Equinix and NVIDIA Launch AI Inference Project in Q1 2027

en.Wedoany.com Reported - Equinix announced a major expansion of its long-term collaboration with NVIDIA, launching Equinix Inference Exchange, a distributed AI inference initiative for global enterprises, while also announcing a new partnership with Together AI.

As AI spreads across models, providers, and geographies, where inference runs is becoming a critical variable affecting performance, cost, and governance. Equinix Inference Exchange aims to give enterprises a faster path from AI experimentation to production, enabling them to securely access the data, users, and ecosystems they depend on with low-latency connectivity at the edge. On the delivery side, NVIDIA's enterprise reference architecture is layered with the Together AI inference platform, supporting over 200 open-source models; the full suite runs within Equinix's global data centers, connecting clouds, networks, and AI providers via Equinix Fabric.

The solution was unveiled at Equinix Horizon, the company's inaugural customer and partner event, alongside Equinix Fabric One, designed to simplify networking for globally distributed AI environments. Equinix currently operates more than 280 data centers across 77 metropolitan areas worldwide, offering 230 cloud on-ramps, with over 10,500 enterprises interconnected on its platform; eight of the top ten AI model providers and nine of the top ten AI clouds are deployed on Equinix.

Together AI thus becomes a new member of the Equinix AI ecosystem, bringing the flexibility and choice of open models to enterprises requiring AI deployment at scale. The solution is structured across three complementary layers: Equinix provides underlying support including power, advanced cooling, and Day 2 operations; NVIDIA underpins it with enterprise reference architecture and infrastructure optimized for AI factory throughput and token cost; Together AI runs the platform at the top layer, accommodating both multi-tenant and dedicated single-tenant deployments. Its use cases span metro-edge inference, open model migration, and sovereign AI, corresponding respectively to low-latency edge inference, migration from proprietary models to open-source alternatives, and workloads meeting data residency and sovereignty requirements.

The platform connects inference providers across major metropolitan regions via Equinix Fabric to reduce time-to-first-token. Equinix Inference Exchange will be available starting in Q1 2027.