WEBVTT - The Story: The UK’s Race for Second Place in the Chip Wars 0:00:14.800 --> 0:00:18.000 Welcome to tech stuff. I'm os Voloshen here with Cara Price. 0:00:18.200 --> 0:00:22.280 Hi Cara, hi os So. I recently talked with a fellow, 0:00:22.360 --> 0:00:26.400 Britt who reminded me of something that I think you 0:00:26.520 --> 0:00:28.600 Americans often take for granted. 0:00:29.120 --> 0:00:33.479 The center of gravity for the tech industry is so 0:00:33.760 --> 0:00:36.240 much on the West coast of the United States. It's 0:00:36.280 --> 0:00:37.920 you know, you could imagine the whole thing just tipping 0:00:37.960 --> 0:00:40.640 over into the Pacific Ocean. It's so it's so heavy. 0:00:41.200 --> 0:00:44.199 And that's great for California, it's great for Seattle at 0:00:44.320 --> 0:00:47.040 Washington State. But it's worrying and it should be of 0:00:47.080 --> 0:00:48.839 a great concern to the rest of the world. What 0:00:49.000 --> 0:00:50.880 is our future? Are we just going to be takers 0:00:50.960 --> 0:00:51.519 not makers? 0:00:51.800 --> 0:00:54.880 You know, it's really interesting because as an American like 0:00:55.120 --> 0:00:58.240 I rarely imagine what it's like to be a taker 0:00:58.560 --> 0:01:03.160 rather than a maker, because so much innovation comes from 0:01:03.200 --> 0:01:04.040 our tech sector. 0:01:04.400 --> 0:01:07.520 You know, that's right. Nick McEwan has seen the story 0:01:07.520 --> 0:01:10.280 play out from both sides of the Atlantic. He built 0:01:10.360 --> 0:01:14.400 rather an impressive roster of credits, Professor of computer science 0:01:14.520 --> 0:01:20.960 and electrical Engineering at Stanford, Unicorn founder, and a meaningful 0:01:21.000 --> 0:01:25.039 claim to having built the Internet. His work formed the 0:01:25.040 --> 0:01:29.080 basis of Cisco System's GSR router, which at one time 0:01:29.200 --> 0:01:32.080 made up seventy five percent of the backbone of the Internet. 0:01:32.720 --> 0:01:36.520 But Nick recently returned to the UK and he's become 0:01:36.600 --> 0:01:41.040 particularly concerned about the fate of Britain's technology sector because 0:01:41.160 --> 0:01:44.120 he in fact sits on the Prime Minister's Council on 0:01:44.160 --> 0:01:45.840 Science and Technology, which. 0:01:45.640 --> 0:01:50.200 Is like an incredibly powerful position to have. But I'm 0:01:50.200 --> 0:01:53.080 also biased because of how much power this US administration 0:01:53.200 --> 0:01:56.880 handed over to Elon Musk and Doge briefly, very briefly. 0:01:57.560 --> 0:02:00.320 But I'm glad you brought Doze up because I couldn't 0:02:00.320 --> 0:02:03.279 resist asking Nick when we first met on the phone 0:02:03.840 --> 0:02:07.880 whether he was Britain's Doge. He laughed in that classic 0:02:08.639 --> 0:02:11.640 English way, which I think was his way of deflecting 0:02:11.639 --> 0:02:14.200 and saying I think you just did it, his way 0:02:14.240 --> 0:02:17.440 of saying get lost. But of course when I got 0:02:17.480 --> 0:02:19.840 him into the recording studio, I had to ask him 0:02:19.840 --> 0:02:22.200 what he thought about Doge and how Elon ran it. 0:02:22.600 --> 0:02:26.160 Immense respect for him as an entrepreneur and as a technologist. 0:02:26.240 --> 0:02:28.160 I think that he's brought about some of the most 0:02:28.160 --> 0:02:31.320 incredible technology, brought it to market. I think it turned 0:02:31.320 --> 0:02:33.520 out that he wasn't so good in the political sphere. 