WEBVTT 00:00:00.000 --> 00:00:03.220 Welcome to the Deep Dive. We're diving into something 00:00:03.220 --> 00:00:06.540 pretty crucial today. It's the whole global race 00:00:06.540 --> 00:00:09.720 for AI dominance. And we always hear about the 00:00:09.720 --> 00:00:11.800 big tech companies battling it out. But this 00:00:11.800 --> 00:00:13.939 is bigger. This is about countries going head 00:00:13.939 --> 00:00:17.280 to head. And the stakes are, well, they're massive. 00:00:17.620 --> 00:00:19.559 We know you guys want to get up to speed on these 00:00:19.559 --> 00:00:22.960 big issues quickly, but also thoroughly. And 00:00:22.960 --> 00:00:25.839 that's what we aim. to do here. Right. To get 00:00:25.839 --> 00:00:28.640 to the bottom of this, we're looking at a really 00:00:28.640 --> 00:00:32.079 interesting discussion by a security PhD. And 00:00:32.079 --> 00:00:34.039 they've kind of laid out how all these national 00:00:34.039 --> 00:00:36.479 interests are getting tangled up with the AI 00:00:36.479 --> 00:00:38.340 revolution. I mean, they touch on everything. 00:00:39.100 --> 00:00:41.859 Economic power, digital sovereignty, even the 00:00:41.859 --> 00:00:44.460 semiconductor industry, which is a huge bottleneck 00:00:44.460 --> 00:00:46.380 right now. Yeah, and we're going to unpack all 00:00:46.380 --> 00:00:48.799 of that for you. Basically, why are countries 00:00:48.799 --> 00:00:51.039 so obsessed with AI? What are the potential consequences? 00:00:51.359 --> 00:00:53.840 So let's just jump right in. Why is AI such a 00:00:53.840 --> 00:00:55.859 big deal for nations right now? Well, the most 00:00:55.859 --> 00:00:58.000 obvious factor, the one that everyone immediately 00:00:58.000 --> 00:01:00.539 thinks of, is the military implications. You 00:01:00.539 --> 00:01:03.000 know, the possibility of smarter drones, better 00:01:03.000 --> 00:01:05.040 cyber weapons, that kind of thing. And yeah, 00:01:05.040 --> 00:01:06.780 those are serious concerns. It gives countries 00:01:06.780 --> 00:01:09.099 a huge strategic advantage. You could take away 00:01:09.099 --> 00:01:11.340 all the economic arguments and nations would 00:01:11.340 --> 00:01:13.760 still be fighting tooth and nail over AI just 00:01:13.760 --> 00:01:15.650 for the military. applications it's not even 00:01:15.650 --> 00:01:17.689 just about having fancier weapons right it's 00:01:17.689 --> 00:01:20.629 about like completely changing the whole strategic 00:01:20.629 --> 00:01:23.890 game but the economic side of this is obviously 00:01:23.890 --> 00:01:26.010 gigantic too oh absolutely I mean you see that 00:01:26.010 --> 00:01:30.230 New York Times estimate of AI adding like 4 .4 00:01:30.230 --> 00:01:32.409 trillion dollars to the global economy every 00:01:32.409 --> 00:01:34.709 year that's something like 4 % of the entire 00:01:34.709 --> 00:01:37.469 world's GDP right now it shows you the scale 00:01:37.469 --> 00:01:40.650 of the opportunity but and this is key these 00:01:40.650 --> 00:01:42.430 benefits aren't going to be spread out evenly 00:01:42.620 --> 00:01:45.060 Some countries are in a much better position 00:01:45.060 --> 00:01:47.359 to capitalize on AI than others just imagine 00:01:47.359 --> 00:01:50.159 countries that control key AI tech They could 00:01:50.159 --> 00:01:52.879 use that economic power to swing trade deals 00:01:52.879 --> 00:01:55.640 influence international policy, you know It creates 00:01:55.640 --> 00:01:57.920 a