Major Tech Firms Form EBO Coalition for AI Data Center Optical Standards

A coalition of technology companies announced a new multi-source agreement (MSA) on May 12, 2026, aimed at standardizing expanded beam optical (EBO) connectivity for AI data centers. The initiative addresses growing physical layer bottlenecks as hyperscale AI deployments push optical interconnects beyond traditional multi-fiber contact connector limitations.

3M led the announcement, revealing participation from AMD, Arista Networks, Cisco, Meta, Oracle, and 12 other industry players. The group will develop open, interoperable specifications for EBO connector solutions across the data center ecosystem. The official press release details the technical framework and member roster.

Expanded beam optical technology differs fundamentally from physical contact connectors. Instead of fiber-to-fiber contact, EBO systems use lenses to expand and collimate light beams before transmission. This design eliminates the strict hygiene requirements that plague traditional connectors—dust, scratches, or contamination on contact surfaces cause signal degradation that requires meticulous cleaning protocols (a maintenance nightmare for anyone who's managed a live data center).

The coalition's stated goal is reducing deployment complexity while improving reliability in high-density environments. As AI workloads scale, the physical layer faces new constraints. Bandwidth density increases demand for more resilient Layer 1 infrastructure. Current multi-fiber physical contact connectors introduce operational overhead through ongoing link triage and strict cleaning requirements that slow network builds.

Oracle serves as co-chair of the MSA formation. Rajagopal Subramaniyan, senior vice president of OCI networking, emphasized the operational friction points: "Strict connector hygiene requirements slow network builds and add operational overhead for ongoing link triage." The EBO approach aims to enable more resilient cluster topologies and future rack-scale optical architectures.

Member diversity spans the entire supply chain. Connector manufacturers like Senko, TE Connectivity, and Amphenol sit alongside hyperscalers including Meta and Oracle. Network equipment vendors Arista Networks and Cisco join chipmakers AMD and optical specialists viaPhoton, Source Photonics, and Xscape Photonics. This cross-ecosystem participation signals industry-wide recognition that proprietary solutions won't scale.

Jim Hasegawa, president of the Optical Communications Division at SENKO Advanced Components, noted the timing: "As optical data networks scale and evolve rapidly, the industry faces increasing demand for solutions that deliver not only high performance, but also reliability and ease of deployment and operation." The MSA framework enables seamless integration across transceivers, backplanes, and cable assemblies.

The initial technical working group has begun developing the first connector specification. The MSA remains open to additional members across the data center and networking ecosystem. More information is available at www.ebomsa.org, with Richard Ward serving as administrator and co-chair.

This standardization effort arrives as AI infrastructure spending accelerates. Data center operators face a practical problem: proprietary optical solutions create vendor lock-in while increasing maintenance complexity. An open standard approach theoretically reduces costs through multi-sourcing while improving interoperability. The physical reality matters—technicians installing thousands of connections need solutions that work reliably without specialized cleaning equipment or precision alignment tools.

Industry analysts note this positions the EBO MSA differently from previous optical standardization attempts. Earlier efforts focused primarily on transceiver form factors. This coalition addresses the connector layer itself, where physical contact issues create cascading reliability problems. The expanded beam approach trades some insertion loss for dramatically improved tolerance to contamination and misalignment.

3M's participation builds on broader materials science commitments to data center innovation. The company's solutions help enable reliable connectivity, manage heat and power, and support resilient infrastructure at scale. Alex An, vice president of 3M's data center vertical, stated the MSA helps enable "an open, standards-based approach that can accelerate adoption, improve reliability, and support the next generation of AI infrastructure."

The coalition's success depends on adoption rates across hyperscale operators. Standardization efforts often face friction from companies with proprietary investments. However, the member list includes major cloud providers who collectively represent significant deployment volume. Their participation suggests EBO technology has reached a maturity threshold where standardization benefits outweigh proprietary advantages.

Whether this coalition actually delivers interoperable products at scale remains the real question. Standards groups frequently announce ambitious frameworks that take years to materialize in shipping hardware. The initial working group's progress on the first connector specification will determine momentum. Operators watching this space will want concrete timelines, not just member announcements.

For now, the EBO MSA represents a coordinated industry response to physical layer constraints in AI infrastructure. The technology addresses genuine pain points in data center operations. Whether it becomes the dominant standard or remains a niche solution depends on execution speed and member commitment to the open specification framework.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn