We need product cos, not just services delivery resources: Jaswinder Ahuja

Business Standard

15-05-2025

As a partner in India's semiconductor mission, California-based Cadence Design Systems has played an important role in the country's tryst with semiconductors. Its managing director for India, and corporate vice-president, International Headquarters, Jaswinder Ahuja, who has been at the helm here since 1996, now has a bird's eye view. He has seen India's transition. In a video interview with Surajeet Das Gupta, Ahuja sheds light on the challenges and opportunities in India in this space. Edited excerpts:
While the manufacturing side of semiconductor has taken off, the design-led incentive scheme (DLI) faces many challenges. What is your assessment on the progress?
We have seen a great beginning in the manufacturing side, as the new plants being built are pulling in an entire ecosystem to support them, from equipment to chemicals and gases etc. In DLI, it's a chicken and egg situation, and its progress is not in line with the expectations. There are companies with intent to work in this space but they find the incentives inadequate or there are strings attached to it.
As you talk to many of the prospective companies, are there any other challenges which need to be tackled under the DLI programme?
For every project, the government — under DLI — is offering ₹15 crore. It is too little, almost inconsequential. If you look at the total project cost like manpower, tools and cost of buying some IP and building a prototype, when you add up, it is too little. For a non-complex design, the subsidy of 50 per cent should be around ₹100 crore.
Are there any other issues?
Yes. For instance, under the scheme you get the reimbursement only after spending, and considering the delays they create working capital challenges. Second, there are questions as to why it should be 50 per cent and not more? And many bigger players like L&T say they should also be eligible and not just smaller companies. There is also a demand that MNCs should be allowed to participate, but I think that should be a red line.
So what is the feedback from the government?
My understanding is that the government is analysing all the feedback and plans to use them to work on a revamped DLI 2.0, it is expected to address some of these concerns. But the real challenge is where are those companies and entrepreneurs who have compelling business ideas, that is what we have to watch once the DLI scheme is updated. We don't want to be only a services delivery resource like the GCCs. For our aspirations in the semiconductor space we need to have product companies.
Why have PE funds stayed away from funding design-led companies, but have a bet on AI?
PE funds see a clear line of sight in AI application companies on generating revenues and profitability in the next 12-24 months. If you invest in a semiconductor design it takes five years to get a product out, seven years to generate money and 10 years for Ebitda. But most PE funds exit in 7-8 years.
While we talk of the great chip design talent pool available in India for the world, we don't see too many entrepreneurs or startups in the space. What is the disconnect?
We do have amazing design talent, but that design talent is focused on execution. We do not have product business expertise in India, who understand the market, identify an opportunity, define a product and build a business case around it. In MNCs, it is undertaken by a separate team in the global headquarters. Once this is done, then a team builds the architecture that is again done in the headquarters.
The government wants to design and build a 'made in India' GPU. Do you see it to be feasible?
The US has placed certain restrictions on GPU exports which are designed mostly by Nvidia and AMD. So there is a possibility of a supply demand mismatch, especially when GPUs are key for AI and machine learning compute. Developing GPUs in India… if we start from now, producing it in volumes will take five to seven years. Can India do it, yes as like others, it does not have to start from the scratch. For the time being, there is a larger opportunity for startups in AI, in segments like automobiles and others. This compute architecture will solve the problems but at a different price point and performance.