The UK government? On the right track with its semiconductor strategy? The UK government may be onto something with its strategy to support the domestic semiconductor industry. In a strange twist of fate, some experts are starting to say that its approach makes sense. Britain's long-awaited semiconductor strategy was published in May, promising funding and a focus on areas where the country is …
"We're never going to spend $10 billion on building a state-of-the-art 2nm fab for making chips to go into iPhones"
That sounds a bit defeatist. Intel have just started high volume production on their Intel 4 process node (which they say is equivalent to 4nm) fab in Ireland, having spent €17bn on developing it.
Ireland has spent an enormous amount of time cultivating its relationship with Intel, building strategic partnerships with multiple universities over decades (e.g. Intel funded kit for the electronic engineering department for UCC when I attended a long time ago).
People believe that Ireland's success is purely down to tax breaks, but fail to see what the IDA (Industrial Development Agency) has been helping deliver for decades, in areas like semiconductors, pharmaceuticals (7 of the world's top 10 pharma companies have manufacturing and development facilities in Cork) and now datacentres.
Saying it'll never happen is a bit silly, it takes time, patience and planning to achieve it, but it absolutely can be done.
Just not with the current corrupt lot.
If you ever find a significantly better alternative please let me know. I've voted in at least 9 general elections by now and have always voted for 'the least useless of the bunch' or 'the bunch likely to do the least damage'.
I don't think I've ever had the experience of voting for someone who represents a competent collection of politicians. They are all to some extent taking a short termist view and looking after number one.
We need to accept that we get the governments that we deserve. We as a population are fickle, lazy and divided. A sensible government will never be voted in, because we prefer "charismatic politicians", tax "giveaways", or "powerful leaders" (IOW ones that are hard to get rid of) etc. We actively reward the bad ones.
Not a very optimistic analysis, but these things are ultimately OUR fault. The politicians are merely opportunists.
Not sure I completely agree with that.
Winning elections has become far more about Marketing and mass profiling/targeting/manipulation than actual good policy and governance, and i'm not sure how you can place the blame for that on the people. Sensible politicians cannot win elections anymore.
Most people are not aware just how much they are manipulated by the likes of Facebook, Google and TikTok, and how their sentiments are sold to the highest bidder
I live in a seat that has been labour since the peasants revolt.
There was a conservative candidate once, but somebody ate him
This means I don't have any control over the Tory government - but apparantly am personally responsible for Gulf war 2
Not entirely true. We are technically a democracy, but have little control over what governments actually do.
People can only vote for what is on the ballot paper, and there simply isn't any competence on offer. It's not a talent pool, but a puddle. Most people don't vote because they consider them all to be rubbish. Those who do, tend to vote against those they dislike the most, not for anyone. Nobody bothers with policies because they are lies. Starmer will walk the next election not because of Sunak, but because of the consequences of what Johnson and Truss did, primarily Brexit. I can't recall anyone calling Starmer 'charismatic'.
Everyone likes freebies from government, but money in the UK does not flow to tech startups, because rich people in the UK are rarely bright enough to see the potential. They prefer a quick buck from the financial markets and being friends with the right people, and concentrate on avoiding tax. We are not a wealthy country. Most Britons have small amounts of savings, and the flip away from globalisation back to nationalist tribes, 25% off sterling, high inflation and high interest rates has beggared more of the country. Brexit has been as economically damaging as WWII.
The tech policy isn't as idiotic as the US one. The US can magic up billions from additional debt and replicate Chinese factories. It will then sell everything at higher prices and block the cheap stuff from China. That will just beggar us all more. In the UK, we throw a few quid at early stage and will have to leave the rest to SoftBank and the like, as the Irish are with Intel.
We are no longer an Imperial power. Our monarch is currently touring ex colonies apologising for all that. In a world of large trading blocs, we are isolated. After 10 years of political idiocy, our services are collapsing. we don't have enough staff having sent them home with Covid repatriations and Brexit, the economics no longer work due to high costs and high poverty. It's a multiple whammy. Everything that could have gone wrong, has. Government (either) can only paper over the cracks. We barely even have an EV industry, and that is state funded. Add EU sanctions on imports and US sanctions on Chinese EVs, and a 'fast' transition just won't happen.
