Superpower Daily
Enflame’s Retail IPO Tranche Is Subscribed 4,073 TimesAnthropic’s Fable 5.1 Takes Benchmark Lead, but Its Top Setting...Google Launches Fairwind to Give Selected Defenders AI Tools fo...Loudoun County Confronts Data-Center Backlash After Reaping a T...EU Seeks AI Act Answers From 30-Plus Companies, With Fines for...Job Seeker Sends ChatGPT to AI Recruiter After Five Interviews...Enflame’s Retail IPO Tranche Is Subscribed 4,073 TimesAnthropic’s Fable 5.1 Takes Benchmark Lead, but Its Top Setting...Google Launches Fairwind to Give Selected Defenders AI Tools fo...Loudoun County Confronts Data-Center Backlash After Reaping a T...EU Seeks AI Act Answers From 30-Plus Companies, With Fines for...Job Seeker Sends ChatGPT to AI Recruiter After Five Interviews...
Infrastructurepublished

Equinix Sets Q1 2027 Start for Distributed Enterprise AI Inference Service

The planned service combines NVIDIA infrastructure designs, Together AI model serving and Equinix connectivity. Its test is whether one package can simplify placing AI workloads across providers and locations.

By 2 min read
Equinix Sets Q1 2027 Start for Distributed Enterprise AI Inference Service
Equinix Sets Q1 2027 Start for Distributed Enterprise AI Inference Service
Sign in to save and react

Listen to this story

The audio brief

About 1:28
0:001:28
Read transcript
Equinix plans to launch Inference Exchange in the first quarter of 2027, giving enterprises a proposed way to run AI workloads across different locations, clouds, and model providers. It is not available yet. The service brings together three layers: NVIDIA’s validated infrastructure designs, Together AI’s model-serving platform, and Equinix’s data centers and network connections. Customers would connect through Equinix Fabric to clouds, networks, and AI providers, while Together AI would provide access to more than 200 open-source models. The pitch is about placement and flexibility, rather than introducing a new model. Companies could put inference closer to users or data to reduce latency, move from proprietary models toward open-source alternatives, or choose processing locations to meet data-residency and sovereignty requirements. Equinix says its footprint includes more than 280 data centers across 77 metropolitan areas, 230 cloud on-ramps, and over 10,500 interconnected businesses. Together AI’s platform is also planned to support both multitenant deployments, where capacity is shared, and dedicated single-tenant environments. That combination could simplify a fragmented infrastructure problem, but the important details remain unresolved. The test is whether Equinix can deliver these deployment choices across the promised geographies and provider connections when the service is supposed to start in 2027.

Story brief

3 key points

Equinix is planning a distributed AI inference service for Q1 2027, combining NVIDIA infrastructure designs, Together AI’s model-serving platform, and Equinix data centers and network links. Inference Exchange would let enterprises place workloads closer to users or data, shift between proprietary and open models, and choose processing locations for residency requirements. The proposal covers more than 200...

  1. 01

    Equinix says the service will connect customers through Equinix Fabric to clouds, networks, and AI providers.

  2. 02

    Together AI’s platform supports more than 200 open-source models and planned multitenant or single-tenant deployments.

  3. 03

    Equinix cites 280-plus data centers across 77 metros, 230 cloud on-ramps, and 10,500 interconnected businesses.

Equinix has announced Inference Exchange, a planned service for enterprises running AI workloads across locations, clouds and providers. Built with NVIDIA and Together AI, the offering is scheduled to become available in Q1 2027, making it a future deployment rather than a service customers can use today.

A distributed serving stack

The service combines NVIDIA Enterprise Reference Architectures, Together AI’s inference platform and Equinix’s global data-center infrastructure. NVIDIA supplies validated infrastructure designs, Together AI runs the platform layer, and Equinix supplies the facilities and network connections around it.

Equinix says customers will connect through Equinix Fabric to clouds, networks and AI providers. Together AI’s platform supports more than 200 open-source models, giving the proposed service a model-serving layer alongside its infrastructure and connectivity components.

Where enterprises may use it

Equinix identifies three intended scenarios: putting inference nearer to users and data, moving workloads from proprietary models to open-source alternatives, and selecting processing locations for data-residency or sovereignty needs.

  • Metro edge inference is intended to lower latency by placing AI processing closer to users and data.
  • Open-model migration is intended to provide a production path from closed models to open-source alternatives through Equinix’s interconnected fabric.
  • Sovereign AI is intended to let regulated or geographically constrained organizations choose where data and inference processing occur.

A larger connectivity strategy

Equinix says it operates more than 280 data centers in 77 metros, with 230 cloud on-ramps and more than 10,500 interconnected businesses. It places Inference Exchange alongside Fabric One in its Connected Cloud strategy, which the company describes as a neutral foundation for distributed AI architectures.

Together AI’s platform is planned to support multitenant deployments, where capacity is shared, and dedicated single-tenant environments. The commercial question is whether Equinix can deliver those options across the promised locations and provider connections when the service reaches its planned 2027 start.

Sources

  1. prnewswire.comEquinix Accelerates AI Inference for Enterprises with NVIDIA and Together AI
  2. blog.equinix.comThe Coordination Economy: How Enterprises Really Build AI Value

Related coverage

Continue reading