whole new set of dependencies. So it becomes 00:01:57.920 --> 00:02:00.500 this kind of winner -take -all situation Why 00:02:00.500 --> 00:02:03.260 is AI development so skewed like that? What makes 00:02:03.260 --> 00:02:06.659 it that way? It all comes down to the resources 00:02:06.659 --> 00:02:09.219 needed to build these advanced AI models. We're 00:02:09.219 --> 00:02:12.039 talking immense computing power, tons of data. 00:02:12.539 --> 00:02:14.539 Most players simply don't have the resources 00:02:14.539 --> 00:02:17.240 to even get in the game. Smaller companies, even 00:02:17.240 --> 00:02:19.340 some entire countries. Now, once you do have 00:02:19.340 --> 00:02:22.659 a trained AI model, using it, that's called inference 00:02:22.659 --> 00:02:25.259 becomes super cheap and easy, so you have this 00:02:25.259 --> 00:02:28.560 huge upfront investment, then a really low cost 00:02:28.560 --> 00:02:30.639 to actually use it. And that just makes a winner 00:02:30.639 --> 00:02:33.020 take all dynamic even more extreme. So you absolutely 00:02:33.020 --> 00:02:35.219 have to be in it to win it at the very highest 00:02:35.219 --> 00:02:37.560 level. And if you are, you get this huge advantage 00:02:37.560 --> 00:02:39.699 that lasts a long time. The source we're looking 00:02:39.699 --> 00:02:42.159 at even compares it to like the pharmaceutical 00:02:42.159 --> 00:02:44.280 industry. How does that work? Oh, that's a great 00:02:44.280 --> 00:02:47.039 comparison. Think about it. You have these pharmaceutical 00:02:47.039 --> 00:02:51.069 companies spending billions on research and development, 00:02:51.870 --> 00:02:54.389 then the actual manufacturing of the drug is 00:02:54.389 --> 00:02:57.210 relatively cheap. And the legal systems, especially 00:02:57.210 --> 00:02:59.090 in certain parts of the world, are designed to 00:02:59.090 --> 00:03:02.009 protect those initial R &D costs. It gives the 00:03:02.009 --> 00:03:04.430 big pharma companies huge control. And we might 00:03:04.430 --> 00:03:07.189 see something very similar happen with AI. What 00:03:07.189 --> 00:03:09.370 kind of legal frameworks will pop up to protect 00:03:09.370 --> 00:03:11.830 AI developers? How will governments regulate 00:03:11.830 --> 00:03:14.909 AI development and its use? It could end up looking 00:03:14.909 --> 00:03:17.469 a lot like drug approvals, which would solidify 00:03:17.469 --> 00:03:20.039 the power of the early leaders even more. Yeah, 00:03:20.039 --> 00:03:22.199 it seems like a few powerful nations could end 00:03:22.199 --> 00:03:24.099 up holding all the cards. And then there's this 00:03:24.099 --> 00:03:26.199 whole other consequence of AI we haven't even 00:03:26.199 --> 00:03:28.240 touched on yet, you know, mass job displacement, 00:03:28.500 --> 00:03:30.819 automation. Yeah, that's a big one. It raises 00:03:30.819 --> 00:03:33.740 this huge question about the future of work, 00:03:34.199 --> 00:03:36.419 the future of our economic systems. And sure, 00:03:36.419 --> 00:03:38.139 there's some talk about things like universal 00:03:38.139 --> 00:03:40.699 basic income, you know, as a safety net for people 00:03:40.699 --> 00:03:43.539 who lose their jobs to automation. But honestly, 00:03:43.699 --> 00:03:46.099 most countries seem way more focused on winning 00:03:46.099 --> 00:03:51.000 the AI race than on figuring out the They want 00:03:51.000 --> 00:03:52.979 to make sure their own companies and economies 00:03:52.979 --> 00:03:55.060 are at the front of the pack, not really worried 00:03:55.060 --> 00:03:57.400 about global job losses. And they're doing all 00:03:57.400 --> 00:03:59.699 they can to attract and support their tech companies 00:03:59.699 --> 00:04:02.039 to get there. Absolutely. We're seeing governments 00:04:02.039 --> 00:04:04.080 throwing everything they've got at tech companies. 