There are opportunities in next gen tech - notably distributed software - but governments won't allow it. They are taking down much of our tech ecosystem as it empowered individuals at their expense. The net is morphing into a system of state surveillance, and unlike Internet 1.0 that won't drive decades of global growth. A period of global recession, climate decay and conflict looms with no real end in sight.
The good times don't sound like that. We stuck a knife in our own economy whilst other governments were cheerfully taking down everything around us. Some cash for whatever out there is still showing promise is the best you are likely to see from any British government.
I agree with all of the above, except that I can't see how mandating a switch to EVs will help either the UK economy or the environment (except in terms of particulate air pollution perhaps, but we still burn coal and subsidise Drax).
It seems like another nail in the UK's coffin and another big win for China to me. They are not getting any cheaper except via 'dumping' tactics of selling at a loss into specific markets. The amount of mineral resources needed by EVs (except direct oil consumption obviously) is vastly more than petrol cars, and the only way I can see we will reduce emissions is by reducing consumption i.e. reduced car ownership, and that will inevitably worsen the recession.
I'm not sure what you mean by "next gen distributed software" either. Can you elaborate? Portable offline generative AI, maybe? That's a Pandora's Box for sure.
>We need to accept that we get the governments that we deserve. We as a population are fickle, lazy and divided.
While Ireland is populated by a united (other than in a hatred of the British), wise, thoughtful population of political scientists who over the last 100 years have chosen governments with the aim of fostering an economic and political utopia ?
But much of Eire's success is a combination of low corporate and high personal taxes.
In effect they made s deal with the electorate 40 years ago. --- we will get you jobs but you will pay a s***t load of taxes.
Never going to happen on a country that thinks Liz Truss is leadership material.
Ireland has spent 40 years dragging itself out of the dark ages, looking to the future, investing in infrastructure and making the most of joining the European and World economic stages.
Britain, meanwhile, has spent 40 years strangling and neglecting its infrastructure, selling national assets off cheap to any overseas investor, lining the pockets of a few at the top, removing itself from the European and World economic stages and getting ready to re-enter the dark ages.
We (britain) did have a semiconductor industry in the 60's and 70's (Mullard, Ferranti, etc, plus US companies like TI and HP, IBM) but just gave up sometime in the 80's and let everyone else do it better and cheaper.
Ireland spent 20 years persuading US company execs that it was a nice place with no tax, a guarantee of no regulatory oversight (see el'reg story about the Irish DPA being a single room over a sweetshop) and an accomodating government. It's basically Dubai but with golf courses and access to the Eu market.
It's also the richest country in Europe GDP/capita but somehow at the head of the queue for bailouts when its housing market screws up.
My question is, how much cheaper is it to build a 32nm fab today than a 3nm fab?
If you build a brand new one, you have to be able to charge enough for you chips to get back the construction costs. The 13 year old one has already earned that money back, and just needs to cover the costs of staff, materials, and maintenance. That is at least part of the reason why 32nm chips are cheaper now than they were in 2010.
Sure, you can buy a brand new DUV machine a lot cheaper than an EUV machine, that will certainly help. But that is just one part of the overall cost.
No it doesn't. There's no such thing as "200nm chips". That's a type/misunderstanding of "200mm wafers" that keeps getting repeated by low-knowledge journos and certain commentards.
The UK in general is a low-knowledge nation when it comes to semiconductors. Punching well below its weight.
Their previous owner’s website which I can’t find now, probably it is on archive.org, and I think I linked to it in comments to previous articles on this subject, both here and on ft.com, mentioned a process node of 0.2 microns, which is 200nm.
When I first started posting these comments, none of the journalists were talking about that, just parroting the government’s line about it being super-advanced technology that the Chinese couldn’t obtain without copying it. I converted the units to nm to compare it with Apple/TSMC’s then latest 5nm chips, and SMIC’s then latest 14nm chips.
Newport makes power electronics afaik. They won't have the same "process nodes" as microprocessors, because each transistor will be sized individually for its power handling requirements.
200nm feature size could well be something they use.
It's a bit like saying "there's no such thing as a 1.1mm 3D printer nozzle"
Here is an attempt to answer your question about the relative costs.
TSMC is spending $11 on building a 28nm fab in Germany.
