WEBVTT - Bloomberg Businessweek Weekend - February 27th, 2026 0:00:00.280 --> 0:00:07.240 Bloomberg Audio Studios, Podcasts, radio News. 0:00:08.640 --> 0:00:13.000 This is Bloomberg business Week Daily reporting from the magazine 0:00:13.000 --> 0:00:17.599 that helps global leaders stay ahead with insight on the people, companies, 0:00:17.640 --> 0:00:22.520 and trends shaping today's complex economy. Plus global business finance 0:00:22.560 --> 0:00:26.280 and tech news as it happens. The Bloomberg Business Weekdaily 0:00:26.400 --> 0:00:31.320 Podcast with Carol Masser and Tim Steneveek on Bloomberg Radio. 0:00:31.880 --> 0:00:34.760 Hi everyone, Welcome to the Bloomberg Business Week Weekend Podcast. 0:00:35.200 --> 0:00:37.280 It was a week where the political theater of the 0:00:37.280 --> 0:00:40.520 state of the Union and the whisperers over that AI 0:00:40.640 --> 0:00:44.680 scare trade dominated headlines. Even as President Trump touted a 0:00:44.680 --> 0:00:48.599 new era of American energy independence to power the AI boom, 0:00:48.880 --> 0:00:51.560 A nervous Wall Street held its breath for any sign 0:00:51.840 --> 0:00:54.520 that the multi billion dollar AI hype cycle could be 0:00:54.600 --> 0:00:55.320 hitting a wall. 0:00:55.640 --> 0:01:01.120 The results mah lackluster salesforce gave it just appointing sales outlooks. 0:01:01.120 --> 0:01:05.240 Snowflake failed to impress investors. Then came in Vidia. Despite 0:01:05.240 --> 0:01:09.039 another bullish quarterly revenue forecast, it actually drew a lukewarm 0:01:09.080 --> 0:01:12.240 response from investor, signaling that concerns over a potential bubble 0:01:12.280 --> 0:01:13.800 continued away on the company. 0:01:14.200 --> 0:01:17.080 So with the backdrop of the AI trade. Ahead, on 0:01:17.160 --> 0:01:20.280 our program, we lean on a familiar voice to discuss 0:01:20.280 --> 0:01:23.920 the race to dominate the semiconductor supply chain. That race 0:01:24.080 --> 0:01:28.520 no longer your usual suspects, but now include many several members. 0:01:28.560 --> 0:01:31.280 In fact of the mag seven, the author of Chip War, 0:01:31.520 --> 0:01:34.280 The Fight for the World's most critical Technology, joins us 0:01:34.280 --> 0:01:35.080 a little later on. 0:01:35.480 --> 0:01:39.800 Plus, you've got that bonus. Now what Bloomberg pursuits gives 0:01:39.880 --> 0:01:43.720 us the smartest ways to spend and invest that yearly check. 0:01:44.160 --> 0:01:46.480 And it's become less common in men's wear in recent years. 0:01:46.480 --> 0:01:49.720 But Chris Rouser says, now is the time to care 0:01:49.760 --> 0:01:52.040 about belts. He joins us a little later. 0:01:51.880 --> 0:01:53.520 I have a love hate relationship with belts, so I 0:01:53.520 --> 0:01:55.560 can't wait. I know this is from a man's perspective, 0:01:55.600 --> 0:01:56.640 but yeah, I do. 0:01:56.600 --> 0:01:58.400 I do. I just bad bunny. I'll look it up. 0:01:59.320 --> 0:02:01.800 Matter of that to come. We begin, though, with the 0:02:01.920 --> 0:02:05.440 rise of prediction markets. Now on Tuesday night, while millions 0:02:05.480 --> 0:02:08.280 tuned into the State of the Union for all of 0:02:08.280 --> 0:02:10.560 the President's policy talk, or what they hoped would beat 0:02:10.560 --> 0:02:14.200 policy talk, millions of dollars, we're moving on platforms including 0:02:14.200 --> 0:02:16.960 polymarket and Calshi for the prop bets. 0:02:16.880 --> 0:02:19.960 Well, the President's book of a quote, National Turnaround Traders 0:02:20.000 --> 0:02:22.760 wagered over ten million dollars on everything from whether the 0:02:22.760 --> 0:02:25.720 President would mention China to how many times he'd say 0:02:25.800 --> 0:02:29.840 America to the exact count of jd Vance's collapse. 0:02:30.320 --> 0:02:31.120 It's kind of wild. 0:02:31.440 --> 0:02:33.480 Yeah, you can't even like you're kind of speechless. 