What would it take for the UK to become a global leader in semiconductor technology? - Engineering Discussions - IET EngX - IET EngX

What would it take for the UK to become a global leader in semiconductor technology?

Semiconductors have come up in the news again recently and seem to be a big focus for govt at the moment. It seems like something the IET should have a policy interest in, so we are keen to find out more.

 As a quick summary, the UK government has launched a semiconductor advisory panel and strategy having identified semiconductors as one of the top five technologies of tomorrow, and the strategy has three main strands:

  • Grow the domestic sector
  • Mitigate the risk of supply chain disruptions
  • Protect our national security

 I'm keen to hear your views on the challenges that  need to be solved in order to be world leading in this sector to help inform a policy position and identify areas to explore in more depth.

 Jayne

IET Lead Policy Officer

Parents
  • As a general point I do have a concern about any question including "UK to become a global leader". Could this be rephrased to "UK to become a global player" (or partner or similar word).  "Global leader" is a good tub-thumping political phrase, but in the real world of engineering (and semiconductor technology is a particularly fine example of this) we are in a global industry and the most effective (and most realistic) thing to do for everyone's benefit is to be a key partner. Beware of hubris! (Plus the IET is striving to be a multi-national organisation, should it really be promoting UK self-interest?)

    So more specifically to the question but springing off that point, what I would like to see the UK Government doing to support the UK semiconductor research, development and manufacturing is to press for better engagement with overseas partners, specifically striving to retain engagement with the EU academic partnership programmes and build relationships with other similar programmes.

    And leading on from that, how much IP it is reasonable to expect to be "UK based" I think is a challenging question - and it always has been right back to the days of e.g. INMOS when such questions were first seriously raised. I've been one step removed from the semiconductor industry for the last few years, so I may be wrong (and as ever happy to be corrected), but my impression is that it is still heavily dominated by the phones / gaming / entertainment / volume computing market, requiring huge volumes (and rapid development) to be a serious player. So again, I'd suggest the most sensible question for any one nation to consider is where they can fit into that behemoth. Often it's pure luck that a group of people happen to form in a particular country with expertise in a particular area of value to the wider industry, which is hard (read: darn near impossible) to predict or deliberately nurture, what does matter is allowing that to grow by providing support without excessive bureaucracy.  

    I know none of that helps with the last two bullets "Mitigate the risk of supply chain disruptions" and "Protect our national security", but again I suspect a multitude of international links and hence diversity is probably the best bet?

    Thanks,

    Andy

  • Thanks for this, Andy! I appreciate you challenging the question - you are right, that as a multinational organisation, global partnership and collaboration is important to the IET. Just wondered if you would consider the UK to be a a global player already, or not quite there yet?

    On partnership research, would specific funding be needed, for example a research call on semiconductors that stipulates/encourages international collaboration, or do you think ensuring support for existing programmes would address it sufficiently? 

    Jayne

  • We are already playing - it is just we are in the shallow end, and doing some very cunning things at the edge of the possible, surprisingly cheaply.

    It is not reasonable, and barring some catastrophe, probably  never will be reasonable to compete in the mass production sub $1 per chip car electronics and musical birthday cards and heated socks end of the market, as to do so would require effort on the scale of TMSC.

    (By the way for whom spending $40 billion on a couple of stateside fabs is a small change compared to their Tiawan efforts.- Those fabs are set to produce ~ (merely!) 600k wafers a year - their home effort is more like a couple of million wafers with several thousand chips per wafer,  per month. )

    So rather than chasing work we will always be outbid on, we need to be encouraging and training the next wave of cunning makers of expensive devices.

    The place to invest is to fund students and probably also older engineers wishing to stay near the cutting edge, time on the tools and spaces on the shared wafers to play and to understand what they can and should be doing to make the next wave of more integrated clever stuff, at the 'sure it is a £500 chip the size of a thumbnail, but it will replace a £1000 PCB the size of a playing card'  level.
    Mike.

    PS for those wondering about those shared wafer MPW costs.

    https://europractice-ic.com/schedules-prices-2023/

    shows that costs on a shared wafer on say Xfab varies by exact process a bit but is a couple of K euros per square mm, for a drop of 50 identical devices based on a minimum device area of 10mm2 (say 3mm by 3mm and a bit of a scribing border) So $20k gets you 50 chips...

    There is probably some scope for economy of scale or smaller minimum quantities with a central purchaser and aggregator.

    M.

Reply
  • We are already playing - it is just we are in the shallow end, and doing some very cunning things at the edge of the possible, surprisingly cheaply.

    It is not reasonable, and barring some catastrophe, probably  never will be reasonable to compete in the mass production sub $1 per chip car electronics and musical birthday cards and heated socks end of the market, as to do so would require effort on the scale of TMSC.

    (By the way for whom spending $40 billion on a couple of stateside fabs is a small change compared to their Tiawan efforts.- Those fabs are set to produce ~ (merely!) 600k wafers a year - their home effort is more like a couple of million wafers with several thousand chips per wafer,  per month. )

    So rather than chasing work we will always be outbid on, we need to be encouraging and training the next wave of cunning makers of expensive devices.

    The place to invest is to fund students and probably also older engineers wishing to stay near the cutting edge, time on the tools and spaces on the shared wafers to play and to understand what they can and should be doing to make the next wave of more integrated clever stuff, at the 'sure it is a £500 chip the size of a thumbnail, but it will replace a £1000 PCB the size of a playing card'  level.
    Mike.

    PS for those wondering about those shared wafer MPW costs.

    https://europractice-ic.com/schedules-prices-2023/

    shows that costs on a shared wafer on say Xfab varies by exact process a bit but is a couple of K euros per square mm, for a drop of 50 identical devices based on a minimum device area of 10mm2 (say 3mm by 3mm and a bit of a scribing border) So $20k gets you 50 chips...

    There is probably some scope for economy of scale or smaller minimum quantities with a central purchaser and aggregator.

    M.

Children
  • ...so investing in education and collaboration (particularly supporting research) will help the UK stay agile and competitive in this space - working smarter not harder! Thanks everyone for all the useful feedback on this discussion. I am on leave after today but please do continue to discuss further, someone from the policy team will be monitoring the chat so we can gather any feedback. 

    Jayne