00:04:05.080 --> 00:04:07.800 Deregulation, tougher cybersecurity, big tax 00:04:07.800 --> 00:04:10.139 breaks, direct financial incentives, you name 00:04:10.139 --> 00:04:13.259 it. That whole bidding war over Amazon's HQ2 00:04:13.259 --> 00:04:15.699 a few years back, that's a perfect example. It 00:04:15.699 --> 00:04:17.620 shows you just how fiercely these regions are 00:04:17.620 --> 00:04:19.980 competing to become the home of these tech giants. 00:04:20.240 --> 00:04:22.740 Like a modern day race to be the technological 00:04:22.740 --> 00:04:25.819 superpower. And the prize is economic growth, 00:04:26.060 --> 00:04:28.300 strategic advantage, the whole nine yards. And 00:04:28.300 --> 00:04:30.519 as these tech companies get bigger and more important, 00:04:30.860 --> 00:04:32.720 their relationship with governments has to change 00:04:32.720 --> 00:04:35.019 too, right? Yeah, there's a definite power shift 00:04:35.019 --> 00:04:37.860 happening as tech companies become essential 00:04:37.860 --> 00:04:40.300 for economic growth and even national security. 00:04:40.680 --> 00:04:42.920 They have a lot more leverage with governments. 00:04:43.740 --> 00:04:46.040 What they want, what they need starts to shape 00:04:46.040 --> 00:04:48.800 policy decisions more and more. And this whole 00:04:48.800 --> 00:04:51.680 nationalistic competition seems to be changing 00:04:51.680 --> 00:04:54.759 how globalization works. The source we're looking 00:04:54.759 --> 00:04:57.100 at suggests we're moving away from this idea 00:04:57.100 --> 00:05:01.000 of a completely open, interconnected world. That's 00:05:01.000 --> 00:05:02.680 one of the key takeaways here. We're seeing a 00:05:02.680 --> 00:05:04.620 shift away from the kind of globalization we've 00:05:04.620 --> 00:05:06.699 had for the last few decades, you know, the US 00:05:06.699 --> 00:05:09.160 -led system. Countries are starting to realize 00:05:09.160 --> 00:05:12.620 that these big global supply chains are vulnerable. 00:05:12.779 --> 00:05:15.279 They're worried about being too reliant on other 00:05:15.279 --> 00:05:17.500 countries for critical technologies. I mean, 00:05:17.600 --> 00:05:20.100 the pandemic and the war in Ukraine really exposed 00:05:20.100 --> 00:05:21.959 those vulnerabilities, didn't they? Suddenly 00:05:21.959 --> 00:05:24.000 we couldn't take access to all these essential 00:05:24.000 --> 00:05:27.410 things for granted anymore. Exactly. It showed 00:05:27.410 --> 00:05:29.649 everyone just how fragile these global supply 00:05:29.649 --> 00:05:31.970 chains are, especially when you're relying on 00:05:31.970 --> 00:05:34.850 a handful of producers for essential stuff. And 00:05:34.850 --> 00:05:36.970 naturally, this thinking has spilled over into 00:05:36.970 --> 00:05:39.089 AI. Countries want to make sure they can build 00:05:39.089 --> 00:05:41.149 their own AI systems that they're not dependent 00:05:41.149 --> 00:05:43.850 on anyone else. And it's not just about having 00:05:43.850 --> 00:05:46.670 a stable supply of AI itself. The discussion 00:05:46.670 --> 00:05:49.509 also brings up the issue of bias. Bias in the 00:05:49.509 --> 00:05:52.129 data used to train these AIs. That's a really 00:05:52.129 --> 00:05:54.529 important point. We're realizing that the data 00:05:54.529 --> 00:05:58.689 used to AI models carries the biases of its origins. 