The Register on the 2 3nm fabs in the US "TSMC has confirmed it will build not one, but two advanced chip manufacturing plants in Arizona, more than tripling the Asian foundry giant's original investment to $40 billion,"
https://www.techpowerup.com/308184/tsmc-and-partners-to-invest-usd-11-billion-into-german-based-factory
https://www.techpowerup.com/308184/tsmc-and-partners-to-invest-usd-11-billion-into-german-based-factory
It would therefore appear that 28nm fab costs about 45% less than a 3nm fab. It appears that the larger size does reduce costs beyond lithography machines.
I have no particular knowledge on this topic, I just did a quick bit of research (i.e. used a search engine).
"They can build factories in the USA, but where will they find people with the needed knowledge to work there?"
The plants are mostly automated so they don't have to find many. The bigger question is how they will make any money when they have to petition the US State Department to get a license every time if the customer isn't the US government. I can't remember the cut-off, but processes smaller than a certain size are considered a National Security concern. Build the fab in Asia somewhere and export is no problem.
The USSR was building ICBM's with Z80's and it was a big concern for the US. It's not like they need the latest cpu's for something like that. Perhaps the worry is RADAR, SONAR and signal analysis or supercomputers decrypting communications.
"how they will make any money when they have to petition the US State Department to get a license every time if the customer isn't the US government. I can't remember the cut-off, but processes smaller than a certain size are considered a National Security concern. Build the fab in Asia somewhere and export is no problem.
The USSR was building ICBM's with Z80's and it was a big concern for the US. It's not like they need the latest cpu's for something like that. Perhaps the worry is RADAR, SONAR and signal analysis or supercomputers decrypting communications."
Is there a 21st century equivalent of Systime? Some folk who were there have suggested that Systime existed in part to bypass US export controls and get cloned//cheaper DEC-compatible kit sold to places where the US didn't want it to be.
Things like fast digital signal processors and software defined radio with modern RF electronics make the design and manufacture of surprisingly powerful flexible RF receivers (for comms, and maybe also radar, sonar, etc) a whole lot simpler than it used to be. But again it may not need to be built on bleeding edge fab technology.
https://clearborder.co.uk/resource/understanding-itar-regulations-in-the-uk/
"My question is, how much cheaper is it to build a 32nm fab today than a 3nm fab?"
The cost to build and staff a fab is so expensive that it makes little sense to not build a new facility to the latest spec. I think it's hitting hard limits on scale at this point and what's needed is a new approach to the overall layout. Another factor is how many quarters (not years) before there's a return on the investment. It can be faster if one spends a few more billion to be able to charge much more per chip.
"The cost to build and staff a fab is so expensive that it makes little sense to not build a new facility to the latest spec. "
Yes, on the whole, that's been the process for TSMC and related bleeding edge chip shop outfits for decades.
Global hightech companies like ASML (https://en.wikipedia.org/wiki/ASML_Holding) actually build stuff like the "lithography" kit that is one of the critical pieces of the chip manufacturing process, which is very closely related (both ways) to the chip design process. But there are other sectors where the advanced design and manufacturing are less critical,
"I think it's hitting hard limits on scale at this point and what's needed is a new approach to the overall layout."
It's been said for decades and in some parts of the market it may finally be becoming unavoidable, But advances in semiconductor technology have had a tendency to move forward in ways which have surprised some folks, and some sectors benefit more from robustness than absolute performance.. One such sector is the now ubiquitous SoC-type chips and systems used in e.g. the automotive sector, alleged shortages of which have allegedly restricted supply (and definitely inflated prices) of various vehicles in recent years. Those don't necessarily need the latest design rules or associated fab technologies for something based on an 8051 descendant (yeah I know, unfair example).
"return on the investment [...] can be faster if one spends a few more billion to be able to charge much more per chip."
Maybe. The ability to charge a huge premium for relatively low volumes of Xeon chips and in due course backfill with guaranteed volume revenue (not so much profit) from older x86 chips has helped keep Intel afloat for years longer than might otherwise have been expected - just look at the ever-growing list of Intel's failed non-x86 activities.
