1 00:00:02,480 --> 00:00:21,600 Speaker 1: Bloomberg Audio Studios, podcasts, radio news. 2 00:00:24,000 --> 00:00:27,080 Speaker 2: This is Bloomberg Tech coming up the biggest AI export 3 00:00:27,160 --> 00:00:30,920 Speaker 2: restriction yet and Tropic disables access to its most advanced 4 00:00:30,920 --> 00:00:34,600 Speaker 2: models for all foreign nationals. The US concern jail breaks 5 00:00:34,600 --> 00:00:36,720 Speaker 2: in the code in video, is seeking to raise at 6 00:00:36,760 --> 00:00:39,320 Speaker 2: least twenty billion dollars from its first corporate bond sales 7 00:00:39,320 --> 00:00:42,720 Speaker 2: since twenty twenty one. We'll have the details. And SpaceX 8 00:00:42,720 --> 00:00:45,600 Speaker 2: shares throttle up on day to trading, adding to a 9 00:00:45,640 --> 00:00:49,920 Speaker 2: blockbuster public market's debut on Friday. And this is what 10 00:00:50,240 --> 00:00:54,640 Speaker 2: markets look like. Hello, Welcome to Bloomberg Tech Monday, June fifteenth. 11 00:00:55,120 --> 00:00:58,280 Speaker 2: There is a big focus on what is happening between 12 00:00:58,280 --> 00:01:00,000 Speaker 2: the US and Iran Later in the show. We'll get 13 00:00:59,960 --> 00:01:02,480 Speaker 2: to that. But SpaceX in its second day of trading 14 00:01:02,760 --> 00:01:05,480 Speaker 2: up eight percent and it's putting a big spotlight on 15 00:01:05,560 --> 00:01:08,480 Speaker 2: the AI trade, on AI infrastructure. Then aw's that one 16 00:01:08,560 --> 00:01:11,800 Speaker 2: hundred's off session highs, but we're again close to records, 17 00:01:11,840 --> 00:01:14,280 Speaker 2: and the narrative is that there is feel good in 18 00:01:14,319 --> 00:01:16,720 Speaker 2: this equity market. Let's get to our top story. A 19 00:01:16,800 --> 00:01:20,919 Speaker 2: major escalation in Washington's approach to AI. The US government 20 00:01:20,959 --> 00:01:24,560 Speaker 2: has ordered Anthropic to block foreign access to its most 21 00:01:24,640 --> 00:01:28,320 Speaker 2: advanced models after officials flagged security vulnerabilities and it's new 22 00:01:28,360 --> 00:01:31,640 Speaker 2: to released Fable five system and Mephos five to two. 23 00:01:32,000 --> 00:01:35,800 Speaker 2: Talks r underway between Anthropic and the Trump administration joining 24 00:01:35,840 --> 00:01:38,640 Speaker 2: us now our AI editor Bloomberg, Seth Figman and in 25 00:01:38,760 --> 00:01:41,280 Speaker 2: DC Bloomberg's Mike Shepard and Shepp I want to start 26 00:01:41,319 --> 00:01:44,480 Speaker 2: with you. This came from the Commerce Department. Explain the 27 00:01:44,480 --> 00:01:47,600 Speaker 2: export restriction and what the government asked of Anthropic. 28 00:01:49,240 --> 00:01:52,240 Speaker 3: Well, this all dropped right as everyone was still so 29 00:01:52,440 --> 00:01:56,800 Speaker 3: focused on the SpaceX ip IPO. Five twenty one pm 30 00:01:56,880 --> 00:02:00,800 Speaker 3: on a Friday, a letter a directive to Anthropic saying, look, 31 00:02:01,360 --> 00:02:04,560 Speaker 3: we are concerned about foreign nationals having access to two 32 00:02:04,640 --> 00:02:07,760 Speaker 3: of your most advanced models, essentially the key products that 33 00:02:07,800 --> 00:02:10,440 Speaker 3: the company would want to pitch as it's heading into 34 00:02:10,480 --> 00:02:14,240 Speaker 3: this nearly trillion dollar IPO. And this really marked a 35 00:02:14,320 --> 00:02:17,440 Speaker 3: huge shot across the bow once again for this AI 36 00:02:17,560 --> 00:02:21,080 Speaker 3: developer from the Trump administration, and in essence it forced 37 00:02:21,120 --> 00:02:24,959 Speaker 3: the company's hand and Thropic had to deny access worldwide 38 00:02:25,000 --> 00:02:28,280 Speaker 3: to these two products Fable five and Mythos five in 39 