0:02:33.520 --> 0:02:33.720 I know. 0:02:34.600 --> 0:02:37.440 Once again, I know, repeat the refrain that we hear 0:02:37.480 --> 0:02:39.720 over and over again. I cannot believe people bet on 0:02:39.760 --> 0:02:40.200 this stuff. 0:02:40.240 --> 0:02:40.919 Oh but they do. 0:02:41.000 --> 0:02:42.400 That's what we hear. That's what we hear. But I 0:02:42.440 --> 0:02:43.560 believe everything, which is. 0:02:43.480 --> 0:02:46.440 Why as the volume and popularity of these platforms grow, 0:02:46.720 --> 0:02:50.040 so does the scrutiny, or maybe a little bit more scrutiny. 0:02:50.240 --> 0:02:54.120 Critics call it unregulated gambling and where sometimes the truth 0:02:54.360 --> 0:02:57.200 is whatever the highest bidder says. It is. Writing for 0:02:57.240 --> 0:03:01.360 Bloomberg BusinessWeek about how Polymarket and Calshi are gamifying truth 0:03:02.120 --> 0:03:06.200 is BusinessWeek contributing writer Chris Beem. Chris's story is the 0:03:06.240 --> 0:03:09.760 cover story for the March issue of Bloomberg Business Week magazine. 0:03:10.200 --> 0:03:12.600 One reason I was drawn to this topic is that 0:03:12.600 --> 0:03:15.400 there's so many different ways to come at prediction markets. Yeah, 0:03:15.440 --> 0:03:18.960 but to me, the most interesting question here was, so 0:03:19.080 --> 0:03:23.880 these companies Calshi polymarket, the way they pitch themselves is 0:03:23.919 --> 0:03:27.240 as sources of truth that this is going to be, 0:03:27.560 --> 0:03:30.000 you know, in a world where you know, we can't 0:03:30.000 --> 0:03:33.560 agree on what's true, what's real, what's not, you know, 0:03:33.639 --> 0:03:37.160 deep fakes, what have you, prediction markets are going to 0:03:37.280 --> 0:03:41.400 be something that we can trust more than the news, 0:03:41.480 --> 0:03:44.520 more than pundits. And I was just interested in to 0:03:44.600 --> 0:03:49.040 what extent that is true. And as it turns out, 0:03:50.000 --> 0:03:53.000 in a lot of cases, prediction markets can be quite accurate. 0:03:53.160 --> 0:03:56.800 You know, they're really good at predicting the outcomes of elections, 0:03:56.920 --> 0:04:02.040 especially election right, Yeah, and especially at the top of 0:04:02.080 --> 0:04:06.840 the ballot. Scholars have shown that prediction markets are a 0:04:06.840 --> 0:04:09.400 lot more accurate than polls less so as you go 0:04:09.520 --> 0:04:13.880 down the ballot. But I was also interested in the 0:04:13.960 --> 0:04:18.080 various ways that these prediction markets can be flawed. There 0:04:18.120 --> 0:04:21.320 can be market manipulation, there can be insider trading. So 0:04:21.600 --> 0:04:23.000 that's what I wanted to explore. 0:04:23.120 --> 0:04:25.679 We want to talk about those flaws. But funnily enough, 0:04:25.839 --> 0:04:28.560 as they say, we were actually doing a segment on 0:04:28.600 --> 0:04:31.520 Warner Brothers Discovery earlier, like Who's ultimately going to get 0:04:31.520 --> 0:04:33.679 the deal, and we did go to Polymarket to see 0:04:34.000 --> 0:04:36.039 on that. I just want to mention some headlines crossing 0:04:36.080 --> 0:04:39.360 the Bloomberg terminal because they continue to come. Warner Brothers 0:04:39.400 --> 0:04:42.520 says Paramount has boosted its bid to thirty one dollars 0:04:42.600 --> 0:04:46.000 a share in cash, so they're concerning their risk confirming 0:04:46.080 --> 0:04:49.799 excuse me, the receipt of a revised Paramount proposal. Warner 0:04:49.800 --> 0:04:52.120 brother says the revise bid includes a seven billion dollar 0:04:52.200 --> 0:04:56.600 termination feed, and Warner brother says Paramount proposal could lead 0:04:57.040 --> 0:05:01.320 to a superior bid. We know the Warner Brothers board 0:05:01.400 --> 0:05:05.080 still recommending the Netflix transactions, so this continues on. So anyway, 0:05:05.160 --> 0:05:07.080 this is some These are the kind of things though, 0:05:07.200 --> 0:05:08.080 that are showing up. 0:05:08.240 --> 0:05:11.360 Is there predation for when this story ends? Because this 0:05:11.440 --> 0:05:12.800 is going on for a long time. 0:05:12.880 --> 0:05:14.920 I feel like it just keeps going over new and 0:05:15.000 --> 0:05:17.640 new bids. You get into the fraud aspect, and I 0:05:17.680 --> 0:05:22.200 think we are thinking about people who are maybe you know, 0:05:22.520 --> 0:05:25.640 creating a bid or something and has some control over 0:05:25.680 --> 0:05:28.080 the outcome. That's a real issue. 