Published: October 16, 2026 • Category: Indian Tech • AI Policy and Infrastructure • By Mahesh
NEW DELHI: Union Minister Ashwini Vaishnaw said on October 8 that India will publish a consultation paper on AI regulation within one month. The Ministry of Electronics and Information Technology has opened a request for empanelment seeking up to 10,000 AI compute units under the IndiaAI Mission.[6][5] The Centre for Development of Advanced Computing, the implementing agency, issued the request on October 7. The minister said the first tender will cover 5,000 GPUs.[5][6] The minister spoke at the India launch of the World Bank World Development Report 2026. He said safety is the most important point for the proposed rules.[6][7][8]
Two Announcements, One Strategy
| Track | What Was Announced | Status |
|---|---|---|
| AI regulation | Consultation paper on AI safety and governance | Due within one month of Oct 8, 2026 |
| AI compute | C-DAC request for empanelment for up to 10,000 AI compute units | Issued Oct 7, 2026; first tender for 5,000 GPUs |
| Capacity | National total to rise above 50,000 GPUs | 38,000+ onboarded in three earlier rounds |
| Indigenous chips | Roadmap for Indian-made inference GPUs | Targeted in three to four years |
Source: MediaNama and The Defense News reports on the minister's remarks and the C-DAC request.[5][6][7]
Medianama reported that the minister said the time for AI regulation has come.[6] He also said the government wants to use AI itself and may rely on empanelment rather than traditional tenders, after which selected firms and startups could bid for work from any department.[6]
The AI Regulation Consultation Paper
The minister outlined five themes for the paper: safety, a human-first approach, scaling, inclusion so that no person is left behind and protection from harms such as deepfakes.[6] He said AI cannot be regulated by a law alone and called for a techno-legal approach, with industry taking the bulk of responsibility on cybersecurity, safety and user harm.[6] MediaNama noted that it is unclear whether the consultation will be open to the public or limited to selected stakeholders.[6]
| Date | MeitY and Government Position on AI Regulation |
|---|---|
| April 2023 | Government told Parliament it was not considering a law to regulate AI[6] |
| November and December 2025 | MeitY Secretary S Krishnan said the aim was to avoid a new law and use existing frameworks such as the DPDP Act[10] |
| April 2026 | AI Governance and Economic Group created as a central coordination mechanism, without independent regulators[13] |
| June 2026 | Vaishnaw publicly said a new AI law is needed[11] |
| July 2026 | Krishnan said stakeholder discussions and drafting would begin[12] |
| October 8, 2026 | Consultation paper on AI regulation promised within a month[6] |
By Depth Grid's count, 30 days from October 8 falls around November 7, so the paper could appear in the first week of November. No date has been fixed. The sources reviewed do not describe whether the paper would propose a standalone law or amendments to existing rules.
The 10,000 GPU Empanelment
C-DAC defines an AI compute unit as a GPU, TPU, accelerator or similar hardware used for AI workloads. It may scale capacity beyond 10,000 units during the empanelment period depending on demand.[5] The request lists model training, fine-tuning, inference and high-performance computing as intended uses. It names ministries, academia, research institutions, startups and industry as users.[5]
| Parameter | Requirement in the Request for Empanelment |
|---|---|
| Capacity sought | Up to 10,000 AI compute units, scalable beyond 10,000 |
| Cluster sizes | 4,000, 1,000 or 512 units (Stage 2 may specify others) |
| Minimum cluster | At least 512 units in a single Indian data centre |
| Geography | Capacity spread across multiple locations; cities not specified |
| Data residency | All infrastructure, data, metadata, logs, backups and disaster recovery within India |
| Isolation | C-DAC data separated physically and logically from other customers |
The request does not specify how much of the 10,000 units will be bought in the first round, although the minister gave a figure of 5,000 for the first tender.[5] A report says the additional capacity would take the national total above 50,000 GPUs after more than 38,000 were onboarded in three rounds involving Jio Platforms, Tata Communications and Yotta Data Services.[7] On Depth Grid's calculation, 10,000 units would add about 26% to the 38,000 already onboarded.