00:02:28,400 --> 00:02:31,679 Speaker 3: response to the government's move. Now, this in a way 40 00:02:31,760 --> 00:02:36,359 Speaker 3: It really marked not only an unprecedented step for any administration, 41 00:02:36,680 --> 00:02:39,320 Speaker 3: it really marked a break from what the Trump administration 42 00:02:39,480 --> 00:02:41,840 Speaker 3: had laid out just a little more than a week 43 00:02:41,880 --> 00:02:46,120 Speaker 3: ago with that order, that executive order, the President Donald 44 00:02:46,120 --> 00:02:49,480 Speaker 3: Trump sign that was aimed at cybersecurity and AI and 45 00:02:49,520 --> 00:02:53,200 Speaker 3: that it included language that said this does not should 46 00:02:53,280 --> 00:02:57,560 Speaker 3: not be interpreted as establishing a licensing regime. Yet an effect, 47 00:02:57,600 --> 00:02:59,840 Speaker 3: what we are seeing is that the government is blocking 48 00:03:00,160 --> 00:03:03,519 Speaker 3: the release of these models over the kinds of safety concerns, 49 00:03:03,560 --> 00:03:06,240 Speaker 3: in essence, the vetting that the government had wanted to 50 00:03:06,240 --> 00:03:08,920 Speaker 3: see done on a voluntary basis, and yet it is 51 00:03:08,960 --> 00:03:12,840 Speaker 3: setting its foot down now over security concerns and essentially 52 00:03:13,080 --> 00:03:15,399 Speaker 3: holding off the release of these two products. 53 00:03:16,320 --> 00:03:19,840 Speaker 2: Seth this was an action directed at the use of 54 00:03:19,880 --> 00:03:23,920 Speaker 2: the models Fable five me PUS five by foreign nationals, 55 00:03:23,919 --> 00:03:27,600 Speaker 2: but Anthropic made the decision to close off access to everyone. 56 00:03:28,200 --> 00:03:30,680 Speaker 2: What was the explanation that the AI Lab gave. 57 00:03:31,160 --> 00:03:34,240 Speaker 4: Yeah, Well, they're saying access to foreign nationals not just 58 00:03:34,400 --> 00:03:37,520 Speaker 4: in other countries, but those residing in the United States. 59 00:03:37,600 --> 00:03:39,440 Speaker 4: And it seems that Acanthropic made the decision that this 60 00:03:39,600 --> 00:03:42,400 Speaker 4: was unworkable to allow this model to remain on the 61 00:03:42,440 --> 00:03:45,000 Speaker 4: market for those restrictions. So they've pulled up while they 62 00:03:45,040 --> 00:03:48,440 Speaker 4: continue to talk with the government. Now there's a certain 63 00:03:48,520 --> 00:03:51,800 Speaker 4: irony here. You know, two days before this move, Anthropics 64 00:03:51,800 --> 00:03:53,640 Speaker 4: and Dario Medea had come out and said, you know, 65 00:03:53,720 --> 00:03:56,200 Speaker 4: he thinks the government should have the ability to ban 66 00:03:56,360 --> 00:03:59,440 Speaker 4: models that pose certain risks. Two days later, the government 67 00:03:59,440 --> 00:04:01,800 Speaker 4: took a step has effectively caused a model or two 68 00:04:01,840 --> 00:04:04,280 Speaker 4: models to be pulled from the market. But in this case, 69 00:04:04,400 --> 00:04:07,920 Speaker 4: you know, Anthropic is saying it's essentially misguided and the 70 00:04:08,000 --> 00:04:10,640 Speaker 4: risks outlined may not be as severe as the government's right. 71 00:04:10,600 --> 00:04:12,920 Speaker 2: I think they are seth and Frobic saying there's a 72 00:04:12,920 --> 00:04:15,640 Speaker 2: misunderstanding here. The issue that the government laid out is 73 00:04:15,680 --> 00:04:18,200 Speaker 2: something called a jail break in the code. It has 74 00:04:18,279 --> 00:04:20,920 Speaker 2: to do with guardrails that are placed on the models themselves. 75 00:04:20,960 --> 00:04:23,600 Speaker 2: Could you explain to the Bloomberg Tech audience what the 76 00:04:23,720 --> 00:04:24,680 Speaker 2: jail break concern? 