00:05:59.230 --> 00:06:01.610 So if a country relies on AI models developed 00:06:01.610 --> 00:06:03.750 somewhere else, they might be importing ways 00:06:03.750 --> 00:06:06.490 of thinking, priorities that don't really match 00:06:06.490 --> 00:06:08.629 their own. You know that anecdote about European 00:06:08.629 --> 00:06:11.069 friends thinking ChatGPT sounded American? It's 00:06:11.069 --> 00:06:13.550 like the AI has this cultural fingerprint based 00:06:13.550 --> 00:06:16.389 on the data it was trained on. So this push for 00:06:16.389 --> 00:06:18.629 what's it called, digital sovereignty, it's leading 00:06:18.629 --> 00:06:21.339 to some real action from countries. Absolutely. 00:06:21.459 --> 00:06:23.300 We're seeing all kinds of measures to protect 00:06:23.300 --> 00:06:25.639 data and build up domestic tech capabilities. 00:06:26.259 --> 00:06:29.639 The EU's GDPR, that's a big one. It affects companies 00:06:29.639 --> 00:06:32.339 way beyond Europe. It's a clear example of a 00:06:32.339 --> 00:06:34.600 region saying, hey, we're going to control our 00:06:34.600 --> 00:06:36.579 citizens' data, even if it's being handled by 00:06:36.579 --> 00:06:39.720 companies based elsewhere. And Canada has rules 00:06:39.720 --> 00:06:41.980 that stop public institutions from being forced 00:06:41.980 --> 00:06:44.579 to use foreign -hosted services for sensitive 00:06:44.579 --> 00:06:47.600 data. It's all part of this trend. And then there's 00:06:47.600 --> 00:06:50.120 China. They've set some really ambitious goals 00:06:50.120 --> 00:06:53.160 with that Made in China 2025 plan. Oh, yeah. 00:06:53.480 --> 00:06:55.379 China wants to be totally self -sufficient in 00:06:55.379 --> 00:06:58.519 technology by 2025. It's incredibly ambitious. 00:06:58.980 --> 00:07:00.839 Now, they still rely on global supply chains 00:07:00.839 --> 00:07:02.980 for some raw materials. But when it comes to 00:07:02.980 --> 00:07:05.959 AI, they've got some advantages. Like their approach 00:07:05.959 --> 00:07:08.139 to individual rights and privacy, it lets them 00:07:08.139 --> 00:07:10.060 roll out technologies like facial recognition 00:07:10.060 --> 00:07:12.620 really quickly and collect tons of data. Those 00:07:12.620 --> 00:07:14.699 examples from Shanghai are pretty eye opening. 00:07:14.990 --> 00:07:17.250 They are. I mean, they've got cameras everywhere 00:07:17.250 --> 00:07:19.569 tracking everything from traffic violations to 00:07:19.569 --> 00:07:22.089 jaywalking. They even display people's personal 00:07:22.089 --> 00:07:24.730 information on public screens. It's a level of 00:07:24.730 --> 00:07:26.449 data collection that would be really hard to 00:07:26.449 --> 00:07:28.769 imagine in many other parts of the world. But 00:07:28.769 --> 00:07:31.250 it gives them this huge edge in terms of building 00:07:31.250 --> 00:07:34.730 the massive data sets needed to train these advanced 00:07:34.730 --> 00:07:37.790 AI models. And meanwhile, the EU has its own 00:07:37.790 --> 00:07:40.829 set of challenges in this whole AI race. Right. 