Equinix has expanded its NVIDIA partnership and formed a new collaboration with Together AI to launch Equinix Inference Exchange. Built as a distributed enterprise‑grade AI inference architecture, the platform moves compute workloads nearer core data stores, end‑users and operational applications. Unveiled alongside Equinix Fabric One at Equinix Horizon, it unites NVIDIA enterprise hardware, Together AI’s open‑source model platform and Equinix’s global interconnection infrastructure to resolve latency, data‑sovereignty and network bottlenecks.

Addressing Distributed Inference and Data Gravity As enterprise AI moves from prototypes to high‑throughput production, inference placement shapes total cost, response latency and compliance. Still, running distributed infrastructure across heterogeneous environments brings operational overhead and network complexity.

Equinix mitigates these pain points using its 280+ data centers across 77 metros, 230 cloud on‑ramps and over 10,500 connected‑enterprise ecosystem. Eight of the top‑ten AI‑model providers and nine of top‑ten AI clouds already reside inside Equinix sites, making the platform a neutral aggregation layer for distributed‑AI pipelines.

Architecture and Stack Integration Equinix Inference Exchange comprises three layers: physical infrastructure, compute, and software. Equinix operates the underlying physical‑interconnect foundation covering power delivery, advanced liquid‑cooling, ongoing facility operations, plus hybrid‑data connectivity via Equinix Fabric. NVIDIA delivers the compute‑and‑validation layer with enterprise‑AI hardware optimized for high token throughput and lower per‑query inference expenses. Together AI supplies the inference‑serving software layer supporting 200+ open‑source models. It supports shared multi‑tenant setups for compute efficiency and isolated single‑tenant deployments for workloads requiring dedicated capacity.

Traffic routed over Equinix Fabric lowers regional users’ time‑to‑first‑token (TTFT), and creates private peering links between on‑premises storage and model endpoints.

Enterprise Deployment Targets and Availability Three core deployment modes are supported. Metro‑edge inference runs models within local metro hubs for low‑latency API responses alongside centralized management and private interconnects. Open‑source‑model migration offers a smooth transition path away from closed proprietary APIs toward open foundation‑models, cutting vendor lock‑in while re‑using existing network fabrics. For regulated sovereign‑AI use‑cases, its distributed design lets enterprises contain inference and data processing inside defined geographic boundaries to satisfy strict data‑residency and governance rules.

Executives from Equinix, NVIDIA and Together AI state this integration delivers a vendor‑agnostic high‑throughput inference fabric balancing compute performance and operational agility.

Equinix Inference Exchange will reach general enterprise availability in Q1 2027.

Beijing Qianxing Jietong Technology Co., Ltd.
Sandy Yang/Global Strategy Director
WhatsApp / WeChat: +86 13426366826
Email: yangyd@qianxingdata.com
Website: www.qianxingdata.com/www.storagesserver.com
Business Focus:
ICT Product Distribution/System Integration & Services/Infrastructure Solutions
With 20+ years of IT distribution experience, we partner with leading global brands to deliver reliable products and professional services.
“Using Technology to Build an Intelligent World”Your Trusted ICT Product Service Provider!