0:02:33.800 --> 0:02:36.440 I think the sort of the slash and burn approach 0:02:36.600 --> 0:02:39.880 to trying to streamline government just showed that he didn't 0:02:39.880 --> 0:02:41.919 have the experience didn't work, and I think it was 0:02:41.960 --> 0:02:44.720 a pretty much in a visible failure. I think the 0:02:44.760 --> 0:02:48.760 real sad thing about that is the wasted opportunity. I 0:02:48.760 --> 0:02:52.240 think everyone who looks at the US government the British 0:02:52.280 --> 0:02:55.280 government can say, of course it can be streamlined, of 0:02:55.320 --> 0:02:58.840 course it can be improved, and by the abysmal way 0:02:58.880 --> 0:03:01.520 that he handled it, it set us back a long 0:03:01.560 --> 0:03:03.720 way because it's going to be very hard for someone 0:03:03.760 --> 0:03:06.400 else to step in with the same role and say 0:03:06.400 --> 0:03:08.240 that they'll give it a go. I don't think anyone's 0:03:08.240 --> 0:03:10.120 going to trust that someone could do that. 0:03:10.639 --> 0:03:14.640 Well. I'm curious, like how exactly is Nick helping the 0:03:14.760 --> 0:03:17.440 UK government be more efficient and like, how does he 0:03:17.600 --> 0:03:22.080 see you know, this role being something that could actually 0:03:22.120 --> 0:03:24.240 be effective as opposed to dismal. 0:03:24.360 --> 0:03:28.800 Yeah, unlike Dough, she doesn't have executive function. He's literally 0:03:28.880 --> 0:03:31.160 an advisor, right, and one of the things he's most 0:03:31.200 --> 0:03:35.280 interested in is how Britain best positions itself for the future. 0:03:35.880 --> 0:03:39.760 We spoke a lot about AI chips, and interestingly, Nick's 0:03:39.760 --> 0:03:43.520 great ambition for the UK is to take second place 0:03:43.920 --> 0:03:47.520 behind the US in chip manufacturing. He wants, he wants 0:03:47.520 --> 0:03:49.440 to bet, but lifts on the rise right now. 0:03:49.600 --> 0:03:51.200 Some people would argue it's better to be left. 0:03:51.400 --> 0:03:54.120 Nick also sits on the border of this government agency 0:03:54.200 --> 0:03:58.480 called ARIA, the Advanced Research and Invention Agency. For me, 0:03:58.560 --> 0:04:00.280 it was just cool to get an inside view on 0:04:00.320 --> 0:04:03.240 how my own country's government is positioning itself for this 0:04:03.320 --> 0:04:07.400 new technological age, especially the time when something else I'm 0:04:07.440 --> 0:04:10.640 fascinated in, which is how many nation states are starting 0:04:10.680 --> 0:04:14.120 to get deeply involved in the private sector, starting to 0:04:14.200 --> 0:04:17.479 kind of be market participants themselves, like you know, the 0:04:17.600 --> 0:04:20.400 US investment in Intel under Trump. And actually Nick, funny 0:04:20.480 --> 0:04:23.040 enough was a senior figure at Intel for many years. 0:04:23.320 --> 0:04:26.000 But without further ado, here's the rest of our conversation. 0:04:27.560 --> 0:04:31.760 So on Earth persuaded you to depart from sunny California 0:04:31.960 --> 0:04:37.080 to rainy Britain to advise the British government to move 0:04:37.120 --> 0:04:39.560 from the future to a place that often looks like 0:04:39.600 --> 0:04:40.080 the past. 0:04:40.680 --> 0:04:42.800 You know, I left the UK in nineteen eighty nine 0:04:43.360 --> 0:04:47.560 as a somewhat angry twenty six year old feeling there 0:04:47.600 --> 0:04:50.360 was no future for someone that was interested in technology 0:04:50.360 --> 0:04:53.400 and the UK. It was the time of the Thatcher government, 0:04:53.520 --> 0:04:56.359 where all of the emphasis and focus was on the 0:04:56.400 --> 0:04:59.960 city and growing the financial institutions. You could look at 0:05:00.120 --> 0:05:02.280 and say that that paid off, but they came at 0:05:02.279 --> 0:05:06.040 the cost for technology, and it just seemed that engineer 0:05:06.160 --> 0:05:08.120 was still a dirty word. Back in nineteen eighty nine, 0:05:08.480 --> 0:05:10.960 I left saying Okay, I'm done here. My tribe is 0:05:11.000 --> 0:05:14.080 in California. But in the meantime, you know, it would 0:05:14.080 --> 0:05:17.280 be easy to say, oh, the UK just languished. You know, 0:05:17.360 --> 0:05:19.200 we had our ups and downs, and there was certainly 0:05:19.279 --> 0:05:22.839 some big missed opportunities we had over the years. But 0:05:22.920 --> 0:05:27.200 I really felt as though something was changing. There was 0:05:27.520 --> 0:05:31.799 a much more of an optimism about our opportunity, both 0:05:31.839 --> 0:05:35.520 in AI and in technology more generally. I'd seen that 0:05:35.560 --> 0:05:40.680 firsthand through interactions with UK companies, UK universities, so I 0:05:40.760 --> 0:05:43.640 felt sort of felt the ingredients seemed to be coming together. 