00:07:40.939 --> 00:07:42.899 The EU was worried about becoming too dependent 00:07:42.899 --> 00:07:45.899 on American AI tech, and they're having trouble 00:07:45.899 --> 00:07:49.339 attracting and keeping top AI talent. Salaries 00:07:49.339 --> 00:07:51.879 are often lower, and there aren't as many cutting 00:07:51.879 --> 00:07:53.839 edge research opportunities compared to North 00:07:53.839 --> 00:07:56.800 America. This brain drain is a big problem for 00:07:56.800 --> 00:07:59.579 their AI ambitions. It's interesting to look 00:07:59.579 --> 00:08:02.500 at all this in a historical context. The US came 00:08:02.500 --> 00:08:04.639 out of World War II as the dominant global power, 00:08:04.660 --> 00:08:06.560 and at first they were all about globalization. 00:08:07.040 --> 00:08:10.420 Exactly. After World War II, the US was the leader. 00:08:10.600 --> 00:08:14.319 both economically and militarily. And they pushed 00:08:14.319 --> 00:08:17.500 for this interconnected global marketplace with 00:08:17.500 --> 00:08:20.199 themselves right at the center. Technological 00:08:20.199 --> 00:08:22.100 independence wasn't a big concern because this 00:08:22.100 --> 00:08:24.019 system worked for them, and they were driving 00:08:24.019 --> 00:08:26.860 most of the innovation anyway. U .S. universities 00:08:26.860 --> 00:08:29.300 were thriving. They attracted the best minds 00:08:29.300 --> 00:08:31.660 from around the world because they had the funding 00:08:31.660 --> 00:08:34.139 and the opportunities for groundbreaking research. 00:08:34.299 --> 00:08:36.399 But that way of thinking has definitely changed 00:08:36.399 --> 00:08:38.799 over time, hasn't it? Oh, yeah. The US started 00:08:38.799 --> 00:08:41.360 seeing industries moving overseas, they tightened 00:08:41.360 --> 00:08:44.039 immigration policies, and other countries started 00:08:44.039 --> 00:08:47.200 catching up technologically really fast. So in 00:08:47.200 --> 00:08:49.440 recent years, the US has become more protectionist. 00:08:49.639 --> 00:08:51.919 They're focused on maintaining their own technological 00:08:51.919 --> 00:08:54.039 edge. Which brings us to what the discussion 00:08:54.039 --> 00:08:57.139 calls the semiconductor Cold War. I think for 00:08:57.139 --> 00:08:59.139 a lot of people, semiconductors are this kind 00:08:59.139 --> 00:09:01.980 of abstract thing, but they're the foundation 00:09:01.980 --> 00:09:05.000 of everything in AI, right? The bedrock, absolutely. 00:09:05.759 --> 00:09:08.559 And the key player in this Cold War is the Taiwan 00:09:08.559 --> 00:09:11.779 Semiconductor Manufacturing Company, or TSMC. 00:09:11.879 --> 00:09:14.159 Most people haven't heard of it, but as our source 00:09:14.159 --> 00:09:16.259 points out, it's the 10th largest company in 00:09:16.259 --> 00:09:18.460 the world by market cap. They're not a consumer 00:09:18.460 --> 00:09:21.139 facing brand. They operate behind the scenes, 00:09:21.639 --> 00:09:24.539 making the advanced chips that power, well, everything. 00:09:25.059 --> 00:09:27.580 Your phone, the most sophisticated AI systems, 00:09:28.120 --> 00:09:30.139 the servers running giant companies like Apple, 00:09:30.460 --> 00:09:33.919 Nvidia, AMD. It's all built on TSMC chips. And 00:09:33.919 --> 00:09:36.259 they basically have a monopoly on these advanced 00:09:36.259 --> 00:09:38.659 chips, something like 90 % market share. It's 00:09:38.659 --> 00:09:41.139 a huge amount of power concentrated in one company. 