0:05:28.600 --> 0:05:34.320 Absolutely. There's a lot of markets, especially we can get 0:05:34.320 --> 0:05:36.480 into some of the differences between Calci and poly Market. 0:05:36.640 --> 0:05:40.040 There are differences, right, there are huge differences. CALCI is 0:05:40.240 --> 0:05:44.599 entirely US based. It's regulated by the CFTC, so every 0:05:44.720 --> 0:05:48.600 market that they post needs to be at least run 0:05:48.800 --> 0:05:53.159 by the CFTC for approval, whereas poly Market, most of 0:05:53.240 --> 0:05:56.839 their activity is overseas and so it's not regulated by 0:05:56.839 --> 0:05:59.120 the CFTC. So when you hear about some of the 0:05:59.160 --> 0:06:03.000 fringier markets out there, like you know, will Jesus return 0:06:03.080 --> 0:06:06.200 in twenty twenty six? Will the US discover aliens in 0:06:06.200 --> 0:06:06.880 twenty twenty sex? Wait? 0:06:06.920 --> 0:06:11.880 Does this fringy only? It depends some release from the 0:06:11.880 --> 0:06:13.960 White House my understanding when it comes to aliens, But 0:06:14.080 --> 0:06:15.560 that's another story, but go ahead. 0:06:15.720 --> 0:06:18.440 Yeah, I mean those those kinds of markets tend to 0:06:18.440 --> 0:06:20.840 be more on poly market, Yeah, because they would not 0:06:21.000 --> 0:06:22.600 pass muster with the CFTC. 0:06:22.800 --> 0:06:24.919 Does it make though the other stuff that's on the 0:06:25.080 --> 0:06:28.400 like any less real? Like I don't know, I don't 0:06:28.400 --> 0:06:31.280 even know how to like, how do you gauge the 0:06:31.320 --> 0:06:32.640 credibility of what the. 0:06:32.680 --> 0:06:34.680 Credibility of what, like if Jesus is back? 0:06:36.279 --> 0:06:39.000 So part of it, Like I'm just it is it 0:06:39.080 --> 0:06:39.719 just a game? 0:06:40.720 --> 0:06:44.000 Yeah? I would say, like is there any feel information? 0:06:44.320 --> 0:06:46.760 Like are there people who know what the outcome. And 0:06:46.800 --> 0:06:47.480 this goes back to. 0:06:48.320 --> 0:06:51.760 There's a spectrum of markets here, so you'll not be 0:06:51.800 --> 0:06:55.320 surprised to learn that these companies Calci poly Market they 0:06:55.400 --> 0:06:59.240 like to emphasize the more serious markets where you can 0:06:59.320 --> 0:07:02.359 predict outcome of elections, or you can predict what the 0:07:02.800 --> 0:07:04.960 what inflation is going to be, or you know what 0:07:05.200 --> 0:07:08.800 the Fed is going to do next quarter, and they 0:07:08.920 --> 0:07:14.720 argue that those markets actually can provide useful predictive information 0:07:15.120 --> 0:07:18.800 to help people make decisions in the world. That's one 0:07:18.880 --> 0:07:20.560 end of the spectrum. The other end of the spectrum 0:07:20.720 --> 0:07:24.280 is much more kind of almost entertainment, like predicting whether 0:07:24.320 --> 0:07:25.400 aliens are going to appear. 0:07:25.440 --> 0:07:27.960 Well, there's also the sports betting part of this, or 0:07:28.000 --> 0:07:31.920 maybe predicting outcomes of sporting events. I don't want to 0:07:31.920 --> 0:07:34.080 say betting because I don't want to I don't know 0:07:34.080 --> 0:07:36.520 if it is betting, but you you have told me 0:07:36.560 --> 0:07:39.440 in the past that in reporting this story, you did 0:07:39.440 --> 0:07:41.960 place a bet on one of these right for the Patriots. 0:07:42.200 --> 0:07:45.320 Yeah, yeah, it turned out to be a huge mistake, 0:07:46.160 --> 0:07:50.400 but yes, this is so. The vast majority of the 0:07:50.440 --> 0:07:54.800 activity that happens on CALSHI is sports betting. And I 0:07:54.840 --> 0:07:58.520 think it's it's okay to use that word. Something like 0:07:58.600 --> 0:08:03.320 eighty or ninety percent of Calshi is sports poly market 0:08:03.960 --> 0:08:06.880 one hundred percent of its US activity is sports focused. 