0:05:44.600 --> 0:05:47.359 And I'd started to read about ARIA, you know, a 0:05:47.360 --> 0:05:51.320 whole new bold way of funding research from government. Thought okay, 0:05:51.640 --> 0:05:55.640 you know there's there's something that's changing there. And you know, 0:05:55.720 --> 0:05:58.560 I'd been it fairly involved in a number of government 0:05:58.560 --> 0:06:03.280 funded research projects Stanford, from NSF, from Darper, and various 0:06:03.320 --> 0:06:05.840 other sources, and so it was very familiar with how 0:06:05.880 --> 0:06:08.960 that worked in the US, and you know, as a consequence, 0:06:09.040 --> 0:06:11.479 I then got more involved with ARIA and joined their board. 0:06:12.040 --> 0:06:17.400 The paradigm established by Elon and Doge was basically, technologists 0:06:17.680 --> 0:06:24.040 are better than bureaucrats at deploying technology and creating opportunities 0:06:24.120 --> 0:06:29.200 for technological growth, which should be kind of national strategic priority. 0:06:29.640 --> 0:06:32.400 I mean, it's interesting because the mission of DOSE was 0:06:32.440 --> 0:06:36.919 really around slashing and cutting, whereas you know, the lesson 0:06:37.000 --> 0:06:40.480 you might take from Silicon Valley and the extent to 0:06:40.480 --> 0:06:44.120 which it was built on the foundation of government investment 0:06:44.360 --> 0:06:46.919 from the United States government over the last decades and 0:06:46.960 --> 0:06:51.720 the extraordinary concentration of wealth and power that investment created, 0:06:51.760 --> 0:06:54.279 both the Silicon Valley, but you know, more widely for 0:06:54.640 --> 0:06:57.599 the United States. As you think about what you're trying 0:06:57.600 --> 0:06:59.960 to do in the UK, it sounds like it's much 0:07:00.160 --> 0:07:02.000 more in that latter camp. 0:07:02.360 --> 0:07:08.800 Yeah. Absolutely, I'm an enormous fan of careful, thoughtful, bold 0:07:08.839 --> 0:07:11.320 investment by government in order to be able to get 0:07:11.360 --> 0:07:15.240 things started and wherever possible, for government then get out 0:07:15.240 --> 0:07:19.680 of the way, because the free market and startups and 0:07:19.920 --> 0:07:24.360 entrepreneurship are a wonderful way to create economic growth. But 0:07:24.480 --> 0:07:29.320 you can kick start them with investment in university's investment 0:07:29.360 --> 0:07:31.640 in small companies in order to be able to get 0:07:31.640 --> 0:07:34.520 this going. And Silicon Valley is a fantastic example of 0:07:34.560 --> 0:07:37.560 this all across silicon value. We've seen how that model works. 0:07:38.160 --> 0:07:40.040 So you know, I think what we can do in 0:07:40.080 --> 0:07:43.200 the UK is several fold. First of all, we have 0:07:43.280 --> 0:07:46.840 to look at strategically important areas that we can invest in. 0:07:47.320 --> 0:07:49.640 One that I've been focused on in particular over the 0:07:49.720 --> 0:07:53.760 last year is the AI chips, so chips for the 0:07:53.880 --> 0:07:58.040 AI market. Nearly everybody that's making money, serious money and 0:07:58.080 --> 0:08:00.960 serious economic growth out of the AI market are the 0:08:00.960 --> 0:08:03.640 people building the chips. In Vidia being the bright shining 0:08:03.680 --> 0:08:06.360 example as the most valuable company in the world, just 0:08:06.400 --> 0:08:08.920 top five trillion dollars in terms of its market cap. 0:08:09.560 --> 0:08:13.080 So you know, where is our role in all of this. 0:08:13.520 --> 0:08:16.120 What is the opportunity for the UK. We have a 0:08:16.240 --> 0:08:21.000 strong history in the semiconductor industry. We have a strong 0:08:21.120 --> 0:08:24.480 history in computer architecture. You know the stories we built 0:08:24.520 --> 0:08:27.280 the first commercial computer we built the first electronic memory. 