision-time circuit switching.
And There’s More!
These highlights are only part of the 22-track technical program. The 2026 OCP Global Summit also features dedicated content across AI Open Data Center, Security, Short Reach Optical Interconnect, Sustainability, Systems Management and the Symposium.
Beyond the main Summit program, co-located workshops bring together industry experts and technical communities for deeper dives into manageability, SONiC, scalable AI infrastructure, optical interconnects and more. Official co-located workshops are included with Summit registration.View the co-located workshops and add them to your Summit plans.
Browse all 22 tracks and the full 2026 Global Summit schedule to start building your agenda.
As the OCP Community continues to grow, attendance for the 2026 OCP Global Summit may reach capacity before the event. Registration will close once maximum capacity is reached, and onsite registration may not be available. Register today to secure your spot.