0:08:07.200 --> 0:08:09.760 Overseas is a different story. 0:08:09.800 --> 0:08:12.720 One of what you see on poly Market in the 0:08:13.520 --> 0:08:14.680 is sports focused. 0:08:14.400 --> 0:08:16.760 Correct, because those are the only markets they've gotten to 0:08:16.760 --> 0:08:17.800 approved by the CFTC. 0:08:17.920 --> 0:08:23.040 So then so then why are why are people using 0:08:23.080 --> 0:08:25.320 that instead of fan Duel or draft Kings. 0:08:25.840 --> 0:08:28.440 Part of it is just the interface is a lot simpler. 0:08:28.720 --> 0:08:31.200 Like if you go into FanDuel or draft Kings, there's 0:08:31.240 --> 0:08:36.000 a lot of complex measurements and you know, you have 0:08:36.040 --> 0:08:37.960 to know what a spread is and you have to 0:08:38.720 --> 0:08:41.960 It gives you the options to parlay different bets within 0:08:42.040 --> 0:08:45.439 a football game, whereas with Calshi or poly market, it's 0:08:45.520 --> 0:08:47.840 very simple. It's do you think this team is going 0:08:47.840 --> 0:08:51.040 to win or not? And you can make more complex bets, 0:08:51.080 --> 0:08:54.000 but it's it's a lot more user friendly. 0:08:54.360 --> 0:08:57.199 You earlier reference and we'll talk about mentioned markets in 0:08:57.200 --> 0:08:59.880 a few minutes, but you you earlier referenced the accuracy 0:09:00.120 --> 0:09:02.520 these especially at the top of the ballot, and the 0:09:02.520 --> 0:09:05.560 scholarly articles that have cited this as being more accurate 0:09:05.600 --> 0:09:10.120 than polling. Why is that? Is it? Because, as proponents say, 0:09:10.640 --> 0:09:13.000 people are putting money where their their mouth is. 0:09:13.080 --> 0:09:15.240 And that was kind of to my question, like is 0:09:16.320 --> 0:09:18.440 but what's the value? And like do they really have 0:09:18.520 --> 0:09:22.360 some insight that is valuable because it was pretty remarkable 0:09:22.360 --> 0:09:23.560 in some of the elections, Right. 0:09:23.559 --> 0:09:26.760 Yeah, I think what it comes down to is polling 0:09:26.880 --> 0:09:30.880 and prediction markets are just completely different ways of collecting information. 0:09:31.040 --> 0:09:36.439 Polling you're asking people what their personal opinion is, and polling, 0:09:36.600 --> 0:09:39.040 especially in the last decade or two, has in some 0:09:39.080 --> 0:09:42.280 ways gotten worse and worse because of you know, cell 0:09:42.320 --> 0:09:46.760 phones and you know who responds who doesn't. Whereas prediction 0:09:46.840 --> 0:09:50.280 markets you're asking a different question. You're asking them regardless 0:09:50.320 --> 0:09:54.280 of your opinion of who should win an election, what 0:09:54.480 --> 0:09:56.959 is going to happen? What do you think other people 0:09:57.000 --> 0:10:00.559 are going to do? And so really the idea is 0:10:00.600 --> 0:10:05.520 that you're aggregating information like any market, but here you're 0:10:05.640 --> 0:10:10.320 aggregating different understand pieces of understanding about the election and 0:10:10.360 --> 0:10:11.200 how it will play out. 0:10:11.240 --> 0:10:14.000 You know, as as you speak about this, I'm thinking 0:10:14.120 --> 0:10:18.120 to a site like Wikipedia that uses the wisdom of 0:10:18.200 --> 0:10:20.920 the crowd. Supposedly this is also supposed to use the 0:10:20.920 --> 0:10:23.720 wisdom of the crowd, but Wikipedia gets in trouble for bias. 0:10:24.280 --> 0:10:26.160 Is there by is it possible to have bias in 0:10:26.160 --> 0:10:27.040 prediction markets? 