0:08:27.680 --> 0:08:30.760 We have ARM designs ninety percent of the world CPUs, 0:08:31.040 --> 0:08:33.959 so we've got a strong history in this in this area. 0:08:34.679 --> 0:08:37.040 How do we rebuild that in order to be able 0:08:37.120 --> 0:08:39.720 to take a sort of a chunk of this opportunity. 0:08:39.800 --> 0:08:44.560 To put it into perspective, the market value of Nvidia 0:08:45.440 --> 0:08:48.760 is ten x that of open Ai. Yet when we 0:08:48.840 --> 0:08:52.320 think of AI, we tend to think of companies like 0:08:52.360 --> 0:08:56.920 open Ai and chat GPT obviously fantastic growth, but those 0:08:56.960 --> 0:09:00.880 who are producing the infrastructure upon which it runs, they're 0:09:00.880 --> 0:09:05.199 the ones that are bringing significant value and lasting value. 0:09:05.800 --> 0:09:08.200 It takes a huge investment in order to build a 0:09:08.200 --> 0:09:10.240 business like that, but this is where a lot of 0:09:10.280 --> 0:09:13.800 opportunity is and it's growing currently at about thirty percent 0:09:13.840 --> 0:09:17.720 per year. There are very very few markets that grow 0:09:17.760 --> 0:09:21.360 at thirty percent per year. By twenty thirty three and 0:09:21.440 --> 0:09:26.000 the current growth rate, the AI chip market worldwide will 0:09:26.040 --> 0:09:30.400 be seven hundred billion dollars. That is staggering. Imagine if 0:09:30.480 --> 0:09:35.320 the UK could reach up and grab five percent market 0:09:35.320 --> 0:09:38.960 share of that seven hundred billion dollars global revenue, there 0:09:38.960 --> 0:09:42.920 will be thirty five billion profitable, high margin dollars of 0:09:43.040 --> 0:09:46.520 revenue into the UK. I think that is the opportunity 0:09:46.559 --> 0:09:49.200 before us, and I think if we can be ambitious 0:09:49.480 --> 0:09:53.200 and go after it and invest, train our workforce get 0:09:53.240 --> 0:09:56.640 ready for it, it is an enormous opportunity for economic 0:09:56.679 --> 0:09:57.480 growth for the UK. 0:09:58.000 --> 0:10:01.160 So it's working back from a premise of where you 0:10:01.160 --> 0:10:03.040 need to be in ten years in terms of having 0:10:03.120 --> 0:10:05.880 the best chip talent outside of the US. 0:10:06.280 --> 0:10:09.000 That's right, yes, and I happen to think that we're 0:10:09.000 --> 0:10:11.400 in a very very strong position now. We need to 0:10:12.040 --> 0:10:14.240 double our workforce in order to be able to do it. 0:10:14.280 --> 0:10:17.440 We currently have about twelve thousand chip designers in the 0:10:17.559 --> 0:10:23.000 UK and it's a slightly aging workforce, so we actually 0:10:23.080 --> 0:10:25.360 need to move quite fast just to stay at the 0:10:25.360 --> 0:10:28.240 status quo. But with an industry that's growing at thirty 0:10:28.240 --> 0:10:31.959 percent per year, we probably need to double our workforce 0:10:32.080 --> 0:10:34.400 within the next ten years. So we need to be 0:10:34.440 --> 0:10:37.520 training a lot more people within the country. That's getting 0:10:37.520 --> 0:10:41.560 more people into electrical engineering and computer science degrees. Those 0:10:41.559 --> 0:10:43.960 are mostly people who are doing a level physics right 0:10:44.000 --> 0:10:46.800 now in the UK, so you know they're the seventeen 0:10:46.840 --> 0:10:49.880 eighteen year olds right now, and so you know, I've 0:10:49.920 --> 0:10:52.360 been recommending to the government that they increase their bursary 0:10:52.400 --> 0:10:56.800 scheme to incentivize and encourage more A level students to 0:10:56.840 --> 0:10:59.959 go into electrical engineering so that they can be rare 0:11:00.400 --> 0:11:03.600 to be chip designers in four or five years time. 0:11:04.840 --> 0:11:07.240 And this sends a message, right, It sends a message 0:11:07.240 --> 0:11:10.960 that this is important. We won't get there by training 0:11:11.200 --> 0:11:14.840 our own A level students alone. We need to bring 0:11:14.920 --> 0:11:20.400 in experience chip designers from overseas, So figuring out with 0:11:20.480 --> 0:11:24.320 our visas and encouraging and incentives to entrepreneurs and chip 0:11:24.360 --> 0:11:28.480 