00:09:41.259 --> 00:09:43.980 It is. And it shows how much the industry has 00:09:43.980 --> 00:09:46.559 changed. It used to be that the big players, 00:09:46.779 --> 00:09:49.200 companies like IBM, Texas Instruments, they did 00:09:49.200 --> 00:09:51.460 everything in -house, designed and manufactured 00:09:51.460 --> 00:09:53.600 their own chips. But as chips got more and more 00:09:53.600 --> 00:09:55.740 complex, more expensive to develop, companies 00:09:55.740 --> 00:09:57.940 started specializing, some focused on design, 00:09:58.159 --> 00:10:00.789 others on manufacturing. Yeah, the story of Intel 00:10:00.789 --> 00:10:03.389 and AMD is a great example of that. AMD started 00:10:03.389 --> 00:10:05.730 outsourcing their manufacturing to TSMC, and 00:10:05.730 --> 00:10:08.049 it seems like that gave them an edge over Intel. 00:10:08.529 --> 00:10:11.009 Exactly. Intel, they still make their own chips, 00:10:11.129 --> 00:10:14.009 but they had a lot of trouble moving to the newer 00:10:14.009 --> 00:10:17.470 manufacturing processes. AMD, by partnering with 00:10:17.470 --> 00:10:19.970 TSMC, who had the expertise and the scale from 00:10:19.970 --> 00:10:22.029 serving the entire global market, they were able 00:10:22.029 --> 00:10:24.129 to make those technological jumps more smoothly. 00:10:24.830 --> 00:10:27.149 Of course, AMD's chip designs themselves were 00:10:27.149 --> 00:10:29.330 also really innovative, so that helped too. And 00:10:29.330 --> 00:10:32.009 the founding of TSMC is a fascinating story in 00:10:32.009 --> 00:10:34.929 itself. It was started by this American semiconductor 00:10:34.929 --> 00:10:37.830 veteran who basically created the entire industry 00:10:37.830 --> 00:10:40.009 in Taiwan. And what's really remarkable is that 00:10:40.009 --> 00:10:42.710 Taiwan wasn't a tech hub back then. It was this 00:10:42.710 --> 00:10:44.730 one guy's vision and determination that created 00:10:44.730 --> 00:10:46.850 a company that now accounts for something like 00:10:46.850 --> 00:10:49.929 15 percent of Taiwan's entire GDP. And the amount 00:10:49.929 --> 00:10:53.029 they invest in R &D is mind blowing. Five percent 00:10:53.029 --> 00:10:55.509 of Taiwan's GDP in a single year just to build 00:10:55.509 --> 00:10:58.049 new factories, those fabs. Yeah, that shows you 00:10:58.049 --> 00:10:59.929 how strategically important this industry is 00:10:59.929 --> 00:11:03.509 for Taiwan. TSNC is expanding, building new fabs 00:11:03.509 --> 00:11:06.649 in the U .S., Germany, Japan. But those are still 00:11:06.649 --> 00:11:08.870 just a small part of their overall manufacturing 00:11:08.870 --> 00:11:11.980 capacity. Taiwan's location being close to China 00:11:11.980 --> 00:11:14.320 for assembly and shipping, that's been a big 00:11:14.320 --> 00:11:16.559 advantage for them. But with all the tension 00:11:16.559 --> 00:11:19.059 between the US and China, it's not clear how 00:11:19.059 --> 00:11:20.879 that's going to play out in the long run. And 00:11:20.879 --> 00:11:23.639 TSMC is way ahead of everyone else technologically. 