OneQode’s 14 MW Chennai Agreement: What’s Signed—and What’s Still a Discussion

Three figures, three different statuses

The announcement is easiest to read when each figure is kept in its own category. OneQode’s primary release identifies the 14 MW as the signed agreement. It describes the larger amount as additional India capacity opportunities still being discussed; the over-US$6-billion GPUaaS figure is a potential total-contract-value projection attached to that larger scenario.[1][2]

What can be said about the Chennai site?

OneQode names Siruseri SIPCOT, Chennai as the location of its 14 MW agreement. AdaniConneX separately identifies its Chennai 1 data centre at SIPCOT IT Park. The public OneQode announcement, however, does not name “Chennai 1” or identify a particular building or phase. We therefore treat AdaniConneX’s campus figures as local facility context—not as a confirmed allocation of OneQode’s capacity to a specific phase.[1][3][4]

The units and scope matter: AdaniConneX explicitly labels its 17 MW and 33 MW Chennai 1 figures as IT load; OneQode’s release says only “data-centre capacity.” Without a stated basis and phase mapping, the figures should not be compared one-for-one, subtracted to estimate headroom, or used to claim that 14 MW fills a particular Chennai 1 phase.[3][4]

What the agreement says—and leaves open

OneQode calls the deal its first major AI infrastructure deployment in India. It says the Chennai capacity is intended to support large-scale AI training and inference, enterprise AI and sovereign-AI workloads, and that the site will connect into its global carrier network. Those are company-stated plans and intended uses; the announcement does not document that the capacity is already commissioned or serving customers.[1]

The company’s May 2026 AMD announcement described a broader global rollout anchored by AMD Instinct MI355X GPUs in initial deployments, with AMD Helios planned for a future phase. That release does not say which hardware, if any, is assigned to Chennai. It is therefore context on OneQode’s global AI strategy, not a confirmed bill of materials for this 14 MW agreement.[6]

What the US$6-billion TCV projection means

The distributed announcement says the potential expansion of up to 114 MW could generate more than US$6 billion in GPUaaS total contract value (TCV) for OneQode. This is a company projection tied to the potential expansion scenario—not revenue already earned, a signed customer contract, contracted backlog, or the signed 14 MW agreement’s disclosed value. The release does not publish the assumptions behind the projection, such as contract duration, customer commitments, GPU deployment, utilisation or pricing.[2]

For clarity, this is also not the same as OneQode’s separately announced 110 MW Norway project. In May 2026, OneQode said it had signed a binding letter with Bitzero for a 15-year lease covering 110 MW at a Norway site. That is a distinct geography and separate arrangement; it does not turn the India-wide 114 MW discussion into a signed Chennai or India deployment.[7]

A short timeline—and what is still missing

OneQode’s July announcement is useful context, but it did not quantify the Chennai deployment. The October agreement adds a specific signed capacity figure; it does not, by itself, verify that the deployment is live. As of the 1 November 2026 research cutoff, the sources reviewed give no Chennai service-start or commissioning date. Do not import the Norway project’s schedule into the India announcement.[1][5]

Why the announcement matters—without overstating it

A signed capacity agreement is a more specific signal than an unquantified regional rollout statement: it names an infrastructure counterparty, a Chennai location and an initial 14 MW. OneQode describes the intended workloads and network connection, while its CEO has cited Chennai’s regional position and subsea-cable connectivity as part of the company’s location rationale in comments reported by Data Centre Magazine.[1][8]

Those statements are not evidence of a completed deployment, spare grid capacity, a particular cable path, customer demand, GPU availability or an operating SLA. The company has not publicly specified how the 14 MW is measured, when it will be available, or what portion—if any—of the additional 114 MW will be contracted. The next useful confirmations would be a commissioning milestone, a facility/phase and IT-load definition, the hardware and power configuration, and clearer terms for any expansion.

Quick answers