Equinix launches AI inference exchange with Nvidia

Equinix has launched Inference Exchange with NVIDIA and Together AI for enterprise AI inference deployments.

The service combines NVIDIA's enterprise reference architectures, Together AI's inference platform, and Equinix's global data centre and network estate.

Inference Exchange is designed to bring AI inference workloads closer to enterprise data, applications, and users, rather than relying on more distant centralised deployments. The approach is intended to help companies move AI projects from testing into production while maintaining secure, low-latency connections to the systems and providers those workloads depend on.

Together AI's platform supports more than 200 open-source models. The service will be delivered through Equinix's global data centre footprint and connected to cloud providers, networks, and AI services through Equinix Fabric.

Production focus

The launch comes as large companies look for ways to run inference across multiple clouds, models, and regions while maintaining control over cost, governance, and performance. Equinix argues that where inference runs has become a central infrastructure decision as businesses move from AI experimentation to operational use.

Managing distributed inference adds operational complexity, as businesses must decide not only which infrastructure to use, but also where workloads should run and how they connect to data sources and applications.

"AI is transforming enterprise technology at extraordinary speed, and the infrastructure decisions enterprises make today will define their competitive position for years to come. Equinix is uniquely positioned to deliver what this moment demands based on our nearly three decades building the trusted exchange where the world's enterprises run, connect and orchestrate their most critical workloads," said Adaire Fox-Martin, Chief Executive Officer and President, Equinix.

"Our longtime relationship with NVIDIA delivers the accelerated computing foundation at the heart of modern AI, while Together AI's commitment to open ecosystems gives enterprises the flexibility to scale on their terms. Equinix Inference Exchange will enable architectures that are neutral by design, open by default and engineered for exceptional performance."

NVIDIA said the arrangement extends its existing relationship with Equinix and places its AI infrastructure within a broader distributed deployment model. Its enterprise reference architectures are intended to give customers a tested blueprint for building AI systems in Equinix facilities.

"Equinix Inference Exchange turns the world's leading digital interconnection platform into a global fabric for AI inference," said Raj Mirpuri, Vice President of Global AI Clouds and Infrastructure Ecosystem, NVIDIA.

"As accelerated compute becomes a strategic asset class, combining NVIDIA's infrastructure & technology with Together AI's open-model inference platform and Equinix's global reach gives enterprises a powerful, distributed foundation to bring intelligence closer to their data, applications and customers - accelerating the next generation of intelligent services."

Open models

Together AI's role centres on access to open-source models and on providing shared and dedicated deployment options. Customers will be able to use multitenant environments for shared efficiency or single-tenant environments when they need dedicated capacity.

The service is also aimed at companies that want to move workloads away from proprietary models and towards open-source alternatives. That use case reflects growing enterprise interest in controlling AI costs and reducing dependence on a single model supplier.

"Together AI was built on the conviction that open, accessible AI is what will define the industry moving forward, because enterprises shouldn't have to choose between model performance and operational flexibility," said Vipul Ved Prakash, Co-Founder and Chief Executive Officer, Together AI.

"What we are building with Equinix and NVIDIA proves that model choice and performance are not trade-offs. They are the foundation of enterprise AI done right."

Equinix also outlined several deployment scenarios for the service, including metro edge inference for lower-latency responses closer to users, open-model migration for enterprises shifting from closed systems, and sovereign AI deployments for sectors and regions with data residency requirements.

Scale and reach

Equinix is relying on its existing infrastructure footprint as a differentiator. It operates more than 280 data centres across 77 metros, offers 230 cloud on-ramps, and interconnects more than 10,500 businesses on its exchange platform.

The company added that eight of the top 10 AI model providers and nine of the top 10 AI clouds already deploy with Equinix. That ecosystem density underpins its argument that enterprises can connect inference workloads to data, applications, and partners without building new links from scratch.

Industry analysts say these deployment questions are becoming more urgent as AI systems spread across more environments. The challenge is no longer just training models, but placing inference workloads in locations that meet business and regulatory demands.

"Performance, cost and governance have become strategic considerations as AI workloads grow more distributed across providers, data sources and environments," said Nick Patience, Vice President & Practise Lead, AI Platforms, The Futurum Group.

"Organisations are increasingly focused on where inference runs and how quickly it can be deployed into production. Solutions that simplify inference deployment while preserving flexibility will become increasingly important to achieve business outcomes."