00:11:24.279 --> 00:11:26.659 Only Samsung and Intel are even close to making 00:11:26.659 --> 00:11:28.620 the same kind of advanced chips. And they're 00:11:28.620 --> 00:11:30.779 constantly pushing the boundaries with these 00:11:30.779 --> 00:11:32.960 nanometer advancements. Can you break that down 00:11:32.960 --> 00:11:35.200 a little bit? What does that even mean? Basically, 00:11:35.340 --> 00:11:37.200 nanometers are these tiny, tiny measurements, 00:11:37.539 --> 00:11:39.909 one billionth of a meter. And the smaller you 00:11:39.909 --> 00:11:42.070 can go, the more transistors you can fit on a 00:11:42.070 --> 00:11:44.909 single chip. More transistors means more processing 00:11:44.909 --> 00:11:47.669 power, better energy efficiency, and ultimately 00:11:47.669 --> 00:11:49.929 the ability to handle the incredibly complex 00:11:49.929 --> 00:11:52.950 calculations needed for advanced AI. So this 00:11:52.950 --> 00:11:55.549 race to shrink the nanometer size, going from 00:11:55.549 --> 00:11:57.909 five millimeter to three nanometer, with TSMC 00:11:57.909 --> 00:12:00.330 already working on two nanometer, it shows you 00:12:00.330 --> 00:12:02.429 just how intense the innovation is in this field. 00:12:02.620 --> 00:12:04.940 So obviously all of this makes countries that 00:12:04.940 --> 00:12:07.440 aren't close allies with Taiwan or South Korea 00:12:07.440 --> 00:12:10.399 pretty nervous, especially China. They need these 00:12:10.399 --> 00:12:13.220 chips and they're falling behind. Exactly. And 00:12:13.220 --> 00:12:16.080 the U .S. has already taken steps to limit China's 00:12:16.080 --> 00:12:19.320 access to this tech. The biggest move was blocking 00:12:19.320 --> 00:12:22.320 the sale of EUV lithography machines to China. 00:12:22.879 --> 00:12:26.080 EUV stands for extreme ultraviolet lithography. 00:12:26.259 --> 00:12:28.679 It's this super advanced technology and you need 00:12:28.679 --> 00:12:31.000 it to make the most cutting edge chips. the five 00:12:31.000 --> 00:12:33.399 millimeter and smaller ones. And there's only 00:12:33.399 --> 00:12:35.340 one company in the Netherlands that makes these 00:12:35.340 --> 00:12:37.960 EUV machines. So that's a major bottleneck. But 00:12:37.960 --> 00:12:40.080 the source we're looking at mentions that Chinese 00:12:40.080 --> 00:12:42.679 chipmakers have been making some unexpected progress. 00:12:42.820 --> 00:12:45.220 Yeah, it was pretty surprising. SMIC, a Chinese 00:12:45.220 --> 00:12:47.039 chipmaker that's closely linked to the military, 00:12:47.360 --> 00:12:49.240 they managed to produce seven millimeter chips 00:12:49.240 --> 00:12:52.000 without anyone knowing. It came out because someone 00:12:52.000 --> 00:12:54.879 analyzed the chips in some Bitcoin mining equipment. 00:12:55.159 --> 00:12:57.600 that definitely raised some eyebrows. It means 00:12:57.600 --> 00:12:59.559 that China might be further along in advanced 00:12:59.559 --> 00:13:01.720 chip manufacturing than we thought, even with 00:13:01.720 --> 00:13:04.039 all the restrictions in place. And this plays 00:13:04.039 --> 00:13:08.320 right into the U .S.'s fears about Taiwan. They're 00:13:08.320 --> 00:13:10.860 worried that China might invade and then all 00:13:10.860 --> 00:13:13.039 that crucial chip -making expertise would be 00:13:13.039 --> 00:13:15.580 in their hands. It's a huge geopolitical concern. 00:13:16.379 --> 00:13:18.779 The U .S. knows that having so much of the world's 00:13:18.779 --> 00:13:21.600 advanced chip manufacturing concentrated in Taiwan 00:13:21.600 --> 00:13:24.750 right next to a potential adversary is a huge 00:13:24.750 --> 00:13:27.590 risk. And that's why the US took the step of 00:13:27.590 --> 00:13:29.950 officially classifying microchip manufacturing 00:13:29.950 --> 00:13:32.950 as critical to national security. That happened 00:13:32.950 --> 00:13:35.639 in late 2022. And it was a big deal. It showed 00:13:35.639 --> 00:13:37.960 that the US was