A processor is not made in a day: you need an instruction set like ARM or x86 (called “ISA” in the jargon), an architecture grouping together different logical blocks (CPU, GPU, i/o, etc.) and production plants. Slowly, but surely, the Indian government intends to acquire at least partial autonomy in all these areas.
Protect industries (and break free from China?)
Before even talking about the programs from a technological point of view, it is necessary to place the Indian initiative in its context. Very specialized in terms of software thanks to its technological hub in Bangalore, India is a dwarf when it comes to semiconductors. This even though the country has serious local industries, particularly in the automotive field – Tata Motors, Mahindra, etc.
Its industries are totally dependent on foreign chips, and partly on China. Blocked by American embargoes, the Middle Kingdom cannot advance as it pleases in terms of advanced engraving (EUV and sub 14 nm lithography), but it remains a very large producer of “classic” chips (DUV, 28 nm and +). In addition to India’s policy of non-alignment exacerbated by the nationalist trajectory of the Modi government, the sub-continent is cold, not to say in latent war, with China. Especially around the course of the Himalayan borders.
Read also: IBM engraves the first 2nm chip in history… and Intel could well benefit from it
India had already taken retaliatory measures by banning Chinese applications on its territory and it is starting to attack its tech giants: this week, the Indian government blocked approximately $ 750 million in financial assets from Xiaomi in India.
India is thus a large country which intends to develop its industries with as little technological dependence as possible. And it all starts with intellectual property independence, betting on an open ISA (instruction set): RISC V.
RISC-V, the obvious choice
India is historically part of the non-aligned countries and like others – such as Russia – it wishes to free itself from all tutelage. Without being propped up on this position since it relies in particular, in the field of defence, on France for its aviation (Rafale) or its submarines (Scorpene class).
Far from going for MIPS or ultra-exotic architectures like the one chosen by Russia with its Elbrus processor, India has chosen with its “RISC-V Digital India RISC-V (DIR-V)” program. Resulting from an American university project, RISC-V is an open instruction game in full swing which was organized as a foundation, which sheltered itself from Swiss legislation during the “Huawei-gate”.
Read also: RISC-V: ARM’s competing architecture passes the 5nm engraving milestone (2021)
Compared to x86 or Power, both under total American control (Intel/AMD and IBM), or even ARM which is under Japanese-British control, RISC-V has the obvious advantage of being able to be used and developed without paying royalties. But also and above all, there is no risk of technological blockage by another country.
In return, RISC-V is much less mature than the aforementioned architectures, both in the instruction set itself and in terms of ecosystems. Ecosystems with an “s” because we are talking here about hardware on the one hand, but also software or development (machines for production and verification of designs, etc.). This could be a problem if India wanted to directly develop cutting-edge chips, such as SoCs for smartphones. But this is not (yet) the case.
Simple chips, proven manufacturing nodes
On the production side and therefore factories, the local development project is called Semicon India Program. No race for 2 nm like Japan and the USA, nor even an attempt to jump straight into 14 nm like China.
Virtually non-existent for the time being in the world of semiconductors, India cannot claim to be Taiwan on day 1. And is already beginning to build a non-existent ecosystem for the time being. This is why the consortium which has already been selected – ISMC, a joint venture between an Abu Dhabi fund and the Israeli Tower which have put together a 3 billion dollar plan – is not talking about EUV machines (which are anyway all reserved for three years and incredibly hard to master), but more classic DUV scanners.
Technologies not advanced enough to make smartphone chips, but which are sufficient for its industrial needs, such as in defense or the automobile, where the target engraving finenesses range from 65 nm to 28 nm.
Read also: Europe: Thierry Breton reveals the ambitions and challenges of his “Chips Act”
However, India still has ambitions and wants to push to engrave as finely as possible. Government aid is also indexed to the fineness of the engraving of the factories that will be built: nodes greater than 45 nm only give rise to 30% of the total cost of setting up the fab, while this rises to 50% for any plant 28nm or less.
But between designing a fab project, setting up a supply chain, recruiting (and training) talent, and building the factory, there is work to do. This is all the more so as the roadmap is ambitious: the first operational commercial designs are expected for December 2023!
Industrial successes, but India is far from China
“Make in India” is a doctrine that aims to ensure the creation or strengthening of local industrial fabrics in order, if not to break free, at least to reduce the sub-continent’s dependence on the rest of the world. And India has some great successes including its space program. Although it is not the most ambitious or the most prestigious in the world, it is one of the most effective: controlled costs, many successes on very well-defined “low cost” projects, etc.
However, when we compare the power and success of China with the difficulties the country has in creating its own sectors in semiconductors – DRAM, EUV engraving, etc. –, we imagine a long way to go for India. Admittedly, the country is not in the crosshairs of the USA, which block, or even regularly attack, Chinese development in the field. But even if India does not face so many external hindrances, the country has its own challenges.
Let’s mention climatic challenges – right now, ground temperatures are the highest ever measured! –, often (very) mediocre infrastructure, very strong linguistic fragmentation, major energy supply concerns, or even the absence of an existing sector.
Also see video:
Also see video:
So it’s no surprise that none of the announcements talk about developing an industry capable of flooding the market like the Americans, Koreans, Taiwanese or Chinese are doing. But it is already a question for India of having a sufficiently developed and solid local industry to be able to guarantee the continuity of its industries in the event of hard blows.
And the country could count on its local software industry in Bangalore, but also on its American “diaspora”: whether at Intel or AMD, very many (excellent) semiconductor engineers are Indian or of Indian origin. It remains for the country to find the right partners and avoid the pitfalls of corruption that have plagued so many semiconductor projects in China.