designers from overseas, encouraging outposts from chip design companies to 0:11:28.520 --> 0:11:31.600 move to the UK. So we just increase that talent base, 0:11:31.679 --> 0:11:35.200 we increase the concentration. And frankly, the US is not 0:11:35.320 --> 0:11:37.280 so appealing for many people who are in science and 0:11:37.320 --> 0:11:40.240 technology right now, and they'll say, actually, that's not somewhere 0:11:40.320 --> 0:11:41.880 I want to go. I was just talking to someone 0:11:41.960 --> 0:11:45.480 the other day, very experienced technologist who had recently moved 0:11:45.480 --> 0:11:48.120 to the UK and who decided, I don't actually want 0:11:48.120 --> 0:11:50.120 to move to the US right now. This is not 0:11:50.160 --> 0:11:53.079 an environment that feels welcoming where I want to build 0:11:53.080 --> 0:11:55.240 my career, where I want to move my family. I'd 0:11:55.280 --> 0:11:56.600 actually much rather stay here. 0:11:56.880 --> 0:11:58.800 What do you say to people who are that we're 0:11:58.840 --> 0:12:01.080 in a bubble right now? Obviously you were on the 0:12:01.080 --> 0:12:03.960 inside at Cisco, which was briefly the most valuable company 0:12:03.960 --> 0:12:07.319 in the world before the dot com crash. How durable 0:12:07.320 --> 0:12:09.680 and sustainable is this compute build out? 0:12:10.280 --> 0:12:13.920 It will for sure have peaks and troughs, just as 0:12:13.920 --> 0:12:16.760 we saw and the build out of the Internet. I 0:12:16.760 --> 0:12:19.120 think it would be reasonable to expect that we're in 0:12:19.200 --> 0:12:21.520 some kind of hype cycle at the moment. I have 0:12:21.559 --> 0:12:24.240 no idea when that will plateau or when that might 0:12:24.320 --> 0:12:27.080 have a sort of a correction, just as we saw 0:12:27.120 --> 0:12:29.360 with the Internet in the nineteen nineties. It may be 0:12:30.040 --> 0:12:32.320 a little bit like a small version of the bubble 0:12:32.360 --> 0:12:34.400 that may happen in the future. But just think what 0:12:34.480 --> 0:12:37.600 happened afterwards. The Internet didn't stop. It just took a 0:12:37.720 --> 0:12:40.000 setback for a little while, and then it came back 0:12:40.000 --> 0:12:41.760 with a vengeance. Yeah, we got rid of a lot 0:12:41.760 --> 0:12:44.640 of the fluff that was being invested in at that time, 0:12:45.040 --> 0:12:49.040 but the good companies survived. They continue to grow, and 0:12:49.120 --> 0:12:51.400 all the social media companies that came along, all the 0:12:51.440 --> 0:12:54.560 other Internet companies that came along, and then it made 0:12:54.800 --> 0:12:58.360 AI possible because of the interconnection of compute the amount 0:12:58.400 --> 0:13:01.280 of data that was available, and so I expect we'll 0:13:01.280 --> 0:13:03.600 see the same thing happen with AI. It'll go through 0:13:03.640 --> 0:13:08.600 cycles and plateaus, but long term, it's not a technology 0:13:08.640 --> 0:13:11.800 that we'll forget. It's not a technology we'll say, oh, well, 0:13:11.880 --> 0:13:13.040 we don't need this anymore. 0:13:13.320 --> 0:13:15.520 I mean, it's an interesting position for you because you've 0:13:15.520 --> 0:13:20.400 got to observe and participate in the proliferation of the Internet, 0:13:20.920 --> 0:13:24.680 and now the proliferation of AI is starting in your view, 0:13:25.160 --> 0:13:28.800 and you get to actually advise the government on how 0:13:28.800 --> 0:13:30.680 to think about it and how to prepare for a 0:13:30.720 --> 0:13:34.080 future where the UK actually gets to capture part of 0:13:34.080 --> 0:13:36.680 the value. But you also have to think about downside risk. 0:13:37.040 --> 0:13:40.760 I think you have to be prepared to make old 0:13:40.800 --> 0:13:43.920 bets knowing that some of them won't work out, and 0:13:43.960 --> 0:13:46.320 this is hard for government to do. Whether you're a 0:13:46.760 --> 0:13:50.760 short term elected official or you're a longer term civil servant. 