TL;DR
Adani Group has proposed investing ₹1 trillion (approximately $12 billion) to build a 1 gigawatt (GW) AI-focused data center in Odisha, seeking 250 acres of land from the state government. If approved, it would be the largest AI data center investment by an Indian company and the second-largest overall in India after Google's $15 billion, 1 GW campus in neighboring Andhra Pradesh. The proposal is under evaluation by Odisha's state government as of August 2026 — it has not been approved yet.
The Odisha project sits inside a much bigger story: Adani announced a $100 billion AI data center program in February 2026, targeting 5 GW of renewable-powered capacity across India by 2035. The Odisha proposal, if cleared, would be a cornerstone of that national plan.
What Is Confirmed vs. What Is Speculative
Confirmed by primary sources:
- Adani Group submitted a proposal to invest ₹1 trillion for a 1 GW AI data center in Odisha, seeking approximately 250 acres of land (Business Standard, August 3, 2026)
- The proposal is under evaluation by the Odisha state government, which is assessing land, power, water, and connectivity before clearance (CommunicationsToday, August 2026)
- If approved, it would rank as the biggest AI data center investment by an Indian company and the second-largest in India overall after Google's Andhra Pradesh campus (Business Standard)
- Adani Group announced a $100 billion commitment to renewable-powered AI data centers by 2035 in February 2026, targeting 5 GW total capacity (Reuters, Bloomberg, February 17, 2026)
- AdaniConneX — the 50:50 joint venture between Adani Enterprises and EdgeConneX — is already operational with a Chennai facility (17 MW phase 1, scaling to 33 MW), with additional sites in Mumbai, Pune, Hyderabad, Noida, and Visakhapatnam (AdaniConneX press release; DataCenterDynamics)
Not yet confirmed:
- The exact site location — Cuttack has been reported as the tentative site, but the Odisha government has not finalized land allocation
- Whether the Odisha project would be developed under the AdaniConneX JV (like most Adani data centers) or as a standalone Adani Enterprises project
- Final power sourcing agreements — Adani's broader plan emphasizes renewable energy, but site-specific PPAs for Odisha have not been announced
- Construction timeline — no start date has been set, since the proposal is still under evaluation
Why Odisha? The Steel-and-Mining State Pivoting to AI
Odisha is not an obvious data center destination. It lacks the submarine cable density of Mumbai or Chennai, the hyperscaler anchor tenants of Hyderabad, or the established ecosystem of Pune. So why is Adani betting ₹1 trillion there?
Power surplus. Odisha generates surplus electricity from its coal and hydroelectric plants — a critical advantage when AI data centers draw 30-80 MW per facility and gigawatt-scale campuses need dedicated power procurement. The state's grid can absorb massive new loads without the transmission bottlenecks choking Maharashtra and Tamil Nadu.
Land availability and cost. 250 acres in Cuttack costs a fraction of equivalent land in Navi Mumbai or Chennai's OMR corridor. Odisha's government has been actively courting data center investment, designating Info Valley in Bhubaneswar as a technology hub.
Water resources. AI workloads run hot. Liquid cooling for GPU clusters at gigawatt scale requires enormous water supply. Odisha's river systems — particularly the Mahanadi basin — offer more cooling capacity than the water-stressed states where most Indian data centers currently cluster.
Renewable energy potential. Adani's $100 billion plan is explicitly tied to renewable-powered data centers. Odisha has growing solar capacity and adjacent states (the Adani Group controls massive solar projects in Rajasthan and Gujarat) can wheel renewable power through the national grid.
Existing Adani footprint. Adani already has a smaller ₹800 crore data center project at Info Valley, Bhubaneswar — announced by Karan Adani at a groundbreaking ceremony on April 8, 2026, alongside a ₹30,181 crore thermal power plant near Cuttack and a ₹2,100 crore cement facility. The group also operates a cement grinding unit and ports in the state. The ₹1 trillion AI proposal builds on this existing relationship with Odisha.
A growing Odisha data center cluster. Adani's proposal is not happening in isolation:
- HCLTech announced a ₹14,257 crore ($1.48 billion) AI data center at the Odisha Sovereign AI Park in Bhubaneswar on July 24, 2026, in partnership with Sarvam AI and the Odisha government (HCLTech press release)
- The Reserve Bank of India has a ₹169 crore data center facility in the state nearing completion (CommunicationsToday)
- A ₹266.48 crore state-run next-generation data center is under development
- Adani's existing ₹800 crore hyperscale data center MoU at Info Valley, Bhubaneswar
Odisha is methodically building an AI infrastructure corridor — and Adani's ₹1 trillion proposal would make it a national-tier hub, not a regional one.