0:10:28.720 --> 0:10:31.840 Yes, is the short answer. And the smaller the market, 0:10:32.120 --> 0:10:35.080 the more risk there is for bias. Like a couple 0:10:35.120 --> 0:10:38.120 of years ago, when these markets were very small, some 0:10:38.440 --> 0:10:43.080 academics went onto a few election markets and placed bets 0:10:43.400 --> 0:10:46.480 and that skewed the results. And as a result they said, 0:10:47.040 --> 0:10:49.640 you know, this proves that these markets are not reliable. 0:10:50.000 --> 0:10:52.920 But as they've gotten bigger and bigger, you know, the 0:10:52.920 --> 0:10:56.560 presidential market for the election of twenty twenty four was, 0:10:56.920 --> 0:11:02.040 you know, millions and millions of dollars, that potential for 0:11:02.160 --> 0:11:04.920 manipulation has been greatly reduced. 0:11:05.520 --> 0:11:08.080 We're that with Chris Beam, contributor for Bloomberg Business Week 0:11:08.120 --> 0:11:10.800 here in our Bloomberg Interactive Broker Studio. This is on 0:11:10.800 --> 0:11:14.319 the Bloomberg Business Week cover story. His cover story on 0:11:14.360 --> 0:11:19.120 how prediction markets, polymarket and calshi are gamifying the truth. 0:11:20.320 --> 0:11:22.440 Do they need to be regulated? Does there need to 0:11:22.440 --> 0:11:27.280 be oversight because are people you can make money, so 0:11:27.559 --> 0:11:29.720 are people gaming the system? Potentially? 0:11:30.520 --> 0:11:38.120 Yeah, there is regulation already. So the CFTC currently has 0:11:38.400 --> 0:11:43.880 authority over derivatives trading, and technically these are called event contracts. 0:11:44.000 --> 0:11:47.720 That's a type of derivative. So the CFTC is in 0:11:47.840 --> 0:11:50.800 charge of what contracts are allowed and what are not. 0:11:51.240 --> 0:11:54.679 As you might know, there's this huge debate over whether 0:11:54.760 --> 0:11:59.480 the states should regulate these prediction markets as gambling, and 0:11:59.520 --> 0:12:03.880 that would be much tighter. Uh, it would be probably 0:12:04.400 --> 0:12:06.920 mean taxing them a lot more than they aren't currently. 0:12:07.040 --> 0:12:10.160 Yeah, I mean, because to be fair, the gambling platforms, 0:12:10.240 --> 0:12:12.840 the traditional ones pay are taxed in a way that 0:12:12.880 --> 0:12:15.199 these aren't tax for sports betting. 0:12:15.000 --> 0:12:18.360 Right yeah, yeah, I think when you when you bet 0:12:18.400 --> 0:12:22.480 on Polling Market or Calci, it's taxed, is just regular income. 0:12:22.640 --> 0:12:26.040 So let's just briefly mentioned mentioned mentioned markets. I love 0:12:26.080 --> 0:12:28.679 the way the piece opens talking about this what happened 0:12:28.720 --> 0:12:31.160 with Jerome Powell and this mentioned markets around Jerome Powell. 0:12:31.200 --> 0:12:33.720 That's that's sort of our audience explained, explain what happened 0:12:33.720 --> 0:12:35.920 with the utterance of a word by mistake. 0:12:36.080 --> 0:12:37.800 We only have unfortunately about a minute or so. 0:12:37.880 --> 0:12:42.240 Oh sure, so you know, people know that every word 0:12:42.280 --> 0:12:45.200 that Jerome Powell says in a speech is can can 0:12:45.240 --> 0:12:49.000 move markets. But in this case, you have traders on 0:12:49.080 --> 0:12:52.440 polying market and Calci watching a speech, paying attention to 0:12:52.520 --> 0:12:55.200 literally what words will come out of his mouth, and 0:12:55.280 --> 0:12:58.000 so you know, you had people betting that he was 0:12:58.040 --> 0:13:02.640 going to say the word renovation because they were they 0:13:02.640 --> 0:13:06.560 thought he might be talking about the DOJ investigation. It 0:13:06.640 --> 0:13:10.160 turns out he did end up saying it, but a 0:13:10.160 --> 0:13:12.760 lot of people didn't realize it until after the fact, 0:13:12.800 --> 0:13:14.960 So there was this whole debate did he actually actually 0:13:15.000 --> 0:13:19.880 say the word? Which reveals one of the problems that 0:13:19.960 --> 0:13:22.040 comes up with these markets, which is it could often 0:13:22.080 --> 0:13:26.120 be hard to tell what the actual factual outcome is. 0:13:26.800 --> 0:13:29.440 Chris Bean, he's a contributor for Bloomberg business Week. It's 0:13:29.480 --> 0:13:31.280 the cover story, it's the big take. It's one of 0:13:31.280 --> 0:13:34.040 the most read stories. It's all about prediction markets. We 0:13:34.120 --> 0:13:36.480 barely scratch the surface. Everybody should check it out on 0:13:36.520 --> 0:13:37.400 the Bloomberg terminal. 