CoreWeave leases 240 MW of capacity at AdaniConneX’s Taloja campus

CoreWeave announced that it will lease a total of 240 MW of data‑center capacity in Navi Mumbai.¹ The leased power will be split evenly across three separate buildings, each rated at 80 MW.¹ CoreWeave will be the exclusive tenant of all three structures.¹ The lease gives the company an option to double the amount of space it occupies, potentially raising its total commitment at the site to 580 MW.¹² Together, the three buildings account for the full 240 MW announced in the lease, while each individual building represents one-third of that capacity.¹ Exclusive tenancy means the three buildings are assigned to CoreWeave rather than divided among tenants, giving the company the entire announced capacity at the campus.¹ The possible 580 MW total is an expansion scenario, not the starting lease: CoreWeave has an option to add capacity by doubling its commitment.¹² That option would take the potential site total beyond the initial 240 MW, although the facts specify no timetable for exercising it.¹²

The first phase of the Taloja campus is slated to become operational in mid‑2028, with additional capacity to be delivered in later stages.¹ The campus incorporates high‑density AI infrastructure, including direct liquid cooling and non‑evaporative chilled water systems.¹ CoreWeave plans to install Nvidia Vera Rubin systems at the site, extending its recent rollout of the NVL72 hardware on its cloud platform.¹ The staged delivery means the entire planned capacity is not described as arriving all at once: the initial phase is scheduled for mid‑2028, followed by later tranches.¹ The cooling features are specifically identified as part of the campus’s high-density AI infrastructure, the type of facility planned to host CoreWeave’s Vera Rubin systems.¹ The announcement connects a named hardware plan with a location designed for high-density AI infrastructure, while the delivery schedule places the initial phase in the future.¹ CoreWeave has also announced that Vera Rubin NVL72 systems are available on its cloud infrastructure, so the company’s plans encompass both systems at Taloja and cloud-platform availability.¹

In parallel with the lease, CoreWeave will open an office in India and recruit local staff to support its operations.¹ Company executive Sachin Jain highlighted India’s talent pool, developer community, and ambition as key factors for building an AI hub.¹ Jain also noted that CoreWeave is investing in infrastructure, technology, and a local presence to deliver the 240 MW capacity, Nvidia hardware, and its expertise to Indian customers.¹ The local office and recruitment plans mean the expansion includes a stated local presence, not only a lease of data-center capacity.¹ Jain’s comments identify talent, developers, and ambition as strengths in India; these are the qualities he associated with the country’s potential as an AI innovation hub.¹ His description of the investment spans infrastructure, technology, and local presence, linking the capacity and hardware plans with the company’s intention to serve customers in India.¹ The people directly referenced in these plans include local personnel to be recruited and customers in India whom CoreWeave intends to reach with its capacity, systems, and expertise.¹

CoreWeave’s total contracted power stood at 4.2 GW as of August 2026, indicating that the new Indian lease represents a modest but strategic addition to its global footprint.¹ The company has recently announced expansion into Indonesia, showing a broader focus on emerging markets.¹ The 4.2 GW contracted-power figure provides a company-wide comparison point for the 240 MW India lease, which is a smaller part of CoreWeave’s reported contracted power.¹ The figures describe different scales: one is CoreWeave’s contracted power overall, while the other is the capacity announced for a single campus in Navi Mumbai.¹ The Indonesia expansion is another announced geographic move, alongside the India lease, though no capacity figure for Indonesia is supplied here.¹ Together, the India and Indonesia announcements show that CoreWeave has recently reported activity in both countries, while the facts do not state the size of the Indonesian expansion.¹

The lease is part of a larger partnership with AdaniConneX, a joint venture between Adani Enterprises and EdgeConneX.¹ AdaniConneX’s portfolio already includes facilities in Chennai, Noida, Vizag, Hyderabad, Pune, and Mumbai.¹ The joint venture was originally announced in February 2021 with a target of 1 GW of hyperscale capacity across India.¹ Adani later stated an ambition to reach 5 GW of data‑center capacity in the country by early 2026.¹ The Taloja deal sits within a company whose listed project locations span several Indian cities, with Mumbai among them.¹ The joint venture’s original target was 1 GW, and Adani later stated a goal of 5 GW for data-center capacity in India.¹ Those targets provide context for the 240 MW lease: it is a specific campus commitment within the broader capacity ambitions described for India.¹ The facts distinguish the joint venture’s original hyperscale target from Adani’s later data-center goal, rather than describing them as the same commitment.¹

By securing exclusive tenancy of the three buildings, CoreWeave ensures dedicated access to the high‑density cooling infrastructure required for AI workloads.¹ The option to double its lease provides flexibility to scale up to 580 MW if demand from Indian AI customers grows.¹² The exclusive arrangement combines the full three-building capacity with the campus’s cooling features, all within the facility planned for high-density AI infrastructure.¹ The 580 MW figure describes the potential expanded total, not the amount CoreWeave has already announced as its initial lease.¹² The option therefore gives CoreWeave a route to increase its campus capacity, while the stated facts do not say whether or when the company will use it.¹² For customers in India, the announced lease and the option together describe an initial capacity commitment and a possible larger footprint at the same campus.¹²

The Indian market is seen as a fertile ground for AI development, with local talent and developer ecosystems cited as strengths.¹ CoreWeave’s investment aligns with its strategy to bring advanced GPU‑based cloud services closer to end users in the region.¹ The talent and developer strengths in this account come from Jain’s statement, which also refers to India’s ambition to become a major AI innovation hub.¹ CoreWeave’s stated effort to serve Indian customers involves infrastructure, Nvidia technology, expertise, and a local presence, rather than a capacity lease alone.¹ The company’s cloud-platform announcement about Vera Rubin NVL72 systems complements its plans to install Vera Rubin systems at Taloja, though the facts do not specify when the site installation will occur.¹ The people and organizations most directly identified as affected are local personnel CoreWeave plans to recruit and customers in India it intends to serve.¹

Overall, the 240 MW lease adds a new AI‑focused data‑center capacity layer to CoreWeave’s portfolio while complementing AdaniConneX’s broader goal of expanding hyperscale infrastructure across India.¹ The deal combines three 80 MW buildings, exclusive occupancy, and an option for a larger total, making Taloja a defined site commitment with room for potential expansion.¹² Its initial capacity is smaller than CoreWeave’s reported 4.2 GW of contracted power, but it is a concrete addition located in India and tied to local hiring plans.¹ For AdaniConneX, the lease is one activity at a campus operated by a joint venture with projects in multiple Indian cities and an original national hyperscale target of 1 GW.¹ The 5 GW ambition Adani announced later is larger than that original target, placing the Taloja lease within a wider expansion objective rather than making it the entirety of that goal.¹ The announcement thus describes a relationship involving a cloud provider, a data-center joint venture, local employees, and customers in India, with staged delivery beginning in mid-2028.¹