Elsewhere in the market, if someone wants to build e.g. an ecosystem for resilient computing for control applications (cars, automation, etc) they’re not going to be starting from the latest fab technology, They might benefit from readily available stuff such as dual-redundant ARM-based SoCs such as the stuff available from TI (and presumably others) in recent years.
"But advances in semiconductor technology have had a tendency to move forward in ways which have surprised some folks"
The arguments I see are based on the process getting down to such a small scale that there won't be enough atoms available and/or quantum. Clock speed for CPUs get to a theoretical point where the speed of light enters into whether signal delays getting from one side of the chip to the other is a problem. Looking at that same limitation on a system level also comes into play. I keep expecting to start seeing functional blocks that are far more three dimensional so there's no going outside the CPU chip for memory and other tasks. I've heard of some projects and the sticking point is cooling in many cases. If somebody comes up with a good way to create sheets of diamond to act like a heat tube, that may be an enabling tech to very 3D "chips".
Another way to phrase it is that we're inducing a brain drain.
When someone develops intellectual property, it's often because they're passionate about it and hope to profit from it.
Selling it entirely is akin to selling your home because the neighbourhood has deteriorated to the point where you no longer feel safe even going out to buy groceries.
This suggests that British business owners no longer see a promising future, making it more appealing to cash out and enjoy retirement.
It's also a testament to the Tories' ineptitude and incompetence. After 13 years marred by corruption, they've left the country in a lamentable state.
We could just ban foreign companies form buying brilliant innovative British companies.
Of course this means that brilliant innovative British inventors know that if they start their company in Britain they have no chance of cashing out. But that's OK because the VCs also know this so will know not to invest in a British company.
Of course the brilliant innovative British inventor could just go somewhere else. But that's OK because somehow in politics headline "$$$ US company was founded by a British inventor who moved to US" is a good headline while "$$$$ British company sold to US company" is a bad headline
"Selling it entirely is akin to selling your home because the neighbourhood has deteriorated to the point where you no longer feel safe even going out to buy groceries."
That's a pretty decent analogy, the thing is this is not a new phenomenon. Case in point Maggie Thatcher's government tried to arrange for the sale of INMOS to American investors at a knock down price before brokering a sale to Thorn EMI (yeah, the fire extinguisher folks). INMOS was the company that built the fab in Newport. Total government investmernt in INMOS was £50m (£235m in today's money going by BoE's inflation calculator) which is peanuts really when you compare it to HS2 which comes in at £247m per km.
El Reg reported on the announcements in May and took the line shared with many commentators that not subsidising new fabs was a mistake.
At the time, some of us pointed out that it was a daft idea to try to compete, and that with many other countries throwing money at the problem, it would be a completely wasted investment for our government. It's not without historical precedence either, since the UK went down exactly that route in the 90's and ended up with nothing to show for it.
So it's nice to read that (at last) reports are coming out that vindicate this view. It might be that other countries are subsidising the sector heavily, but throwing money into companies to make them "competitive" is a cost, not a benefit, so though it might hurt the aspirations of some companies in the UK (who wouldn't want the government to create well paid jobs for them?), it's probably better to let those other countries spend billions on stuff that we'll (mainly) be able to buy cheaply from them later on, and spend our money strategically elsewhere.
Of course in the current environment nothing this government proposes is going to be well received. Unfortunately, it's a bit too easy to mistake politics for sound policy, so the biggest risk here is that Labour come in and make radical chances "because Tory" rather than through any rational analysis of what is needed and effective. We've already seen that intuition is not a good guide for how best to support a complex, rapidly changing and expensive industry - it's only taken six months for the initial objections to the government's plans to be re-evaluated and tempered.
El Reg reported on the announcements in May and took the line shared with many commentators that not subsidising new fabs was a mistake.
As I recall, the line many commentards (myself included) took was:
Nice that they've actually developed a strategy (now do the rest of the economy - UKGov has no over-arching industrial strategy), but:
1. It's table stakes. You don't need to be subsidising fabs to take a proactive interest in things like developing connections with universities, putting the right investment in place to develop clusters (whether that's for shiny logic silicon or - very sensibly - leveraging existing expertise in power silicon and other niche sectors). But they're not taking a tremendously proactive approach on that.