0:13:41.480 --> 0:13:45.480 You're listening to the Bloomberg Business Weekdaily Podcast. Catch us 0:13:45.559 --> 0:13:49.040 live weekday afternoons from two to five pm Eastern Listen 0:13:49.040 --> 0:13:52.600 on Apple CarPlay and Android Auto with the Bloomberg Business app, 0:13:52.760 --> 0:13:54.560 or watch us live on YouTube. 0:13:55.120 --> 0:13:58.720 It's arguably the fundamental building block of the twenty twenty 0:13:58.760 --> 0:14:02.520 six Economy talking about the semiconductor Well, it's kind of 0:14:02.520 --> 0:14:05.120 been I feel like the fundamental building block for a 0:14:05.160 --> 0:14:08.720 long time. Everything it seems, has some kind of chip 0:14:08.760 --> 0:14:09.000 in it. 0:14:09.240 --> 0:14:09.520 All right. 0:14:09.559 --> 0:14:12.240 Well, this past week, Meta shook the semiconductor world with 0:14:12.280 --> 0:14:15.080 a multi billion dollar deal to buy six gigawatts of 0:14:15.120 --> 0:14:19.800 AI power from AMD Advanced micro Devices. This comes just 0:14:19.920 --> 0:14:24.360 days after Meta doubled down on its partnership with Nvidia. 0:14:24.520 --> 0:14:26.880 The message is clear. If you don't own chips, you 0:14:26.920 --> 0:14:27.760 don't know own the future. 0:14:28.160 --> 0:14:30.680 Someone who knows the chip industry well as Chris Miller. 0:14:30.760 --> 0:14:33.360 He's professor at the Fletcher School at Tufts University, as 0:14:33.360 --> 0:14:35.600 well as a non resident Senior Fellow at the American 0:14:35.680 --> 0:14:38.920 Enterprise Institute. He's also the author of the book on chips, 0:14:39.160 --> 0:14:41.960 the New York Times bestseller, Chip War, The Fight for 0:14:42.040 --> 0:14:44.800 the World's Most Critical Technology. Chris, it's good to have 0:14:44.840 --> 0:14:46.480 you back with us. Carol and I were just talking, 0:14:46.520 --> 0:14:49.800 and you know, you wrote this book close to four 0:14:49.880 --> 0:14:54.160 years ago at this point. How different is the world 0:14:54.240 --> 0:14:59.840 of semiconductors and the world's reliance on semiconductors now th 0:15:00.120 --> 0:15:01.239 when you wrote the book. 0:15:03.760 --> 0:15:05.960 Well, I think the key difference is just the scale 0:15:06.000 --> 0:15:08.960 of spending that we've got right now on building data 0:15:08.960 --> 0:15:12.480 centers and buying all of the chips that are required 0:15:12.680 --> 0:15:15.440 inside of them, and Vidia first and foremost, but not only. 0:15:15.480 --> 0:15:18.680 We've seen companies transformed by the data center build out, 0:15:18.880 --> 0:15:20.720 but in a lot of ways, not much just changed 0:15:20.760 --> 0:15:24.920 other than that we're still sourcing our keysmi conductors from 0:15:24.960 --> 0:15:28.880 Asia and above all from Taiwan, and so the industry 0:15:28.960 --> 0:15:33.160 is supercharged, its size, growing faster than ever, but dealing 0:15:33.200 --> 0:15:35.360 with some of the same supply chain choke points that 0:15:35.400 --> 0:15:36.560 it had a decade ago. 0:15:36.960 --> 0:15:40.880 You know, we talk a lot with one of our 0:15:40.880 --> 0:15:43.880 brilliant voices on the chip world, and that is our 0:15:43.920 --> 0:15:46.080 own Ian King, who has seen a lot of cycles, 0:15:46.120 --> 0:15:47.840 the ups and downs, and we talk a lot about 0:15:47.840 --> 0:15:51.040 that supply demand imbalance and what happens when there's a 0:15:51.040 --> 0:15:54.040 lot of demand, there's not enough supply. The buildout continues, 0:15:54.080 --> 0:15:56.880 the investment, supply goes up, and then there's a glut. 0:15:57.240 --> 0:16:00.320 So I'm just curious what your view is when it 0:16:00.360 --> 0:16:04.440 comes to the cyclical nature of the semiconductor world. Has 0:16:04.520 --> 0:16:08.760 ai changed it or we will see also an overbuild 0:16:08.800 --> 0:16:10.920 and oversupply and then a glut. 0:16:12.800 --> 0:16:15.680 I think if you look historically, certainly you see cycles up, 0:16:15.760 --> 0:16:19.120 cycles down, but there are moments when you see step 0:16:19.240 --> 0:16:23.000 changes in terms of demand for certain types of chips. 