The deal splits into three roles and one open question: Equinix supplies power, cooling and its 230 cloud on-ramps; Nvidia supplies the reference architecture; Together AI operates the inference platform and, per CNBC’s September 2, 2026 reporting, acts as seller of record billing end customers directly. Equinix isn’t disclosed as taking a cut of that revenue — CNBC noted financing details for the deal weren’t disclosed — which means the company most associated with the announcement may be the one least exposed to its upside.
Equinix’s $100 billion market cap says investors are paying for interconnection real estate, not for a stake in whoever ends up billing the inference workloads that run through it.
The buyer side is enterprises chasing three use cases DCD flagged: metro-edge latency, migration off closed models onto Together AI’s roster of 200-plus open-source models, and sovereign workloads that must stay in-jurisdiction. The competitive read is sharper. CNBC’s September 2 figures put Equinix’s market cap at $100 billion versus Digital Realty’s $68 billion, a gap that’s widened as Equinix’s stock climbed 33% this year — but analyst Vlad Galabov told CNBC that Equinix was “too slow” to react to gigawatt-scale AI-factory demand, ceding that tier to hyperscalers and neoclouds like CoreWeave. Inference Exchange reads as Equinix conceding the training-scale buildout and doubling down on the smaller, denser, interconnection-heavy footprint it already owns.
Watch whether Together AI’s role as billing intermediary becomes the template other neoclouds copy to get distribution without building colocation themselves, and whether Equinix’s companion Fabric One service — entering beta later this year with 2027 general availability — turns out to matter more than Inference Exchange itself, since it’s the piece that actually lets Equinix charge for orchestrating connections rather than just hosting them.
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.