2. Realistically, this government is allergic to investment or infrastructure. It's cynical but not entirely unreasonable- to suggest that them saying "We're going to pursue this niche stuff, so don't expect big press about a new 3nm plant or anything. It's important but very low key" is actually expectation management for "we're promising low because we have no intention of delivering anything anyway" and nobody can be surprised when it all goes quiet and we hear nothing about it again.
3. Good luck delivering even if they do want to - they've gutted the civil service such that DBT would have a real job delivering some of this stuff even with strong ministerial backing.
4. None of this matters because industrial strategy is necessarily long-term. But this government has less than 12months in office, and is then looking at another decade in opposition. And they know it. One can have their own opinions on HS2, but the manner in which Sunak has cancelled it and is then expediting the sale of land, at a loss to the taxpayer is a deliberate attempt to salt the earth and prevent a future government restarting the project. It's scorched earth politics from a party that doesn't give a toss and is busy burning every bridge they can, to make life hard for the next government. They're not good-faith actors. If they do something good, it is purely by accident - but even then, it probably means you haven't looked hard enough to find the donor who is cashing out.
"It might be that other countries are subsidising the sector heavily, but throwing money into companies to make them "competitive" is a cost, not a benefit"
If there were a business advantage for a company to have fabs in the UK, there would be some. Government, no matter where, only has one tool in their box and that's to throw money at things. If they concentrated really hard before their 3 martini lunch, they might develop a new tool that can be used to analyze why there are no fabs in the UK and working on fixing those issues which might cost nothing at all.
In the US, top level politicians are tripping all over themselves to spend taxpayer money on EV charging stations. Going by the recent past for this type of thing, whatever they come up with will be rife with fraud and waste from the onset. A huge problem is when the charging companies have to deal with local government to get planning permission, permits and inspections. The red tape is akin to concertina wire and the locals are always adding to the barrier. The same goes for putting solar on your home. The compliance and permitting costs are a significant percentage of the build and the delays through inspectors showing their importance by nit-picking items or just not being able to visit sites in a timely manner to do their work adds endless costs.
The Tesla Shanghai plant was built in short order partially due to it being a government project (they hold the title) with full time inspectors on site checking and signing off work in real time so the next trade could come in and start their work without delay. If this show of urgency was implemented in other places, loads of money could be saved.
The UK's strategy of focusing narrowly on certain segments of the supply chain where it has a comparative advantage offers an alternative approach
Concentrating on areas where you have comparative advantage was first pointed out as a sound idea by a chap called David Ricardo, about the time Jane Austen was enjoying balls in Bath(*).
(*) This latter may just have been a scurrilous rumour.
Well, after the revelations about the state of the government from the Covid-19 inquiry I can only assume it happened by accident. (The current PM, Rishi Sunk was chancellor at the time of the Johnson 'Government' and was the 'brains' behind 'Eat out to Help out' - the virus - as Chris Whitty had it.)
What truly amazes me about every single one of the politicians and Political Advisors who have given 'evidence' to that inquiry is that nothing was at all their fault, everyone else was an idiot, and NONE OF THEM RESIGNED because of the collective incompetence of the Johnson cabinet. I suppose I should not be amazed by this but I am.
Excuse me while curl up in a corner and cry.
I get abuse because I voted Labour last time.
Not because of any love of labour or its policies, but because I knew what a shit show bozo johnson would be in charge(having seen it where he torpedoed May's and the tories brexit deal simply so he could get the PM's job)
And the covid inquiry.... witness say "bozo johnson could'nt even run a shit show"
But back to the subject.... How much cash are they spaffing on this 'chip' project.... well give it a year of 2 and the funding will be cut in half and we'll endup with a sand quarry for making the silica for the wafers
rather like HS2...............
That matters not! What matters is what will we call it!
Britishfab? Britishchip? Great British Wafers? Britchip? Brexfab? Brexchip?
We'll give it a lovely office in Mayfair and a stately home somewhere Up North, just for a while, until the tory-chum directors have pocketed all the dosh and then they can declare bankruptcy! Top plan! Bosh!
Mao died, and a man called Deng Xiaoping took his place....this would be around 1980.
Deng had a (forty year) plan:
- The Chinese Communist Party would rule the country (one party state)...but....
- The economy would be run on capitalist lines........
