The Information's TITV
00:24:18 · Akash Pasricha, Bradley Tusk
[00:24:18] S0: And before I let you go, I mean, so you you mentioned prediction markets, and that was the other thing people were watching last night. Do... How... What what are your observations about how people were talking about predictions markets last night? I mean, I saw a tweet saying, oh, tonight is a huge win for prediction markets. Think it's quite funny that... [00:24:38] S2: Just because of all the volume and the attention [00:24:40] S0: Yeah. That they Well, even... I think the thing they were pointing to is that... Which I thought was a little bit silly, was the idea that the prediction markets would forecast a race much earlier than, say, you know, news publications would Yeah. Based on the Which is... [00:24:56] S2: They can't [00:24:57] S0: They're do not counting the votes. It's just predictions Right. At the end
00:25:01 · Bradley Tusk
[00:25:01] S2: That doesn't make sense. But I will say, as someone who likes to call myself a political professional, I have come to find that prediction market data is often more useful in assessing a race than polling data, right? Because for a poll, it's a limited sample. You're asking someone, not do you think what will happen, but how do you as one individual feel about something? And the methodology is usually pretty screwed up. Getting people on the phone is hard. Online samples are confusing. Whereas prediction markets, much bigger samples. People are saying what they think will happen based on talking to their friends, their neighbors, looking at other data, whatever else. And, you know, it's just a lot cleaner in many ways actually than than the way a poll is conducted. And so I I do think that the prediction markets were pretty on point for most of what I've been looking at and compared to how the outcomes were. I mean, there were still things that I think people were... Like the Texas, you know, governor's race, I think Abbott was still ahead in the prediction markets heading into last night. But overall, I think prediction markets are very much a force. And I think in the twenty twenty eight presidential race, people are going to spend a lot of time looking at those before deciding, these are more institutional players, who to endorse, major donors, you know, advocacy groups, things like that. And so they did win that way, but I do think that legislation, to regulate them will certainly come out of a Democratic Congress, and
00:26:24 · Mahdi Yahya, Akash Pasricha, CJ Muse
[00:26:24] S3: we'll see what Trump does. [00:26:26] S0: Great. Bradley, want thank you for coming on. As always, that is Bradley Tusk, CEO and founder of Tusk Ventures here on TITV. Stocks are getting a boost following Democrat victories last night in the midterms, but investors remain on edge about the AI trade as tech giants pour billions into infrastructure. AMD's latest earnings offered a bit of a bright spot amid those concerns. Joining me now to discuss the markets and outlook for AI chips is CJ Muse, senior managing director and semiconductor analyst at Cantor Fitzgerald. CJ, welcome to the show. It's great to have you here. [00:26:59] S4: Yeah. Thanks for having me. [00:27:01] S0: Okay. So we're going to get to AMD in a second here, but very quickly, I mean, the market is kind of a bit bifurcated this morning in tech. You've got the hyperscalers are up, the core weaves are down, Oracle's flat. What do you make of this result broadly speaking?
00:27:01 · Akash Pasricha, CJ Muse
[00:27:01] S0: Okay. So we're going to get to AMD in a second here, but very quickly, I mean, the market is kind of a bit bifurcated this morning in tech. You've got the hyperscalers are up, the core weaves are down, Oracle's flat. What do you make of this result broadly speaking? [00:27:17] S4: Yeah. No, I think it's a flight to quality. So I think recurring revenues at the hyperscalers is attractive. I think there's a vision that if data center builds are canceled, that would be a net positive for the hyperscalers in terms of their free cash flow generation. And on the other hand, I think the view is that the amount of data centers approved within The United States is lower. Can debate to what degree. And in that sense, I think that that benefits the lead steer. So I'd kind of put Nvidia in that spot. And at the same time, it begs the question as to what the growth rate will look like for some of the neo clouds who are really requiring that build out to drive their EBITDA. [00:28:08] S0: Yeah. So what do you think happens to CapEx plans then? [00:28:13] S4: Do you [00:28:13] S0: think they change?
00:28:13 · Akash Pasricha, CJ Muse
[00:28:13] S0: think they change? [00:28:14] S4: It's a great question. I think one of the beauties of this cycle is that you started with TSMC sold out. They're under shipping by 20% broadly to their customers. They can't build enough CoWoS. We now have substrate constraints, which was really highlighted by Intel's report last week. And I would say now we have semi equipment shortages. We have shortages for golden screws like ball bearings, power, things that don't come to mind. So if we're really on this path to multi trillion dollar semiconductor market, I think that we're going to be constrained continuously. And so maybe on the margin you see a slowdown, but I still think computers is so sold out that I don't see a slowdown by any of the participants unless it's really regulatory driven. And I guess what I don't know is is there available power and or land shell outside The US that we could build? And I'd say just specifically like to Texas, yes, governor, yes, senate has moved democratic, but you know, the local legislators is is still Republican. And they've already put, you know, moratoriums and increased requirements on data center builds. And locally is really the way to think about data centers, and so they're really gonna drive the bus there. So I don't think much has changed in Texas, to be honest with you. So maybe on the margin, see some cuts, but I don't think it's gonna be broad. You know, today, I think is all about just kind of an assessment of the Blue Wave that I do think caught some people by surprise, including myself.
00:30:07 · Akash Pasricha, CJ Muse
[00:30:07] S0: So that's kind of interesting. You're saying that even despite the results, all of this was priced in and the CapEx plans, that was a reflection of them having done the math on the shortages as well. And they may have been even higher, I guess, is what you're saying, if there had not been shortages and if there had not been moratoriums. That was all baked into the numbers is what I'm hearing from you? [00:30:30] S4: Well, I think there's CapEx and then there's the stocks. I think in CapEx, you're placing orders for your accelerators, your HBM, alongside your planned gigawatt builds. And if you don't have land power shell, you're not signing up for those orders. And so the key question is, does last night cause a rethinking of those plans? I don't think that changes like the next twelve months. I would think it would change kind of beyond twelve months and thereafter. And and, you know, I would hope that cooler heads would would, you know, come to realize how important AI compute is, but that's probably the the wrong thought process for every state. But I think when you think in aggregate, I don't see a huge change. Maybe 10%, 15%, but that's just a guess. [00:31:27] S0: 10%, 15% of what? Sorry. [00:31:30] S4: Of CapEx beyond a year. [00:31:32] S0: Beyond a year. Okay. Got it. Well, I mean, that... But that's... I mean, that's... That could be significant though, you know, given given the volumes that were that were True.