How Adani's Proposal Compares: The Three Gigawatt Projects
India's AI data center landscape is entering a gigawatt era. Three projects stand out — and the comparison reveals distinct strategies.
Adani Group — Odisha (₹1 trillion / ~$12B / 1 GW)
| Dimension | Detail |
|---|---|
| Investment | ₹1 trillion (~$12 billion) |
| Capacity | 1 GW |
| Location | Odisha (site under evaluation, Cuttack reported) |
| Land sought | ~250 acres |
| Power | Renewable energy (per Adani's broader $100B plan) |
| Status | Proposal under government evaluation (Aug 2026) |
| Structure | Unknown — could be AdaniConneX JV or standalone |
Google + AdaniConneX + Airtel — Visakhapatnam, Andhra Pradesh ($15 billion / 1 GW)
| Dimension | Detail |
|---|---|
| Investment | $15 billion |
| Capacity | 1 GW |
| Location | Visakhapatnam, Andhra Pradesh — 600+ acres across Tarluvada, Adavivaram, and Rambilli |
| Timeline | Multi-phase, 2026–2030 |
| Power | Renewable energy; subsea cable landing station included |
| Status | Partnership announced October 2025; construction phased |
| Structure | AdaniConneX + Google + Airtel strategic partnership |
Source: Google's official blog announcement; BlackRidge Research project tracker.
Reliance Industries — Visakhapatnam, Andhra Pradesh (~₹98,000 crore / ~$11B / 1 GW)
| Dimension | Detail |
|---|---|
| Investment | ₹98,000 crore ($11 billion) via Digital Connexion JV |
| Capacity | 1 GW |
| Location | Visakhapatnam, Andhra Pradesh — 400 acres |
| Power | Backed by a planned 6 GW solar project |
| Status | MoU signed with AP Economic Development Board (Nov 2025) |
| Structure | Digital Connexion (Reliance JV) |
Source: Bloomberg, CNBCTV18, November 2025; BlackRidge Research.
The pattern that matters
All three gigawatt-scale Indian AI data center proposals share two traits: renewable energy backing and a coastal or power-surplus location. But they diverge on structure. Google's campus is a hyperscaler-anchored partnership (Google brings the workload, Adani brings the infrastructure). Reliance is building for its own ecosystem (Jio, retail, telecom). Adani's Odisha proposal — if it is a standalone Adani Enterprises project — would be the first gigawatt-scale Indian-built, Indian-owned AI data center without a named hyperscaler anchor. That is the most ambitious and the riskiest bet, because building the facility is the easy part; filling 1 GW of AI compute capacity requires either hyperscaler tenants or a domestic AI services business that does not yet exist at that scale.
Why Adani Is Betting $100 Billion on AI Compute
The Odisha proposal does not exist in a vacuum. On February 17, 2026, Adani Group announced a $100 billion commitment to build renewable-powered AI data centers across India by 2035, targeting 5 GW of total capacity. Reuters reported that the group expects this direct investment to catalyze an additional $150-250 billion in broader AI infrastructure investment across the Indian ecosystem.
The strategy has three pillars:
1. Energy-to-compute vertical integration. Adani is India's largest private power producer and its largest solar developer. AI data centers are essentially power plants that convert electricity into compute. By owning both the energy generation and the data center, Adani can offer something no pure-play data center operator can: guaranteed renewable power at controlled costs over a 10-15 year horizon. In a market where power is 40-60% of a data center's operating cost, this is a structural advantage.
2. Land and real estate advantage. The Adani Group controls port-adjacent land across India's coastline — Mundra, Dhamra (Odisha), Visakhapatnam, Ennore, and Mormugao. Port-adjacent sites are increasingly valuable for data centers because they offer submarine cable landing options, which are essential for latency-sensitive AI inference workloads and for connecting Indian compute to Asian and Western markets.
3. Hyperscaler partnerships without hyperscaler ownership. AdaniConneX already hosts Google's Visakhapatnam campus. Adani has also announced partnerships with Microsoft spanning Hyderabad and Pune. The group is positioning itself as the infrastructure layer — the company that builds and operates the physical facilities — while hyperscalers bring the customers. The Odisha project, if standalone, would test whether Adani can fill capacity without a named anchor tenant.