0:16:23.000 --> 0:16:25.600 We saw that with smartphones, for example, there was no 0:16:25.640 --> 0:16:28.360 demand for smartphone chips, and then now everyone needs a 0:16:28.400 --> 0:16:31.080 new smartphone every couple of years. And what we're seeing 0:16:31.080 --> 0:16:33.120 in AI right now is that type of step change, 0:16:33.160 --> 0:16:36.120 a huge increase in just the baseline amount of chips 0:16:36.120 --> 0:16:39.040 that we're going to need for data centers driven by AI, 0:16:39.160 --> 0:16:41.040 And so I think we shouldn't expect cycles to be 0:16:41.080 --> 0:16:43.920 over Certainly there'll be ups and downs in the future, 0:16:43.960 --> 0:16:46.680 but it's now very clear. I think that we're just 0:16:46.720 --> 0:16:49.640 going to need a lot more compute for AI purposes 0:16:49.680 --> 0:16:51.320 in the future, and as a result, will need a 0:16:51.360 --> 0:16:53.160 lot more of the AI chips that go inside of 0:16:53.240 --> 0:16:53.800 data centers. 0:16:54.200 --> 0:16:57.840 In your view, is the promise that many people think 0:16:58.400 --> 0:17:03.800 AI is suppose to deliver? Will it actually be delivered? 0:17:03.960 --> 0:17:04.040 Like? 0:17:04.160 --> 0:17:08.280 What's the future that you envision after all this capex 0:17:08.320 --> 0:17:08.720 has spent. 0:17:10.920 --> 0:17:13.159 Why, I think it's in some ways funny when we 0:17:13.200 --> 0:17:15.760 ask will AI deliver? Because if you look just three 0:17:15.840 --> 0:17:18.679 years ago, even after chat schipt was released. There are 0:17:18.720 --> 0:17:23.240 so many things that were not possible that are possible today, 0:17:23.600 --> 0:17:27.199 whether it's the number or the scale of hallucinations and 0:17:27.720 --> 0:17:30.639 answers that chatbots give you, or the scale of work 0:17:30.680 --> 0:17:33.800 that you can put together today that just wasn't possible 0:17:34.480 --> 0:17:37.080 in the past. We've already had so much new capability 0:17:37.160 --> 0:17:40.479 generated just in the couple of years since chat schipt 0:17:40.640 --> 0:17:43.280 that in some ways, I think it's an absurd question 0:17:43.359 --> 0:17:45.200 to ask will AI deliver it already has in a 0:17:45.240 --> 0:17:48.000 lot of ways. But I think I understand why. There's 0:17:48.040 --> 0:17:50.600 plenty of questions about what about the investment is happening 0:17:50.680 --> 0:17:54.320 right now, will that investment pay off? And I think 0:17:54.320 --> 0:17:58.880 there are reasons to think carefully about how much each 0:17:58.920 --> 0:18:02.359 company is putting in over what time horizon. But I 0:18:02.400 --> 0:18:04.919 guess when I when I zoom out, I ask myself, 0:18:04.920 --> 0:18:06.200 do I want a world in which there is more 0:18:06.240 --> 0:18:08.760 access to compute or less? And it seems to me 0:18:08.880 --> 0:18:12.479 that we should, rather than being concerned that there's too 0:18:12.520 --> 0:18:16.159 much compute being put in the ground, be excited that 0:18:16.160 --> 0:18:18.760 there's actually companies willing to invest in the infrastructure that's 0:18:18.760 --> 0:18:20.720 going to deliver all these capabilities. 