CoreWeave is entering India with a 240-megawatt AI data-centre deployment in Navi Mumbai, in one of the clearest signs yet that the country’s next wave of AI infrastructure is being built around high-density computing rather than conventional cloud workloads.
The AI cloud provider said on October 6 that it is partnering with AdaniConneX to develop three 80 MW buildings at the latter’s Taloja campus. The facilities are expected to begin coming online in phases from mid-2028, with CoreWeave as the sole tenant across the three buildings. The company also plans to establish an Indian office and hire locally.
CoreWeave is bringing AI infrastructure to Navi Mumbai
The announcement marks CoreWeave’s formal entry into India. Unlike a general-purpose cloud region, the planned campus is being designed specifically for demanding AI workloads, including model training, inference, reasoning and AI agents.
CoreWeave says the deployment will use NVIDIA’s Vera Rubin platform, the company’s newest generation of AI computing infrastructure. The Taloja facilities are also planned with direct liquid cooling for GPU data halls and a non-evaporative chilled-water system, reflecting the much higher power and heat densities of modern AI clusters.
CoreWeave’s announcement describes the initial deployment as 240 MW, with an option to add another 240 MW at the campus. That would take the potential expansion to 480 MW for the CoreWeave deployment, although the additional capacity is not part of the initial commitment.
Why 240 MW matters in India’s AI race
For AI data centres, electricity capacity is increasingly as important as server hardware. A large language model cluster can require enormous amounts of power continuously, while advanced GPUs generate substantially more heat than conventional enterprise servers. That makes access to suitable sites, grid capacity and advanced cooling systems a strategic constraint.
India is already moving quickly on this front. CBRE’s September 2026 data-centre report said India had moved into its “Leading Markets” category in Asia-Pacific alongside Japan, Australia and Mainland China. The firm said cumulative investment commitments in India’s data-centre sector had reached about $173 billion by the end of the first half of 2026, including roughly $38 billion in the first six months of the year.
Mumbai is at the centre of that expansion. CBRE said the city accounted for 52% of India’s operational data-centre stock at the end of the first half of 2026, helped by connectivity, established infrastructure and its position as a major commercial hub.
AdaniConneX is becoming a key AI infrastructure partner
The project also expands the role of AdaniConneX, the 50:50 joint venture between Adani Enterprises and EdgeConneX. The company is building a large data-centre footprint across India, including its Navi Mumbai campus.
According to AdaniConneX’s own data-centre information, its Navi Mumbai site is planned for up to 1,000 MW of IT load and is being designed for very high availability. That makes the location strategically suited to large AI clusters that need substantial power, cooling and network connectivity in one place.
For CoreWeave, the partnership provides something equally important: a local infrastructure platform without requiring the company to build an entire Indian data-centre estate from scratch. CoreWeave can concentrate on its AI cloud software, GPU infrastructure and customer relationships while AdaniConneX supplies the physical environment.
India gets another major AI infrastructure bet
The timing is significant. India’s AI ambitions are increasingly shifting from experimentation toward deployment, which means demand is moving from model access and software to the underlying compute needed to run those systems at scale.
CoreWeave’s entry comes alongside a broader expansion by global technology companies. Amazon, Microsoft and Google are increasing cloud and AI infrastructure investments in India, while domestic operators and infrastructure groups are building additional capacity. The result is a market where data-centre capacity is becoming a strategic part of the AI ecosystem rather than simply a piece of cloud infrastructure.
CoreWeave is also expanding globally as AI demand strains available capacity. The company said it had about 4.2 GW of contracted power as of its August 11 earnings call. It announced its first Asia-Pacific expansion in Indonesia in August, before adding India to that regional footprint.
The real test will be power, delivery and utilisation
The announcement is large, but it is important not to confuse planned capacity with immediately available computing power. The first phase of the Navi Mumbai project is not expected until mid-2028, and additional capacity will arrive later.
That long lead time reflects the difficulty of building AI infrastructure at this scale. Securing electricity is only one part of the challenge. Operators also need high-capacity networking, specialised cooling, construction timelines, reliable power distribution and a steady pipeline of customers capable of using expensive GPU capacity.
CoreWeave’s model makes the project particularly interesting because it is focused almost entirely on AI workloads. If the company can fill the Navi Mumbai deployment with training and inference demand, India could become an important regional base for AI compute rather than simply another market in which global cloud providers sell services.
What this means for India’s AI ecosystem
For Indian startups and enterprises, the most important change will not be the physical buildings themselves but access to specialised compute closer to domestic customers. Locally deployed AI infrastructure can reduce dependence on distant regions for latency-sensitive workloads and give companies another option as demand for GPUs grows.
It also strengthens India’s position in the global AI infrastructure map. The country already has a large developer base and rapidly growing demand for AI applications. What has historically been less abundant is the underlying compute and data-centre capacity needed to support frontier-scale workloads.
The CoreWeave-AdaniConneX project will not solve that gap by itself. But a 240 MW initial deployment, backed by a potential expansion to 480 MW, is substantial enough to show where the market is heading: toward purpose-built AI campuses where power, cooling, networking and accelerated computing are designed together.
For India, the AI race is no longer only about who builds the best model. It is increasingly about who can supply the electricity, chips and data-centre infrastructure needed to run those models at scale.
