OneQode has signed an agreement with AdaniConneX for 14 megawatts of data centre capacity in Siruseri Sipcot, Chennai, India, its first major AI infrastructure deployment in the country, the company announced on October 28, 2026. The Chennai deployment will support large-scale AI training and inference, enterprise AI and sovereign AI workloads, and the two companies are continuing discussions on additional capacity of up to 114 megawatts in India. OneQode says that expansion path could generate over $6 billion in GPU-as-a-service total contract value, a forward-looking figure that depends on those discussions turning into signed capacity.
Why Chennai, Why Now
India’s position between Europe, the Middle East and Asia-Pacific makes it a strategic location for OneQode’s global AI infrastructure platform, the company says. As inference workloads become increasingly latency-sensitive, serving them from India can reach users across both regions without routing everything through a distant hub. The Chennai site will connect into OneQode’s existing infrastructure, extending its ability to serve AI workloads across strategic global markets, and it combines large-scale compute capacity with the company’s own low-latency global carrier network.
“India is becoming one of the most important markets for AI infrastructure globally,” said Johnny Blackfield, OneQode’s chief growth officer and president for Europe, the Middle East and India. “Working with AdaniConneX gives us a strong foundation to build at scale in Chennai. We have secured our initial 14 MW deployment and are continuing discussions with AdaniConneX on additional capacity opportunities of up to 114 MW in India as demand grows. The combination of infrastructure, competitive operating costs and room to scale makes India a compelling market for our customers.”
Part Of A Wider Asia-Pacific Push
The Chennai deployment forms part of OneQode’s wider Asia-Pacific AI infrastructure expansion, alongside its deployment in Thailand. It also builds on OneQode’s collaboration with AMD, announced in May 2026, to deploy AMD Instinct graphics processors, and the company says it plans for AMD’s Helios rack-scale solution as part of its wider AI infrastructure rollout. That hardware context matters: capacity announcements are only useful if the chips to fill them arrive, and OneQode is tying its Chennai story to a named GPU supplier rather than to generic compute.
OneQode describes itself as a global AI infrastructure operator that designs, builds, owns and operates compute and network infrastructure for frontier model providers, enterprise buyers and AI-native companies. Its platform combines high-performance AI compute with its own carrier-grade network, optimised for low latency to support real-time AI applications. Its licensed carrier operations and established compliance framework support sovereign AI deployments in strategic markets, a selling point for governments and enterprises that want AI workloads kept within national borders.
The 14MW Starting Point, And The 114MW Question
Fourteen megawatts is a modest start by hyperscale standards, where single campuses now run into hundreds of megawatts. Its significance lies in what it starts rather than what it is: a first major deployment in a new country, with a named local partner, AdaniConneX, one of India’s fastest-growing data centre providers, and a stated path to eight times that capacity if demand materialises. The $6 billion total contract value figure attached to that path should be read as OneQode’s estimate of what the larger deployment could be worth over its contract life, not as revenue already booked.
That distinction is worth keeping in view across the AI infrastructure boom, where announced capacity, discussed capacity and contracted capacity are often reported as if they were the same thing. OneQode’s announcement is careful on this point: the 14 megawatts are signed, the 114 megawatts are under discussion. Investors and customers should judge the story on whether the second number follows the first, and on how quickly the Chennai site fills with real workloads.
Why It Matters
This matters because the geography of AI infrastructure is shifting. Training clusters once concentrated in a few US locations, but inference, sovereign AI and latency-sensitive applications are pushing capacity into new regions, including India, where power, land and connectivity costs differ sharply from established hubs. OneQode’s bet is that Chennai can serve both Asia-Pacific and EMEA workloads from a single location, a claim that will be tested by the actual customers it signs.
It also matters as a sign of how infrastructure competition is widening beyond the hyperscalers. Smaller operators that combine compute with their own networks, like OneQode, are carving out niches in sovereign and enterprise AI, as in our report on NetBox and NEYON’s AI factory platform and our report on Argentum AI’s $5 billion chip financing. If Chennai fills quickly, it will validate that model in one of the world’s fastest-growing AI markets; if it does not, it will show how hard it is for new entrants to convert announced capacity into used capacity.
Background
OneQode combines high-performance compute with its own global carrier network and deployment capability across more than 30 data centres on five continents. The company supports AI training, inference and enterprise workloads, with deployments designed around customers’ capacity, connectivity, location and sovereignty requirements.
Source: OneQode announcement
0 Comments