0:18:20.760 --> 0:18:22.399 Well, I think it's an important question to ask for 0:18:22.440 --> 0:18:24.639 a few reasons, and one of those is because if 0:18:24.680 --> 0:18:27.120 you look at just the capex that these companies are spending, 0:18:27.160 --> 0:18:30.080 you know, one company two hundred billion dollars in a 0:18:30.119 --> 0:18:33.720 single year. That's a big commitment, and that's something that 0:18:33.720 --> 0:18:36.760 they have to convince investors. That is money that is 0:18:36.840 --> 0:18:40.199 going to the right place. That's money that has to 0:18:40.480 --> 0:18:44.399 be earned back. Plus on companies coming to a hyper 0:18:44.400 --> 0:18:47.879 scaler and saying, yes, we think that this money is 0:18:47.920 --> 0:18:49.920 not only being well spent, but then we can use 0:18:49.960 --> 0:18:53.479 this compute to actually create a product that will provide 0:18:53.480 --> 0:18:55.879 a return on investment. So, yes, we've seen a lot 0:18:55.920 --> 0:18:58.600 of the hyperscalers benefit the you know, the anthropics and 0:18:58.640 --> 0:19:02.240 the open aies and and the mistraws like that's amazing stuff. 0:19:02.280 --> 0:19:04.199 But at the end of the day, there are a 0:19:04.359 --> 0:19:08.040 lot of companies that are not necessarily technology companies that 0:19:08.400 --> 0:19:13.240 maybe aren't yet necessarily seeing an increase in productivity as 0:19:13.240 --> 0:19:16.719 a result of these tools. So my question is do 0:19:16.800 --> 0:19:19.119 those companies start to see that do we live in 0:19:19.119 --> 0:19:21.439 a world where this is a layer just like the 0:19:21.440 --> 0:19:23.240 Internet was a layer of technology. 0:19:25.280 --> 0:19:26.960 I think there's no doubt that we're going to have 0:19:27.000 --> 0:19:30.600 AI as a layer that's embedded into all technology that 0:19:30.760 --> 0:19:33.320 we use. And when I look at the economic question, 0:19:33.520 --> 0:19:36.320 I say, first off, is it profitable to serve AI 0:19:36.400 --> 0:19:39.320 systems today, not train the next generation, but serve today's. 0:19:39.359 --> 0:19:42.800 And if you listen to what's publicly reported about OpenAI 0:19:43.000 --> 0:19:46.760 or Anthropic, their margins on their influence business are not 0:19:46.840 --> 0:19:49.720 just positive but quite good. And so that I think 0:19:49.800 --> 0:19:52.639 is pretty strong evidence that the delivery of already existing 0:19:52.680 --> 0:19:56.760 AI services is a pretty profitable business. The next question 0:19:56.840 --> 0:19:58.600 is should we be investing in R and D in 0:19:58.640 --> 0:20:02.480 the next generation? Which is a co very very expensive, 0:20:02.600 --> 0:20:05.400 But I think it's hard to argue we should dramatically 0:20:05.480 --> 0:20:07.439 slow down R and D and AI, just given all 0:20:07.480 --> 0:20:10.360 of the extraordinary improvements that we've seen over the past 0:20:10.440 --> 0:20:13.000 couple of years in terms of capability. And so when 0:20:13.000 --> 0:20:15.040 you start breaking it down and ask, well, which specific 0:20:15.080 --> 0:20:18.680 investment do we think is excessive? Which specific investment would 0:20:18.760 --> 0:20:21.000 rather not be doing? Would you really like to be 0:20:21.040 --> 0:20:23.719 the CEO of the only big tech company that's not 0:20:23.760 --> 0:20:25.920 investing in AI? That doesn't seem like a very comfortable 0:20:26.440 --> 0:20:28.440 place to be. And I'd rather have a situation in 0:20:28.480 --> 0:20:31.800 which the big technology companies are investing more rather than 0:20:32.440 --> 0:20:34.760 putting money in the bank or buying back their stock 0:20:34.800 --> 0:20:37.320 because they didn't have any profitable by investment opternity. 0:20:37.400 --> 0:20:39.480 Well, you know, I would say Apple's doing it differently, 0:20:39.480 --> 0:20:41.480 and I guess time will tell whether or not their 0:20:41.520 --> 0:20:45.560 approach works out. Chris, I am curious to you know, 0:20:46.440 --> 0:20:48.480 who do you talk to, what research do you follow, 0:20:48.520 --> 0:20:50.760 what are the leading voices, what are the leading companies 0:20:50.760 --> 0:20:55.479 that you watch to figure out kind of how the 0:20:55.480 --> 0:20:59.200 semiconductor space is evolving. I mean, we still know TSMC 0:20:59.560 --> 0:21:02.400 is still the big manufacturer of all chips in the world, 0:21:02.480 --> 0:21:05.200