Fabs, Fabless and Indian Semiconductor Industry

Most of the semiconductor professionals and enthusiasts in India are keenly following the policy, commentary and outcomes from the host of Indian govt decisions taken under the head of SPECS and other incentive-based measures.

While every actor – government, investors, entrepreneurs and engineers- wants to see India emerge as a vibrant location in all aspects of production and value addition in the semiconductors ecosystem - they also bring their own unique insights based on their own domain experience and knowledge. Such an environment is naturally expected to breed some disagreements requiring careful moderation and application of mind. We will try to address one such disagreement voiced in opposition to govt push for domestic semiconductor fabrication units from the Indian fabless start-up ecosystem in this article.

A fabless semiconductor company translates the abstract system requirements coming from either the end customer or a standard into a very large number (in billions) of circuits. Those circuits are verified using sophisticated software tools for required performance on a chosen process supported by a fabrication company. Finalized design is then sent to the fabrication company as an image – which is etched on the Silicon substrate at the fab. The final Silicon (now called a chip) is then sent back the fabless semiconductor company for characterization and commercialization.

During the last 2 decades India has seen an exponential growth in semiconductor design activity, but zero presence in the semiconductor fabrication space. There have been at least two failed attempts at setting up a fab in the past as well.

One must be deeply aware of this overarching reality and context to appreciate the steps taken by the Indian Govt to push for domestic semiconductor manufacturing. Consequences of missing out completely from this crucial space has very obvious strategic and security implications as well.

While most of the mainstream media has been focused on e-commerce and edutech start-ups for the last decade, India has seen some encouraging activity in the space of fabless semiconductor start-ups in the same timeframe as well.

Companies like Cosmic Circuits (acquired by Cadence), Saankhya Labs, Aura Semiconductor, Steradian Semiconductor, and many others have built successful start-ups based on offering design services and building high-value IPs at times complete semiconductor products.

A couple of comments by Parag Naik, CEO of Saankhya labs have caught the attention of many readers, and also require a response.

Speaking to Ken, he said:

Ken summarized his views with the following recommendation.

Speaking to Indian Express, he said the following:

The motivation behind including the quotes by an individual CEO is not to single out the person, but to substantively engage with the thoughts and assumptions embedded in them. Comparing semiconductor production to a mineral like crude oil is a non-starter. No amount of policy or human ingenuity can extract oil from a place that isn’t already blessed with it. Semiconductor fabs exist in Japan, China, Taiwan, Korea, Europe and USA. They exist in wet regions of heavy rainfall like Taiwan to dry regions of Arizona. The only common thread between crude oil and semiconductors is that of trade imbalance, energy and national security; and unlike crude oil, India isn’t destined by geography to live without domestic production of semiconductors!

While the authors completely understand the frustrations of fabless semiconductor companies in getting predictable or preferential treatment by fabs, it is simply not reasonable to expect that the Indian govt can battle global semiconductor and device majors like Apple, Nvidia, Qualcomm, Intel, Samsung and Broadcom to get preferential treatment for Indian fabless start-ups.

Every major design house, device and auto manufacturer is reeling under shortage of chips for more than a year now. If anything, Indian govt can be expected to have more leverage on fabs if and when they are setup on Indian soil. It is an interesting thought to get the Indian govt to subsidize the wafer costs of Indian fabless companies. This is something to be examined on its own merits and govts strategic goals. There is no merit in presenting this as a trade-off against govt incentivising domestic semiconductor fabs.

Symbiotic relationship between fabs and fabless industries can be seen from the blossoming of companies like MediaTek and design hubs like Hsinchu in Taiwan. Mr Naik’s comment noting the advantages available to fabless companies in Korea (due to proximity with fabs) underscores this symbiotic relationship and only makes the case for fabs in India stronger.

Industry has also seen fabs like TSMC cannibalizing the share of design pie by offering IP and design services to their clients. One can also easily see the great geographic divergence between locations of semiconductor design (US, EU and India) and semiconductor production (Taiwan and Korea). It is certainly not a truism to suggest that semiconductor fabs will simply follow the fabless companies.

Authors are deeply aware of lack of availability of risk capital to Indian semiconductor start-ups – especially when compared to their e-commerce peers. There is an encouraging trend in the last few years with a few venture capitalists and PE funds taking interest in the deep tech spaces like semiconductors. For example, one can see 2 semiconductor companies in the portfolio of Endiya here. Last few months have shown that Indian players are indeed focussed on moving up the value chain and are forced to look into areas of 5G, telecom equipment, EV and green energy supply chain etc.

Authors also hope that some of the post-IPO exits from e-commerce companies will free up some more risk capital for fabless ventures in India. However, unlike fabless companies starting a semiconductor fab requires an order of magnitude more capital and risk-taking appetite – and the Indian government must renew their efforts to see this through.

(The author has more than a decade experience spanning across many global semiconductor majors and wishes to remain anonymous)