- As long as the capitalists stayed out of politics
The results took a long time to materialise:
- Western capitalists (e.g. Apple) noticed that Chinese production was VERY cheap (and profitable!)
- And after the forty years planned by Deng, most of the world's manufacturing was in China!
- And of course, there was almost no manufacturing left in the UK
So here we have a stark contrast:
- Deng implements a FORTY YEAR plan....
- And in the UK the plans only last TILL THE NEXT ELECTION (Thatcher, Major, Blair, Brown, Cameron, May, Johnson, Truss, Sunak)
That would be NINE one year plans!!!!!!!
Deng Xiaoping.....a forty year plan......global success!!! Compare!!!!
@cyberdemon
Please apply for support (supplied by UK taxpayers) to my OFFICIAL email address:
- stephen.barclay.mp@parliament.uk
Of course, UK rules about payments to the needy need to be approved.....five pounds won't be a problem....but people like Michelle Mone who need millions.....that's a COMPLETLY different matter!
Link: https://www.theguardian.com/politics/2023/oct/31/genius-among-morons-dominic-cummings-gives-halloween-display-of-his-ego
Yup.....that would be today........For a sample:
Quote: ".......Especially Matt Hancock, Grant Shapps, Gavin Williamson and Liz Truss. Braindead liars every one of them. Weirdly, this was one of several occasions when one thought Cummings actually had a point....."
My experience of reading Dominic Cummings's blog* is that he has lots of points, it is just that none of them is actually thought through, with clear basis and consequences stated or any justification as to why anyone in their right mind would actually want to do what he says. But then I'm just a nit-picking liberal leftie who has only managed to get four letters published in the Financial Times, so what do I know?**
* Don't go there, just don't. Let me put it this way, if you want to leave a comment it costs you a £10 per month subscription (no, I haven't).
**Definitely TIME FOR BED
My Dear Mister Barclay,
Thank you for your good wishes. Indeed you are correct in that the absence of justification does not prevent something being true. As a PhD mathematical logician I am aware that F(alse) => T(rue) as well as T(rue) => T(rue) . I just like to see the left-hand side of the derivation as well as the conclusion.
I did sleep rather well last night. I hope you did as well.
Best wishes,
E.M.
The high energy and water costs here really put any kind of major silicon manufacturing fab as out of the question, without enormous subsidy.
The UK can be competitive regarding R&D, design capability. We're historically good at that, but it is of course dependent on decent education and a bit of random luck as to what the gene pool spits out next to enter the sector.
But what exactly does the strategy mean in terms of actions on the ground. The private sector will get on with R&D anyway off it's own bat (wherever those that are able happen to reside). To me, the only viable actions then must reside education side. The RAAC debacle, T-level nonsense, A-level replacements. Fluff. Not addressing the underlying problems with investment into education.
Lets face it, I'm sure most el reg readers knew more than any school IT staff did throughout the 80's and 90's. The situation has not improved. (And maybe worse, with computers moving to fluffy UI's instead of requiring a bit of thought on how to learn to use).
Fresh, clean water is important to silicon fabs. For examples, see the recent shortfall in water impacting on production inTaiwan.
Water and energy in Britain are relatively expensive per unit compared to say, virtually all of Asia. There is a reason why most of the big fabs are located in Malaysia, Taiwan, Indonesia, etc.
To put the 1 billion figure into perspective, realise that the the uk govt has already ploughed 1.6 billion into the albatross that is the Bibby Stockholm.
Let that sink in: the future of the UK's high-tech industry - our industry - is worth less than two-thirds of a single ship that will hold 222 refugees.
organize a pissup in a brewery even with 5000 £1000 a day consultants to help out.
Note.. this applies to basically any government since around 1987 when Maggie stopped hitting ministers with her handbag. Say what you like about her and there is a lot to dislike but she got things done. Ok, some of them were the wrong things but this lack of clear decisive decision making is what Yes Minister and Yes Prime Minister exposed so well.
We should return to timely decisions and start making contracts for any public works investments fixed price. It is long past time for the 'do nothing and then I can't be blamed for making a mistake' era to end.
Too late. The money's run out.
Doesn't matter who is in charge now, the damage has been done.
No company will accept the risk of a fixed-price contract now. They will just say 'no bid'.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
