Enterprise networks have relied on one connection at a time for decades, and Equinix says that approach simply doesn’t hold up anymore in a world of distributed AI. At its inaugural Horizon event, the company unveiled Equinix Fabric One, a managed connectivity service that lets enterprises specify what they need connected and lets the system handle the rest, rather than forcing teams to assemble networking infrastructure piece by piece across every cloud and AI provider they use.
The service leverages open connectivity specifications AWS and Google Cloud jointly developed, and it lets customers request connections through a portal, APIs, automation workflows, or even natural-language prompts. From there, Fabric One handles routing, cloud connectivity, encryption, resiliency, and failover automatically, packaging what used to require specialized networking expertise into something closer to a single managed request.
Chris Audie, Equinix’s chief product officer, tied the launch to how poorly traditional networking scales when AI workloads demand dynamic, real-time connectivity across distributed environments. IDC’s Andrew Buss backed that assessment, noting that networking remains one of the more manual, labor-intensive parts of enterprise infrastructure, and that organizations increasingly want automated, intent-driven approaches instead of hand-built connections for every partner and provider.
Fabric One runs on Equinix’s existing interconnection footprint, which includes more than 10,500 connected businesses, roughly 3,000 cloud and IT providers, and active deployments from eight of the top ten AI model providers. That neutral positioning sits at the center of the pitch: customers get flexibility to connect across whichever providers they need without locking themselves into a single vendor’s architecture.
Both AWS and Google Cloud helped shape the open specification underlying Fabric One. AWS’s Robert Kennedy said customers shouldn’t have to work out multi-cloud connectivity themselves, they should define what they need and let it happen automatically. Google Cloud’s Rob Enns echoed that, framing cross-cloud interoperability as foundational to how the industry builds AI infrastructure going forward.
Equinix also announced Inference Exchange at the same event, a tool meant to speed up enterprise AI infrastructure deployments. Fabric One itself enters beta later this year, and Equinix plans a full 2027 launch, starting in North America. Hyundai AutoEver America CEO Paul Hager said the service has the potential to help his team manage increasingly distributed AI and multicloud environments more effectively.
