Listen to this article in summarized format
Loading...
××
Subscribe to
Unlock AI Briefing and Premium Content
New Year Offer 24 Hours Left
Subscribe NowAlready a member? Sign In
What's Included
- Exclusive Stories
- Daily ePaper Access
- Smart Market Tools
- Curated Investment Ideas
- Ad-lite Experience
Subscription
Prices of Nvidia’s latest B200 GPUs have dropped 10% in the latest round of IndiaAI Mission tender, reflecting aggressive bidding by some compute service providers.
Executives of some qualified bidders, however, flagged concerns over the sustainability of such pricing.
The letter of intent (LoI) issued for empanelment under the fourth round of GPU procurement tender revealed a sharp correction in lowest (L1) bidding prices of select high-end chips. Hourly rates for B200 have fallen a uniform 10% to Rs 290.70, from Rs 323 in the third round, for a single unit and to Rs 2,325.60 from Rs 2,584 for eight-unit setups.
Newer B300 configurations are priced between Rs 351 and Rs 2,808 per hour.
“Aggressive L1 pricing under the IndiaAI Mission raises viability concerns, as rising global NAND/DRAM costs and a weakening rupee squeeze margins, making long-term contracted demand essential for profitability,” said Anuj Bairathi, chief executive of Cyfuture India, one of the nine companies that cleared the technical round of the tender.
Narendra Sen, chief executive of RackBank Datacenters, another technically qualified bidder, said the latest benchmarks remain highly competitive globally.
“IndiaAI’s latest L1 pricing is still extremely aggressive compared to current global pricing benchmarks. For Blackwell GPUs like B200 and B300, IndiaAI pricing is coming in nearly 20-40% lower than global market standards,” he said.
“With inference demand rising rapidly, especially due to code models and open-source AI deployments, Blackwell demand continues to stay very strong worldwide,” he added.
The LoI, dated April 24, pertains to the ongoing continuous request for empanelment (RFE) and asks selected bidders to accept empanelment at the lowest (L1) discovered rates for AI compute units, along with network, storage, and parallel file system services.
ET had reported on April 9 that according to Government e-Marketplace (GeM) tender data, nine companies have qualified in the technical round. They include Paradigmit Technology Services, Tata Communications, Netmagic IT Services, E2E Networks, Yotta Data Services, Sify Digital Services and UrsaCompute.
As part of the conditions, companies must match L1 rates across categories where they are not the lowest bidder and ensure availability of at least 1,000 AI compute units within six months of the LoI issuance, with responses due by May 5, next Tuesday.
The annexure to the LoI outlines detailed pricing across GPU categories, including Nvidia B200, B300, H100, H200, AMD MI300X, and Google’s TPUv6e chips. Pricing has also been standardised across reserved usage periods: one month, six months, and twelve months, with lower rates offered for longer commitments, reflecting an effort to encourage sustained usage. For instance, a single Nvidia B200 unit drops to Rs 251.10 per hour under a 12-month reservation plan.
Beyond compute, the government has fixed storage and networking benchmarks. High-speed block and file storage is priced at Rs 1.1 per GB per month, while object storage is pegged at Rs 0.78 per GB. Parallel file system services are set at Rs 4.8 per GB, and internal NVMe/SSD storage at Rs 1.10 per GB per month. Network ingress and egress charges have been set at zero, while MPLS services are priced at Rs 16,795.15 per month.
The empanelment forms a key part of India’s broader push to build sovereign AI infrastructure by aggregating compute capacity from multiple providers and offering it at standardised rates to startups, enterprises, and government agencies.
Industry executives said the pricing strategy reflects a deliberate policy push.
“This clearly shows IndiaAI is pursuing a serious sovereign AI infrastructure strategy by making advanced compute significantly more affordable for the domestic market,” Sen said.
Executives of some qualified bidders, however, flagged concerns over the sustainability of such pricing.
The letter of intent (LoI) issued for empanelment under the fourth round of GPU procurement tender revealed a sharp correction in lowest (L1) bidding prices of select high-end chips. Hourly rates for B200 have fallen a uniform 10% to Rs 290.70, from Rs 323 in the third round, for a single unit and to Rs 2,325.60 from Rs 2,584 for eight-unit setups.
Newer B300 configurations are priced between Rs 351 and Rs 2,808 per hour.
“Aggressive L1 pricing under the IndiaAI Mission raises viability concerns, as rising global NAND/DRAM costs and a weakening rupee squeeze margins, making long-term contracted demand essential for profitability,” said Anuj Bairathi, chief executive of Cyfuture India, one of the nine companies that cleared the technical round of the tender.
Narendra Sen, chief executive of RackBank Datacenters, another technically qualified bidder, said the latest benchmarks remain highly competitive globally.
“IndiaAI’s latest L1 pricing is still extremely aggressive compared to current global pricing benchmarks. For Blackwell GPUs like B200 and B300, IndiaAI pricing is coming in nearly 20-40% lower than global market standards,” he said.
“With inference demand rising rapidly, especially due to code models and open-source AI deployments, Blackwell demand continues to stay very strong worldwide,” he added.
The LoI, dated April 24, pertains to the ongoing continuous request for empanelment (RFE) and asks selected bidders to accept empanelment at the lowest (L1) discovered rates for AI compute units, along with network, storage, and parallel file system services.
ET had reported on April 9 that according to Government e-Marketplace (GeM) tender data, nine companies have qualified in the technical round. They include Paradigmit Technology Services, Tata Communications, Netmagic IT Services, E2E Networks, Yotta Data Services, Sify Digital Services and UrsaCompute.
As part of the conditions, companies must match L1 rates across categories where they are not the lowest bidder and ensure availability of at least 1,000 AI compute units within six months of the LoI issuance, with responses due by May 5, next Tuesday.
The annexure to the LoI outlines detailed pricing across GPU categories, including Nvidia B200, B300, H100, H200, AMD MI300X, and Google’s TPUv6e chips. Pricing has also been standardised across reserved usage periods: one month, six months, and twelve months, with lower rates offered for longer commitments, reflecting an effort to encourage sustained usage. For instance, a single Nvidia B200 unit drops to Rs 251.10 per hour under a 12-month reservation plan.
Beyond compute, the government has fixed storage and networking benchmarks. High-speed block and file storage is priced at Rs 1.1 per GB per month, while object storage is pegged at Rs 0.78 per GB. Parallel file system services are set at Rs 4.8 per GB, and internal NVMe/SSD storage at Rs 1.10 per GB per month. Network ingress and egress charges have been set at zero, while MPLS services are priced at Rs 16,795.15 per month.
The empanelment forms a key part of India’s broader push to build sovereign AI infrastructure by aggregating compute capacity from multiple providers and offering it at standardised rates to startups, enterprises, and government agencies.
Industry executives said the pricing strategy reflects a deliberate policy push.
“This clearly shows IndiaAI is pursuing a serious sovereign AI infrastructure strategy by making advanced compute significantly more affordable for the domestic market,” Sen said.