Financial support for GPU infra will increase AI adoption, say analysts

The government proposal of financing up to 50 per cent of the cost of setting up artificial intelligence (AI) compute infrastructure is a welcome move, said industry players and analysts.

Many feel that it will bring down graphic processing unit (GPU) usage costs in the country and will speed up the adoption of AI.

“The Ministry of Electronics and Information Technology’s (MeitY’s) plan to finance AI compute infrastructure development is a positive step, demonstrating commitment to AI technology and incentivising industry investment in GPU capacity,” said Amrish Pipada, Founder & CEO, Mega Networks, an AI server manufacturing company.

“The initiative is likely to attract more investment, leading to advanced home grown infrastructure and furthering the capabilities in the AI sector,” Pipada added.

Recently, at an event organised by CII, MeitY Secretary S Krishnan said that the government was planning to finance up to 50 per cent of the cost of creating AI compute infrastructure, and was willing to partner with private players to develop high-end compute capacity within the country under India AI mission. 

While the industry welcomed the announcement, it also highlighted some other areas that need government intervention, including skilling initiatives and regulatory support.

“Some areas where GPU server players may look for support include training programs for upskilling and reskilling, funding support for research initiatives, ensuring a conducive regulatory environment and building a strong network infrastructure to facilitate the deployment and utilisation of GPU servers effectively,” said Pipada.

Rohit Kumar of the Quantum Hub (TQH), a public policy firm, suggested that it was crucial to conduct a systematic study to identify the binding constraints that may otherwise hinder further investments.

“Many of these constraints including the need for skilled manpower, enhanced R&D capabilities etc., are already recognised as focus areas under the India AI mission but will require prioritisation,” he added.

Analysts also said that the fiscal support will encourage local AI initiatives, especially in the government sector as well as India’s huge MSME sector.

“It is a good initiative, as AI infrastructure is prohibitively costly, with a small cluster costing above a million dollars in capex. This is also in line with the interest shown by the government for sovereign data centres as part of its strategic initiatives,” said Naresh Singh, Senior Director Analyst, Gartner.

Secretary Krishnan, during the CII event, had also said that in order to expand GPU capacity usage, the government will look at providing the viability gap funding (VGF) to support the creation of new GPU capacity, or alternatively, a voucher-based mechanism where vouchers are issued to people for using GPU capacity.

Industry players believe that availability of AI servers from the government at subsidised rates could make it harder for server companies to attract clients at regular prices, but at the same time they also see potential opportunities.

“Server companies can differentiate through value-added services, specialised solutions, and customised offerings. The increased AI adoption will likely boost demand for GPU server solutions, allowing companies to secure clients and grow,” said Pipada of Mega Networks.

Vishnu Vardhan, Co-Founder and CEO, SML India -- the firm that has recently launched homegrown AI chatbot Hanooman -- said that the move to subsidise use of GPU servers will empower a lot of companies to build use cases and applications for Gen AI and build the marker for GPU cloud providers and hence, it is good for the industry in the longer terms.