What This Means for India's AI Data Center Market
India's data center capacity is forecast to grow from 2.2 GW in 2025 to 12 GW by 2030 — a compound annual growth rate of approximately 40%, according to Wood Mackenzie. AI-dedicated capacity specifically is projected to grow from 275 MW to 6,546 MW, a 24-fold increase.
Cushman & Wakefield's 2026 report pegs India's current operational capacity at 1.6 GW, making it the second-largest Asia-Pacific market after Singapore, with 3.1 GW under construction or planned.
The numbers vary by source and methodology — Wood Mackenzie counts total capacity including AI-dedicated; Cushman & Wakefield tracks operational vs. pipeline separately. But every forecast agrees on the direction: India is in the middle of a data center super-cycle, and AI is the primary demand driver.
The capacity gap nobody talks about
India's 2.2 GW of data center capacity sounds large until you compare it to the United States (over 20 GW) or even Singapore (~1.4 GW in a city-state). The 12 GW by 2030 target would make India the third-largest data center market globally, but it requires building roughly 10 GW of new capacity in five years — equivalent to 2 GW per year. For context, India delivered approximately 450 MW of new capacity in all of 2026.
The gigawatt-scale proposals from Adani, Google, and Reliance collectively represent 3 GW of planned capacity. That is 30% of the 10 GW gap. The remaining 70% has to come from dozens of smaller projects — and the constraint is not capital, it is power, land, and skilled labor.
The talent bottleneck
A 1 GW AI data center is not just a building full of servers. It requires specialized expertise in GPU cluster operations, liquid cooling systems, high-voltage power distribution, and AI workload orchestration. India's IT services industry — TCS, Infosys, Wipro, HCLTech, and others — has deep software talent but limited operational experience with hyperscale AI infrastructure. This is why HCLTech's entry into AI data centers via the Odisha Sovereign AI Park matters: it signals that Indian IT companies are building the operational muscle to run these facilities, not just the software that runs on them.
What This Means for You
If you are an enterprise evaluating Indian cloud or AI infrastructure
The Adani Odisha project signals that India will have significant domestic AI compute capacity by 2028-2030 — but not overnight. If you are planning multi-year cloud or AI infrastructure commitments, factor in:
- Data localization compliance. India's Digital Personal Data Protection Act and sector-specific data residency rules are driving demand for domestic data processing. The Adani, Google, and Reliance gigawatt campuses all serve this need.
- Cost arbitrage. Indian AI compute is likely to be 20-40% cheaper than US or Singapore equivalents once these campuses come online, driven by lower renewable energy costs and land costs.
- Vendor diversification. AdaniConneX is positioning as a colocation and build-to-suit provider — an alternative to AWS, Azure, and GCP for companies that want infrastructure in India without hyperscaler lock-in.
If you are an AI startup or researcher in India
The Odisha proposal, combined with HCLTech's Sovereign AI Park investment, signals that India is building AI infrastructure specifically for sovereign AI workloads — government data processing, domestic language models (Sarvam AI partnership), and public-sector AI applications. If your startup works on Indic language AI, government tech, or sovereign AI infrastructure, these projects create both capacity and potential grant/partnership opportunities.
If you are an investor tracking Indian AI infrastructure
The key question is not whether India will build 12 GW by 2030 — the demand is real. The question is who fills the gap between gigawatt-scale headline projects and the actual operational capacity delivered. The track record is sobering: India delivered ~450 MW in 2026 against a multi-GW pipeline. Execution risk — power procurement delays, transformer lead times (24 months globally), and skilled labor shortages — is the real constraint, not capital.
Will the Odisha Government Approve It?
The Odisha government, led by Chief Minister Mohan Charan Majhi, has been actively courting technology investment. The state's Electronics and IT Minister Mukesh Mahaling signed the HCLTech MoU alongside the Adani ₹800 crore Info Valley project earlier in 2026. The state has a CM-headed clearance authority that evaluates large investment proposals on land, power, water, and connectivity.