The same report says 3,000 of the new units would be reserved for government departments and public-interest work, which is 30% of the total. The other 7,000 would go to companies developing AI projects.[7] The minister also pointed to a roadmap for Indian-made GPUs for AI inference within three to four years.[7]
The Delivery Question
MediaNama reported in April that only Rs 400 crore of the IndiaAI Mission's outlay of more than Rs 10,000 crore over five years had been released.[9] Depth Grid has not seen a more recent figure, so the pace of spending is a point to track as the new empanelment progresses. The gap between announced capacity and funds released is the most useful test of whether the new round moves faster than earlier ones.
Strategic Implications for Startups, Enterprises and Policy
For startups and researchers, more subsidised or shared compute lowers one of the largest costs of training and running models. The data residency rules favour providers with Indian data centres. For companies building on AI, the consultation paper will signal which obligations may follow, with safety, cybersecurity and deepfakes already named. Depth Grid's guide to the AI native company covers how businesses are restructuring around AI. Enterprise software moves such as SAP's purchase of TechWolf show where demand is heading.
On the hardware side, the stated goal of Indian inference GPUs links compute policy to chip policy. Investor interest in Indian AI hardware is visible in Volantis' 88 million dollar raise. Our report on Q3 global venture funding gives wider context on how much private capital is flowing into AI.
Four items to watch are the release of the consultation paper, whether consultation is public, the size of the first-round award and how much of the Rs 10,000 crore-plus outlay is released.
Common Questions and Direct Answers
What did Ashwini Vaishnaw announce about AI regulation?
Union IT Minister Ashwini Vaishnaw said on October 8, 2026 that the time for AI regulation has come and asked MeitY to publish a consultation paper within one month. The paper will focus on safety, a human-first approach, scaling, inclusion and harms such as deepfakes. He added that industry must take primary responsibility for cybersecurity, safety and user harm.
When will India's AI consultation paper be released?
The minister said it would come within one month of October 8, which points to early November 2026. No exact date has been announced. Medianama noted that it is unclear whether the consultation will be open to the public or limited to selected stakeholders, so check MeitY's website for the release.
How many GPUs is India adding under the IndiaAI Mission?
The government is seeking up to 10,000 AI compute units through a C-DAC request for empanelment. Reports say this would take national capacity above 50,000 GPUs, after more than 38,000 were onboarded in three earlier rounds. Vaishnaw said the first tender is for 5,000, but the request does not specify the first-round quantity.
What conditions does the GPU empanelment set for providers?
Providers must keep all infrastructure, data, metadata, logs, backups and disaster recovery inside India and isolate C-DAC data from other customers. Capacity is organised in clusters of 4,000, 1,000 or 512 units, with each cluster holding at least 512 units in a single Indian data centre. Capacity must be spread across multiple locations.
Who can use the IndiaAI compute capacity?
The request names ministries, departments, academia, research institutions, startups and industry as users. A report says 3,000 of the new 10,000 GPUs would be reserved for government departments and public-interest work, with 7,000 for companies building AI projects. Permitted uses include model training, fine-tuning, inference and high-performance computing.
Primary Sources and Regulatory Filings
- Ministry of Electronics and Information Technology: official notices and consultation papers
- IndiaAI Mission: compute portal and programme updates
- Centre for Development of Advanced Computing: empanelment documents
- Press Information Bureau: government statements
- MediaNama: MeitY invites AI compute providers to supply 10,000 GPUs
- MediaNama: MeitY to release AI regulation consultation paper in a month
- The Defense News: AI regulation paper, 10,000 GPUs and capacity above 50,000
- Venture Magazine: AI regulation paper in 30 days and the 10,000 GPU plan
- MediaNama: IndiaAI Mission outlay released as of April 2026
- MediaNama: MeitY's December 2025 position on a standalone AI law
- MediaNama: Vaishnaw calls for a new AI law, June 2026
- MediaNama: MeitY to start stakeholder discussions on a separate AI law, July 2026
- MediaNama: AIGEG, MeitY's AI governance body, April 2026
Related Intelligence
By Mahesh | Depth Grid News Desk | depthgrid.in
Discussion (0)
Comments are disabled for this post.