0:13:51.400 --> 0:13:55.200 No one feels that they get rewarded for taking a 0:13:55.200 --> 0:13:58.360 big bet that doesn't work out. So the most important 0:13:58.360 --> 0:14:01.840 thing I think is to help government get comfortable with 0:14:02.120 --> 0:14:06.920 the necessity of taking risk, and therefore the best thing 0:14:07.000 --> 0:14:11.559 to do is to rather than thinly spread what will 0:14:11.559 --> 0:14:15.320 always be a limited budget, particularly in fiscally tight times 0:14:15.440 --> 0:14:19.640 like now, is actually to make a reasonable number of 0:14:19.720 --> 0:14:24.920 bets and to go along with experts who have some 0:14:25.040 --> 0:14:29.160 skin in the game investors usually or entrepreneurs, and work 0:14:29.240 --> 0:14:33.160 with them or invest along with them in order to 0:14:33.280 --> 0:14:37.120 be able to use their expertise. And I had said before, 0:14:37.160 --> 0:14:39.800 I think the biggest growth potential that we have in 0:14:39.840 --> 0:14:44.160 this country is in aichips, the design of aichips. I 0:14:44.160 --> 0:14:46.240 don't think we should be manufacturing in them in the UK. 0:14:46.400 --> 0:14:50.520 That's a nation state activity which is well beyond our budget. Actually, 0:14:50.560 --> 0:14:54.320 the opportunity is to do what AMD and Invidio and 0:14:54.480 --> 0:14:58.280 Broadcom do, three of the most successful companies in the world. 0:14:58.480 --> 0:15:02.000 They design them. They create the intellectual property for the designs. 0:15:09.680 --> 0:15:13.480 After the break the race for second place in AI 0:15:13.880 --> 0:15:34.040 chip design, stay with us. Welcome back. I'm as Velocian 0:15:34.120 --> 0:15:36.600 and you've been listening to my conversation with Nick McEwen, 0:15:36.680 --> 0:15:39.880 who sits on the Prime Minister's Council on Science and Technology. 0:15:40.600 --> 0:15:43.360 Before the break, Nick mentioned that he believes the UK 0:15:43.480 --> 0:15:46.640 government should help Britain be a power player in AI 0:15:46.760 --> 0:15:50.560 chip design. But what surprised me is that Nick is 0:15:50.640 --> 0:15:52.120 vying for second place. 0:15:52.480 --> 0:15:55.040 What I'm looking to do is to see how can 0:15:55.120 --> 0:15:58.920 the UK be not first in the world. I think 0:15:58.960 --> 0:16:01.600 we have to concede that the US has as pretty 0:16:01.680 --> 0:16:06.000 much licked at this point, with a vast majority of 0:16:06.040 --> 0:16:09.120 the AI chips being designed there and the products are 0:16:09.160 --> 0:16:12.600 sold from the US. But who's going to be second? 0:16:13.160 --> 0:16:15.360 Which country in the world is going to be second 0:16:15.600 --> 0:16:19.440 in most areas of technology, that's China, We always say China, 0:16:20.080 --> 0:16:24.280 and in fact often they come in, they commoditize, they invest, 0:16:24.480 --> 0:16:27.880 like they did with solar panels and photovoltaics right, and 0:16:27.920 --> 0:16:29.600 then anyone else that was trying to be in that 0:16:29.680 --> 0:16:33.800 space basically was shut out and gave up. They can't 0:16:33.800 --> 0:16:36.040 do it here. And the reason they can't do it 0:16:36.080 --> 0:16:41.280 here is because AI chips use the most advanced processing nodes, 0:16:42.080 --> 0:16:46.120 which in the semiconductor industry right now is three nanometers 0:16:46.200 --> 0:16:50.880 or two nanimeter nodes. China can't use nodes below seven 0:16:50.920 --> 0:16:55.000 nanometers at the moment because of export restrictions that don't 0:16:55.040 --> 0:16:58.840 give them access to the technology that they need, so 0:17:00.160 --> 0:17:03.800 held at Bay. So beyond the US, who are the 0:17:03.840 --> 0:17:07.000 countries that are in the best position with the most 0:17:07.040 --> 0:17:11.000 experience of doing chip design, that have an understanding of AI. 0:17:11.560 --> 0:17:14.200 We are, after all the only country outside the US 0:17:14.240 --> 0:17:17.040 and China that has a foundational model company deep mind. 0:17:17.440 --> 0:17:21.000 We have a long experience of being architects and designers 0:17:21.320 --> 0:17:24.840 of chips. If we can put those together, couple with 0:17:24.840 --> 0:17:29.359 our universities, then we can be the second best in 0:17:29.400 --> 0:17:31.480 the world. So we should be able to get much 0:17:31.520 --> 0:17:33.840 more than that five percent that I talked about earlier. 