Broadcom presents AI networking innovations at the 2026 OCP Global Summit, demonstrating industry leadership in custom silicon solutions.

Broadcom will showcase its latest AI networking innovations at the 2026 Open Compute Project Global Summit, running October 12–15 in San Jose. The company will present its portfolio for scaling AI infrastructure across open systems, including Tomahawk 6, Tomahawk Ultra, and Jericho 4 Ethernet switches; Thor Ultra 800G and Thor 2 400G AI Ethernet NICs; and its third-generation TH6-Davisson Co-Packaged Optics, designed to address the exponential bandwidth demands of frontier AI models.

"Together, the OCP Community and Broadcom have a long history of driving computing at massive scale, and this year's event theme—'Scaling Innovation for the AI Era'—underscores the growing importance of open collaboration," said Charlie Kawwas, Ph.D., president of Broadcom's Semiconductor Solutions Group. "AI infrastructure demands are scaling faster than anything our industry has seen, and Ethernet is the foundation that makes that growth sustainable. With our industry-leading portfolio of networking solutions and new breakthroughs in integrated optics, Broadcom is giving AI labs, hyperscalers and cloud providers the bandwidth and power efficiency they need to scale-up, scale-out, and scale-across next-generation AI clusters."

George Tchaparian, CEO of the Open Compute Project Foundation, characterized OCP's collaborative model as essential to solving the AI data center infrastructure challenge. "The current pace of AI data center infrastructure development depends on an open ecosystem that can support rapid innovation," he said. "Broadcom has been a longstanding contributor to OCP, and we appreciate the technologies and contributions they're bringing to this year's Summit that leverage the open collaborations of the OCP Community, from ORV3 rack-based designs to advances in ORW wide racks and new wide co-packaged optics."

Asad Khamisy, Ph.D., senior vice president and general manager of Broadcom's Core Switching Group, will deliver a keynote on Monday, October 12 at 5:30 p.m. Pacific Time titled "Scaling AI Infrastructure Through Open Innovation." His presentation will outline Broadcom's vision for scaling AI infrastructure, highlighting breakthroughs in open networking, optical interconnects, and modular rack designs capable of supporting clusters of over 100,000 accelerators. Khamisy will address the technical challenges of building switch hardware for AI networks and discuss innovations that enhance bandwidth, reliability, and efficiency for distributed AI computing.

Broadcom experts will lead multiple technical sessions throughout the conference, covering topics including 224G Automated Ethernet Cables for AI data centers, next-generation AI fabric design, open cluster designs, firmware updates, multi-path reliable connections, network configurations for optical AI, Ethernet transport for scale-up networking, co-packaged optics, deep buffers for AI training and inference, and precision timing protocols.

At the show, Broadcom will partner with 15 plus companies to demonstrate its solutions. Key partner exhibitions include Accton/Edgecore Networks (Tomahawk 6 demos), Alpha Networks (liquid-cooled Tomahawk 6), Celestica (Tomahawk 6-Davisson 102.4T CPO), Delta Electronics (Tomahawk 6-Davisson 102.4T CPO), and Micas Networks (Tomahawk 6-Davisson 102.4T CPO).