77 00:04:24,800 --> 00:04:27,200 Speaker 4: Is important to note that at this point we have 78 00:04:27,279 --> 00:04:31,120 Speaker 4: not seen the research that Anthropic and former White House 79 00:04:31,200 --> 00:04:33,240 Speaker 4: or David Stacks have alluded to that implies that there 80 00:04:33,279 --> 00:04:35,400 Speaker 4: is some kind of jail break former ability. What we 81 00:04:35,480 --> 00:04:38,080 Speaker 4: understand is that it seems to be pretty narrow in nature. 82 00:04:38,080 --> 00:04:42,880 Speaker 4: It may actually be something that enables cyber defenders to 83 00:04:42,960 --> 00:04:46,200 Speaker 4: use the technology to spot and protect against flaws. But 84 00:04:46,240 --> 00:04:48,880 Speaker 4: what Anthropic is saying is, if not some universal jail break, 85 00:04:48,920 --> 00:04:52,000 Speaker 4: some wider flaw that could be breached in a potentially 86 00:04:52,080 --> 00:04:54,200 Speaker 4: damaging way. And it seems to believe that the government 87 00:04:54,320 --> 00:04:57,640 Speaker 4: is either overstaving or misunderstanding the nature of the risk. 88 00:04:57,480 --> 00:05:00,960 Speaker 2: Here, Mike, real quick, what's happened behind the scenes. 89 00:05:02,240 --> 00:05:04,159 Speaker 3: Well, that is what we are trying to find out next, 90 00:05:04,279 --> 00:05:07,040 Speaker 3: of course, and we want to first know what remedial 91 00:05:07,080 --> 00:05:09,839 Speaker 3: steps the company could take at the government's behest, what 92 00:05:10,000 --> 00:05:12,840 Speaker 3: those might look like, how long that might take, and 93 00:05:12,880 --> 00:05:15,400 Speaker 3: whether the company really would follow suit. We don't know 94 00:05:15,440 --> 00:05:18,400 Speaker 3: how long this will unfold or if there will even 95 00:05:18,480 --> 00:05:21,280 Speaker 3: be a legal challenge, And also we have to look 96 00:05:21,440 --> 00:05:25,320 Speaker 3: at the politics of this abroad. The President just arrived 97 00:05:25,320 --> 00:05:28,120 Speaker 3: in Geneva for the summit in Evion with other G 98 00:05:28,240 --> 00:05:30,760 Speaker 3: seven leaders. AI is on the agenda, and now it 99 00:05:30,800 --> 00:05:33,840 Speaker 3: will be a more urgent issue, especially as we see 100 00:05:34,000 --> 00:05:37,880 Speaker 3: economies including the European Union, examining the question of tech 101 00:05:37,960 --> 00:05:41,400 Speaker 3: sovereignty and the concern that they have about perhaps over 102 00:05:41,520 --> 00:05:44,599 Speaker 3: reliance in US technology that may be at risk of 103 00:05:44,640 --> 00:05:47,080 Speaker 3: being taken away from them at a moment's notice. 104 00:05:47,600 --> 00:05:50,800 Speaker 2: Bloombergs Seth Figmann and Bloombergs Mike Shepard in DC, thank 105 00:05:50,839 --> 00:05:53,680 Speaker 2: you very much. The industry weighs in for more in 106 00:05:53,720 --> 00:05:57,120 Speaker 2: the impact across the AI ecosystem. We're joined by Joel Pinot, 107 00:05:57,520 --> 00:06:00,719 Speaker 2: chief AO officer at Cohre, a frontier AI firm focused 108 00:06:00,760 --> 00:06:04,719 Speaker 2: on LMS and AI applications for business customers, and shep 109 00:06:04,839 --> 00:06:07,719 Speaker 2: just set it out pretty clearly. At issue here is 110 00:06:07,760 --> 00:06:10,200 Speaker 2: whether or not the government should be able to say 111 00:06:10,640 --> 00:06:13,760 Speaker 2: you can or you cannot use a particular piece of 112 00:06:13,800 --> 00:06:16,839 Speaker 2: technology because of the concern that was outlined. What is 113 00:06:16,880 --> 00:06:18,080 Speaker 2: Coher's position. 