0:17:34.520 --> 0:17:35.879 We should be able to get more than that, So 0:17:35.880 --> 0:17:37.680 it's a pretty conservative ambition. 0:17:38.240 --> 0:17:41.720 You mentioned ARM, which is one of the top chip 0:17:41.760 --> 0:17:46.840 design companies in the world, founded in Britain, I believe, 0:17:47.119 --> 0:17:51.320 but majority owned by the Japanese company soft Bank. Is 0:17:51.359 --> 0:17:54.120 that a win for the UK chip industry and would 0:17:54.160 --> 0:17:56.520 you be happy if creating the next ARM is part 0:17:56.520 --> 0:17:57.240 of your legacy. 0:17:57.560 --> 0:18:00.200 I'd love to see ARM continue expand it's foot in 0:18:00.240 --> 0:18:03.280 the UK. Has a significant number of R and D 0:18:03.400 --> 0:18:06.199 engineers in the UK and to a large extent, it's 0:18:06.240 --> 0:18:09.320 still very much a British company because of its large 0:18:09.320 --> 0:18:12.520 footprint in Cambridge and other parts of the country, so 0:18:12.840 --> 0:18:16.359 you know it is a true British success. Would we 0:18:16.400 --> 0:18:19.000 have preferred that it's stayed as a UK company, of course, 0:18:19.240 --> 0:18:22.640 but it's still bringing significant revenue and tax base into 0:18:22.680 --> 0:18:25.920 the country. It's still training. They have a very big 0:18:25.960 --> 0:18:30.080 training and internship program, so they are training the next 0:18:30.119 --> 0:18:33.000 generation of chip designers, whether they work for ARM or 0:18:33.000 --> 0:18:35.560 whether they go off and start their own companies. And 0:18:35.800 --> 0:18:37.000 that's how these things work. 0:18:37.440 --> 0:18:39.800 You know, when you think about what it might look 0:18:39.920 --> 0:18:45.080 like to own five percent of the market, what does 0:18:45.119 --> 0:18:48.040 that look like practically in terms of the product people 0:18:48.080 --> 0:18:50.359 will buy. Will it be I can't get access to 0:18:50.440 --> 0:18:53.280 Nvidia and therefore are by British chips? Or I have 0:18:53.400 --> 0:18:55.959 something which is less complicated and therefore I don't need 0:18:56.000 --> 0:18:57.600 the top of the line in video chips, so are 0:18:57.640 --> 0:18:59.680 by British chips? Or is that more ambitious? Is there 0:18:59.720 --> 0:19:03.000 more specialized in applications that you think will be geospace 0:19:03.119 --> 0:19:06.040 or whatever it may be that the Britain can win in. 0:19:06.760 --> 0:19:09.200 So I think it's good to make an analogy here. 0:19:09.440 --> 0:19:12.080 We talked about Cisco earlier in the early days of 0:19:12.240 --> 0:19:15.280 the Internet. They were clearly a very dominant company very 0:19:15.320 --> 0:19:18.640 early on and became for a while the most valuable 0:19:18.640 --> 0:19:21.919 company in the world. And what happened to them was 0:19:21.960 --> 0:19:25.439 what happens to many companies at that scale, and then 0:19:25.640 --> 0:19:27.800 they start to look at markets in terms of are 0:19:27.800 --> 0:19:30.800 they big enough, and so what happened is there would 0:19:30.840 --> 0:19:33.159 be new markets. Frankly, they weren't big enough, so they 0:19:33.160 --> 0:19:35.960 didn't develop them, and so you had this whole ecosystem 0:19:36.000 --> 0:19:39.440 of other companies that popped up producing them and lowerhol 0:19:39.480 --> 0:19:42.320 those markets became a lot bigger over time, and so 0:19:42.760 --> 0:19:46.440 it suddenly diversified the market and there were many opportunities 0:19:46.440 --> 0:19:50.320 for other players in the AI space. The same thing 0:19:50.359 --> 0:19:54.480 will happen. It's sort of inevitable that Nvidia will start 0:19:54.520 --> 0:19:57.640 to look at some adjacent markets or smaller parts of 0:19:57.760 --> 0:20:00.360 the market. Sooner or later, there'll be markets