IndiaAI Mission has officially slashed the benchmark hourly rates for NVIDIA’s latest B200 (Blackwell) GPUs by 10% in its fourth round of compute tenders, as reported on Tuesday, April 28, 2026.
This “price correction” is the result of aggressive bidding by domestic data center providers looking to secure government contracts. The move is designed to make high-end AI compute more accessible for Indian startups, research labs, and government agencies.
1. The New Price Benchmarks
The latest Letter of Intent (LoI) issued on April 24, 2026, reveals a uniform drop in the lowest (L1) discovered rates for the most advanced chips in the market.
| GPU Configuration | Old Hourly Rate (R3) | New Hourly Rate (R4) | Change |
| NVIDIA B200 (Single) | ₹323 | ₹290.70 | ▼ 10% |
| NVIDIA B200 (8-Unit Node) | ₹2,584 | ₹2,325.60 | ▼ 10% |
| NVIDIA B300 (Single) | New Listing | ₹351.00 | — |
- Utility Pricing: At ₹290.70 per hour (~$3.1), IndiaAI’s B200 rates are nearly 20–40% lower than global market standards for Blackwell-class hardware.
- Zero Fees: The tender mandates that network ingress and egress fees must be set at zero, effectively making AI compute function as a standard utility like electricity.
- Long-term Discounts: Under a 12-month reservation, the hourly rate for a single B200 unit drops even further to ₹251.10.
2. Industry Concerns: “Can it Last?”
While the price cut is a victory for startups, industry leaders like Anuj Bairathi (CEO of Cyfuture India) have warned that these rates may be unsustainable in the long run.
- Weakening Rupee: With the INR trading near ₹94/$1, the cost of importing NVIDIA hardware is rising.
- Component Surges: A global “parabolic” spike in DRAM and NAND storage costs in Q1 2026 is putting immense pressure on data center margins.
- The “H100” Comparison: Interestingly, the new B200 rate (₹290.70) is now dangerously close to the market rate for the older NVIDIA H100 (~₹249), which could lead to a rapid shift in demand that outstrips current supply.
3. Expanding the National Compute Pool
This price cut coincides with a massive expansion of India’s sovereign AI infrastructure overseen by MeitY.
- Current Capacity: India’s national AI compute pool recently reached 62,000 GPUs in March 2026.
- Target Goal: The IndiaAI Mission is aiming for a total of 200,000 GPUs by the end of 2027 through public-private partnerships.
- Empaneled Companies: The nine companies that cleared the technical round and must match these L1 rates include Tata Communications, Yotta Data Services, E2E Networks, Netmagic, and Sify Digital Services.