The ₹1 trillion scale — roughly equivalent to Odisha's entire annual budget — means the clearance process will be thorough. Adani is reportedly seeking "considerable incentives" from the state, which likely include power tariff concessions, land at preferential rates, and tax breaks. The political optics are favorable: Adani is already a major investor in Odisha (ports, power, cement), and the state government has positioned itself as business-friendly.
But the 250-acre land allocation for a single private project in a state where land acquisition is politically sensitive will face scrutiny. The Odisha government's response will likely come in phases — an in-principle approval followed by detailed site assessment.
How Does This Compare to China and the US?
India's 12 GW by 2030 target sounds ambitious until you compare it to China, which already has over 30 GW of operational data center capacity and is building gigawatt-scale AI compute zones in Guizhou, Inner Mongolia, and Ningxia. The US has over 20 GW operational, with hyperscalers adding 3-5 GW annually.
The difference is not capacity — it is concentration. In the US, four companies (Amazon, Microsoft, Google, Meta) account for over 60% of hyperscale capacity. In China, the state dictates where capacity goes. India is building a third model: private conglomerates (Adani, Reliance) partner with hyperscalers (Google, Microsoft) while the government sets data localization rules that create domestic demand. The question is whether this model can scale without the execution bottlenecks that have historically plagued Indian infrastructure projects.
FAQ
What exactly has Adani proposed for Odisha?
Adani Group has submitted a proposal to invest ₹1 trillion (~$12 billion) to build a 1 GW AI-focused data center in Odisha, requesting approximately 250 acres of land from the state government. The proposal is under evaluation as of August 2026 and has not been approved. (Source: Business Standard, August 3, 2026)
Is this the same as Adani's existing ₹800 crore data center in Odisha?
No. Adani already has a separate ₹800 crore data center project at Info Valley, Bhubaneswar, announced on April 8, 2026. The ₹1 trillion proposal is a new, much larger project — roughly 125 times the investment of the existing one. (Source: OneIndia, EconomicTimes Manufacturing, April 8, 2026)
How does this compare to Google's data center in Andhra Pradesh?
Google, in partnership with AdaniConneX and Airtel, is building a $15 billion, 1 GW AI data center campus in Visakhapatnam, Andhra Pradesh — India's largest planned AI data center. Adani's Odisha proposal is the same scale (1 GW) but at a lower investment ($12B vs. $15B), and it is under evaluation rather than under construction. (Source: blog.google; BlackRidge Research)
Why is the investment amount different — ₹1 trillion vs. $100 billion?
These are different announcements. The ₹1 trillion (~$12 billion) is the specific Odisha proposal. The $100 billion is Adani Group's total commitment to AI data centers across all of India by 2035, announced in February 2026. The Odisha project would be one component of the larger $100 billion program. (Sources: Business Standard; Reuters, February 17, 2026)
What is AdaniConneX and is the Odisha project part of it?
AdaniConneX is a 50:50 joint venture between Adani Enterprises and EdgeConneX (a US-based global data center operator), formed in February 2021. It operates Adani's existing data centers in Chennai, with additional sites in Mumbai, Pune, Hyderabad, Noida, and Visakhapatnam. It has not been confirmed whether the Odisha project would be developed under the AdaniConneX JV or as a standalone Adani Enterprises initiative. (Source: AdaniConneX press release, February 2021; Adani.com)
What is India's total data center capacity and how much is for AI?
India had approximately 2.2 GW of data center capacity in 2025, projected to reach 12 GW by 2030 (~40% CAGR), according to Wood Mackenzie. AI-dedicated capacity is expected to grow from 275 MW to 6,546 MW — a 24-fold increase. Cushman & Wakefield reports 1.6 GW operational as of mid-2026 with 3.1 GW under construction or planned. (Sources: Wood Mackenzie, July 2026; Cushman & Wakefield Global Data Center Market Comparison 2026)
When will the Adani Odisha data center be operational?
There is no confirmed timeline. The proposal is still under government evaluation. Based on typical Indian data center project timelines — land allocation, environmental clearance, construction — a 1 GW campus would take 3-5 years from approval to first phase operational, suggesting 2029-2031 at the earliest for any live capacity.
Disclosure
This article was researched and drafted with AI assistance (transcript extraction, source verification, and drafting support) and reviewed for accuracy by the author, Sham. All investment figures, capacity projections, and project statuses were cross-checked against at least two primary sources before publication. No figures were sourced from the originating YouTube video without independent verification. The author has no financial interest in any company discussed.
Discussion
0 comments