114 00:06:17,760 --> 00:06:21,320 Speaker 5: On that, indeed, And you know co Here's a company 115 00:06:21,320 --> 00:06:24,440 Speaker 5: that's fully focused on developing models and a gentic platform 116 00:06:24,560 --> 00:06:27,560 Speaker 5: for enterprises and government, and so we've been talking to 117 00:06:28,160 --> 00:06:31,640 Speaker 5: that segment of the market for a number of years now. 118 00:06:32,360 --> 00:06:33,120 Speaker 2: What we've been. 119 00:06:33,040 --> 00:06:37,479 Speaker 5: Hearing is that customers want the ability to really control 120 00:06:37,560 --> 00:06:40,000 Speaker 5: their tech stack. They want to make sure that the 121 00:06:40,080 --> 00:06:43,880 Speaker 5: technology runs in a predictable way, that it's stable, that 122 00:06:43,960 --> 00:06:47,400 Speaker 5: it has high guarantees in terms of confidentiality and security, 123 00:06:48,000 --> 00:06:51,520 Speaker 5: so there's a number of companies and governments already who 124 00:06:51,520 --> 00:06:55,279 Speaker 5: are opting for on premise installation where they can control 125 00:06:55,279 --> 00:06:58,000 Speaker 5: the model and they can control all of the software stack. 126 00:06:59,240 --> 00:07:01,280 Speaker 5: Clearly know all the news from the last few days 127 00:07:01,400 --> 00:07:04,640 Speaker 5: is a huge wake up call for others across the industry. 128 00:07:05,000 --> 00:07:07,640 Speaker 5: The fact that models which a few days before were 129 00:07:07,760 --> 00:07:11,080 Speaker 5: presented as some of the most capable in industry suddenly are 130 00:07:11,120 --> 00:07:15,040 Speaker 5: no longer accessible means that there can be a lot 131 00:07:15,080 --> 00:07:17,840 Speaker 5: of volatility in terms of access to the technology, and 132 00:07:17,880 --> 00:07:20,440 Speaker 5: that's got to be a big concern for companies. 133 00:07:21,280 --> 00:07:24,400 Speaker 2: Joelle, This news hit five twenty one pm Eastern Time Friday, 134 00:07:24,440 --> 00:07:27,520 Speaker 2: So I appreciate it's been just a couple of days 135 00:07:27,560 --> 00:07:31,720 Speaker 2: over the weekend, but has Cohed had any inbound right 136 00:07:32,480 --> 00:07:35,000 Speaker 2: If there are entities that can't get access to meet 137 00:07:35,040 --> 00:07:37,880 Speaker 2: those in fable five, maybe they use you. 138 00:07:39,560 --> 00:07:42,400 Speaker 5: I would say, like huge number of inbounds. It's coming 139 00:07:42,440 --> 00:07:47,119 Speaker 5: from business customers who are looking to diversify the set 140 00:07:47,360 --> 00:07:50,560 Speaker 5: of technology that they can rely on, both for the 141 00:07:50,640 --> 00:07:54,240 Speaker 5: core models and we do build our own models and 142 00:07:54,920 --> 00:07:58,880 Speaker 5: for the orchestration and governance platform. A large number of 143 00:07:58,960 --> 00:08:02,880 Speaker 5: inbounds from governments around the world especially outside of US 144 00:08:03,080 --> 00:08:06,640 Speaker 5: and China. Governments are concerned about their own ability to 145 00:08:06,680 --> 00:08:11,320 Speaker 5: access the technology, but also the economic trajectory of companies 146 00:08:11,320 --> 00:08:15,120 Speaker 5: and citizens within their frontiers. And then of course a 147 00:08:15,160 --> 00:08:17,720 Speaker 5: lot of inbound interest from investors who are wondering what 148 00:08:17,840 --> 00:08:20,400 Speaker 5: does that mean. We know there's of course the big 149 00:08:20,440 --> 00:08:24,120 Speaker 5: IPOs coming, a lot of volatility from that point of view, 150 00:08:24,360 --> 00:08:25,800 Speaker 5: and a lot of people who are just trying to 151 00:08:25,960 --> 00:08:29,800 Speaker 5: understand the situation and see what have the news changed 152 00:08:29,840 --> 00:08:31,840 Speaker 5: in terms of what should be their investment strategy. 