really serious about this. And 00:13:37.960 --> 00:13:40.039 they didn't stop there. They actually expanded 00:13:40.039 --> 00:13:42.100 the rules that prevent Americans from working 00:13:42.100 --> 00:13:44.379 for foreign militaries. They broadened it to 00:13:44.379 --> 00:13:46.879 include any substantial contribution to microchip 00:13:46.879 --> 00:13:49.080 development in countries like China. Which led 00:13:49.080 --> 00:13:51.740 to that really dramatic situation for American 00:13:51.740 --> 00:13:54.240 engineers and managers working at Chinese chip 00:13:54.240 --> 00:13:56.120 companies. Yeah, the US government basically 00:13:56.120 --> 00:13:58.279 gave them an ultimatum. They said, you have to 00:13:58.279 --> 00:14:01.519 choose. Give up your US citizenship or quit your 00:14:01.519 --> 00:14:04.190 job in the Chinese chip industry. And they were 00:14:04.190 --> 00:14:06.710 serious. They threatened legal action. It led 00:14:06.710 --> 00:14:09.309 to a ton of resignations. And it was a big setback 00:14:09.309 --> 00:14:11.669 for China's plans to build up their chip baking 00:14:11.669 --> 00:14:14.190 industry. It shows that the U .S. is not playing 00:14:14.190 --> 00:14:16.470 around anymore. They've taken a clear side in 00:14:16.470 --> 00:14:18.970 this tech competition. So to sum it all up, what 00:14:18.970 --> 00:14:21.909 we're seeing is this fierce global competition 00:14:21.909 --> 00:14:25.610 over AI driven by the need for better weapons 00:14:25.610 --> 00:14:28.610 and the potential for huge economic gains. And 00:14:28.610 --> 00:14:30.970 this competition is pushing countries to become 00:14:30.970 --> 00:14:33.590 more self -sufficient in technology. They want 00:14:33.590 --> 00:14:36.049 to control their own data, build their own AI 00:14:36.049 --> 00:14:38.450 systems, and make sure they have access to the 00:14:38.450 --> 00:14:41.429 essential components, like those advanced semiconductors. 00:14:41.629 --> 00:14:44.250 It's led to what some people are calling a semiconductor 00:14:44.250 --> 00:14:47.529 Cold War, where control over chip manufacturing, 00:14:48.009 --> 00:14:50.830 especially TSMC's dominance, has become the central 00:14:50.830 --> 00:14:54.190 point of geopolitical tension. It's a key battleground 00:14:54.190 --> 00:14:57.470 in the overall fight for AI supremacy. It all 00:14:57.470 --> 00:15:00.889 ties together. The drive to be the leader in 00:15:00.889 --> 00:15:04.009 AI is having these huge ripple effects on global 00:15:04.009 --> 00:15:07.090 politics, on trade, on how technology develops. 00:15:07.500 --> 00:15:09.399 So here's something to think about. We have this 00:15:09.399 --> 00:15:12.200 intense competition between nations, and we know 00:15:12.200 --> 00:15:14.539 that semiconductor technology is essential to 00:15:14.539 --> 00:15:16.440 all of this. Given how much is at stake, and 00:15:16.440 --> 00:15:18.240 how quickly things are changing, what kind of 00:15:18.240 --> 00:15:20.419 unexpected things might happen, what new alliances 00:15:20.419 --> 00:15:22.879 might form, what disruptions might we see in 00:15:22.879 --> 00:15:25.419 this race for AI dominance, and how will all 00:15:25.419 --> 00:15:27.440 of this impact our lives and the stability of 00:15:27.440 --> 00:15:29.340 the world in ways we can't even predict right 00:15:29.340 --> 00:15:31.759 now, that's something to chew on. Thanks for 00:15:31.759 --> 00:15:32.980 joining us on the Deep Dive.