for which 0:20:00.400 --> 0:20:03.080 they say, these aren't really big enough for us to 0:20:03.280 --> 0:20:06.560 go after. They could be in parts of robotics, they 0:20:06.600 --> 0:20:09.000 could be parts of factory automation, they could be in 0:20:09.160 --> 0:20:13.439 autonomous drones, they could be in medical equipment. And someone 0:20:13.480 --> 0:20:16.520 will come up with an innovation that fits that need. 0:20:17.560 --> 0:20:20.399 And we have a number of promising startups, most of 0:20:20.440 --> 0:20:22.520 them in stealth mode in the UK, who are doing 0:20:22.600 --> 0:20:26.440 exactly that they're going after. Particular areas or different sort 0:20:26.440 --> 0:20:29.160 of different techniques that they think will be more efficient, 0:20:29.840 --> 0:20:34.879 that will go after specific portions of that market where 0:20:34.920 --> 0:20:37.840 we can best serve the needs that will be unmet 0:20:37.840 --> 0:20:39.000 by other parts of the market. 0:20:39.800 --> 0:20:42.040 Do you worry about the future of the UK if 0:20:42.040 --> 0:20:45.360 we don't take advantage of this on stony generation opportunity. 0:20:46.400 --> 0:20:49.560 I think it's once in a generation because if we 0:20:49.600 --> 0:20:52.439 don't step up and do it, others will and they 0:20:52.440 --> 0:20:55.040 won't be room left for us. So it's one of 0:20:55.080 --> 0:20:57.080 those things where we have to go and grab a 0:20:57.160 --> 0:20:59.680 stake of the market because you have to be there 0:20:59.680 --> 0:21:02.040 for the long term. You have to be there for 0:21:02.160 --> 0:21:05.399 multiple generations to build the trust and the understanding for 0:21:05.480 --> 0:21:09.280 the customers. Who are those customers. They're big hyperscalers, they're 0:21:09.320 --> 0:21:12.560 people that make cars, they're people that make robots, and 0:21:12.640 --> 0:21:16.280 their system integrators generally people who are building systems out 0:21:16.320 --> 0:21:19.439 of those chips. If we stay in the market for 0:21:19.480 --> 0:21:21.879 a while, we'll have more of those system companies in 0:21:21.920 --> 0:21:25.520 the UK. Today we don't really have a large system integrator, 0:21:25.640 --> 0:21:27.840 someone that takes the chips and builds them up into 0:21:27.840 --> 0:21:31.120 systems in the UK. If we're building the chips, those 0:21:31.119 --> 0:21:33.399 companies will come along. We'll build that expertise. 0:21:33.720 --> 0:21:36.600 You mean cars factory like high tech manufacturing back on 0:21:36.640 --> 0:21:37.640 British shaws. 0:21:37.680 --> 0:21:41.879 Yeah, robots, autonomous vehicles, drones. If we're building the chips 0:21:41.880 --> 0:21:44.880 and we have the ambition they will come. We'll then 0:21:44.920 --> 0:21:48.240 develop packaging technology for putting more of these chips into 0:21:48.240 --> 0:21:53.760 one package, will gain a broader expertise and that's then sustainable. 0:21:53.920 --> 0:21:57.359 It's very difficult for that to then fail once you've 0:21:57.359 --> 0:22:01.840 got that broad expertise across the market as a whole. 0:22:02.280 --> 0:22:04.720 So if we don't do that, we won't have that, 0:22:04.880 --> 0:22:08.159 we won't get that revenue, and we will eventually be 0:22:08.240 --> 0:22:10.719 crowded out and it'll be very very hard for us 0:22:10.760 --> 0:22:11.639 to make in roads. 0:22:12.320 --> 0:22:16.560 Let's talk about manufacturing versus design. You are pushing the 0:22:16.680 --> 0:22:21.040 UK to rush towards second place in terms of the 0:22:21.080 --> 0:22:24.240 second leading chip designer in the world after the US. 0:22:24.760 --> 0:22:27.720 The US is thinking about sovereignty and chips very differently, 0:22:27.760 --> 0:22:31.240 which is thinking about how can we ensure the actual 0:22:31.280 --> 0:22:35.639 manufacturing of the chips. Right, and you have a background 0:22:35.760 --> 0:22:37.520 and work at Intel, I think a still of senior 0:22:37.560 --> 0:22:40.760 advisor there. The US government took a direct stake in 0:22:40.760 --> 0:22:45.399