OneQode signed an agreement with AdaniConneX for 14 MW of data center capacity in Siruseri Sipcot, Chennai, announced on October 28, 2026. The company called it the first major AI infrastructure deployment in India, with discussions continuing for up to 114 MW of additional capacity. The potential expansion could generate over $6 billion in GPU-as-a-service total contract value. For business, this adds another large-scale source of training and inference capacity in Asia.
Terms of the Chennai deployment
Under the signed deal, OneQode takes an initial 14 MW at the Chennai site for large-scale AI training and inference, enterprise AI and sovereign AI workloads. The agreement is only for that initial block, while up to 114 MW remains under discussion as demand grows. OneQode linked the larger figure to more than $6 billion in GPU-as-a-service total contract value, which frames the commercial scale under consideration. Chief Growth Officer Johnny Blackfield pointed to infrastructure, competitive operating costs and room to scale as reasons for building in Chennai.
Technically, the Chennai capacity will plug into OneQode global platform that combines high-performance compute with a carrier-grade network. The company operates deployment capability across more than 30 data centers on five continents and holds sovereign telecommunications licenses in multiple jurisdictions. It describes itself as an operator that designs, builds, owns and operates compute and network infrastructure for frontier model providers, enterprises and AI-native firms. That structure supports sovereign AI deployments where compliance and licensed carrier operations matter alongside raw compute.
The hardware path builds on a collaboration with AMD announced on May 26, 2026. OneQode plans a phased rollout anchored by AMD Instinct MI355X GPUs in initial deployments, with the AMD Helios rack-scale solution as a platform foundation on the open AMD ROCm software stack. OneQode was named an AMD Instinct Cloud Provider in August 2026. The Chennai site is part of a wider Asia-Pacific expansion alongside Bangkok, Thailand, with further regional capacity expected through 2026 and 2027 for governments, financial services and research institutions.
What this means for enterprise AI buyers
For companies using AI, the Chennai buildout widens options for enterprise and sovereign workloads hosted in India rather than in distant regions. OneQode positions India between EMEA and Asia-Pacific as a strategic point for latency-sensitive inference across both regions, paired with its low-latency global carrier network. Large organizations gain an additional vendor combining compute and network under one operator, useful for training runs and production inference. Smaller firms are unlikely to contract megawatts directly, but they can benefit indirectly through cloud and GPU-as-a-service offerings built on this capacity.
The limits are clear: only 14 MW is secured, while the 114 MW and the associated $6 billion figure remain a possibility tied to future demand. No commissioning date for the Chennai AI capacity was disclosed in the announcement, unlike the Norway project targeted for the first half of 2027. AdaniConneX Chennai 1 campus itself offers 17 MW in its initial phase, scaling to 33 MW at full capacity, so fit-out and power details matter for buyers. Enterprises should ask about GPU type and availability, network latency to their users, compliance coverage for sovereign workloads, and whether capacity is reserved or shared.
The marker to watch is whether talks for up to 114 MW convert into signed capacity and how the Thailand and Norway rollouts proceed on schedule. AdaniConneX is also building 240 MW for CoreWeave in Navi Mumbai, with first phase expected in mid-2028, which will show the pace of hyperscale delivery in India. If OneQode adds tranches in Chennai through 2027, enterprise buyers will have firmer evidence of available GPU supply and regional inference coverage.