153 00:08:32,720 --> 00:08:35,240 Speaker 2: I had asked Seth to explain the concept of a 154 00:08:35,320 --> 00:08:38,959 Speaker 2: jail break in the code, right, that is the reported 155 00:08:39,240 --> 00:08:42,920 Speaker 2: issue that the US government had with me FOS in 156 00:08:42,920 --> 00:08:47,400 Speaker 2: Fab five. Explain technically how that works and how you'd 157 00:08:47,400 --> 00:08:51,480 Speaker 2: approach I guess explaining it if it was you that 158 00:08:51,720 --> 00:08:54,400 Speaker 2: was dealing with the Commerce Department on a restriction like this. 159 00:08:55,200 --> 00:08:58,160 Speaker 5: Yeah, And so the concept of a jail break can 160 00:08:58,200 --> 00:09:02,560 Speaker 5: be used pretty loosely, but quite specifically. There's a sense 161 00:09:02,559 --> 00:09:06,520 Speaker 5: in which a particular system has some guardrails around the 162 00:09:06,559 --> 00:09:10,280 Speaker 5: intended use of that system, and a jail break happens 163 00:09:10,320 --> 00:09:14,040 Speaker 5: when someone manages to infiltrate the system in ways that 164 00:09:14,120 --> 00:09:17,520 Speaker 5: weren't intended. So in this specific case, we can imagine, 165 00:09:17,559 --> 00:09:21,160 Speaker 5: you know, from Meethos to Fable, anthropic hasts put in 166 00:09:21,240 --> 00:09:26,200 Speaker 5: some guardrails that circumvent that prevent the ability for the 167 00:09:26,240 --> 00:09:29,040 Speaker 5: model to be used to create cybersecurity attacks. So we 168 00:09:29,080 --> 00:09:32,840 Speaker 5: can presume that the specific jail break that was identified 169 00:09:32,920 --> 00:09:36,240 Speaker 5: is the ability to go around these defenses and in 170 00:09:36,400 --> 00:09:41,400 Speaker 5: fact make the model produce outputs that would that would 171 00:09:41,559 --> 00:09:45,280 Speaker 5: leverage some of the cybersecurity capabilities of the model. 172 00:09:46,679 --> 00:09:49,800 Speaker 2: Something the Center of Anthropics response is that many of 173 00:09:49,800 --> 00:09:53,040 Speaker 2: their own staff wouldn't be able to access this technology 174 00:09:53,160 --> 00:09:55,520 Speaker 2: based on the restriction, right because many of their staff 175 00:09:55,520 --> 00:10:00,400 Speaker 2: are foreign nationals. It also highlights a world dependent right 176 00:10:00,440 --> 00:10:04,199 Speaker 2: now on US technology. You're a Canadian company, I get that, 177 00:10:04,200 --> 00:10:08,559 Speaker 2: But what's your thinking on that issue, the concentration of 178 00:10:09,040 --> 00:10:12,160 Speaker 2: leadership in the field of small group of US companies. 179 00:10:12,160 --> 00:10:13,800 Speaker 2: Maybe you dispute. 180 00:10:13,400 --> 00:10:17,360 Speaker 5: That, No, I don't fully dispute in the sense that, 181 00:10:17,440 --> 00:10:19,520 Speaker 5: like where we sit today, there's literally, you know, four 182 00:10:19,559 --> 00:10:21,560 Speaker 5: countries in the world that have the ability to train 183 00:10:21,640 --> 00:10:26,840 Speaker 5: Foundation model US China, France, Canada. Right now, even in 184 00:10:26,880 --> 00:10:29,319 Speaker 5: those companies in the US, there's a large number of 185 00:10:29,360 --> 00:10:33,520 Speaker 5: foreign nationals. The you know, inbound talent into the US 186 00:10:33,760 --> 00:10:37,719 Speaker 5: has been formidable force for the progress of this technology. 187 00:10:38,040 --> 00:10:40,199 Speaker 5: So this is going to be a big challenge. And 188 00:10:40,559 --> 00:10:43,280 Speaker 5: the other thing that's worth mentioning is, in many ways, 189 00:10:43,280 --> 00:10:45,360 Speaker 5: a lot of the developments in AI of the last 190 00:10:45,720 --> 00:10:49,160 Speaker 5: few years, you know, are at this level of ideas 191 00:10:49,200 --> 00:10:51,959 Speaker 5: and concepts. Sure we have the models, but really the 192 00:10:52,800 --> 00:10:55,720 Speaker 5: recipe for training these models sits in the head of 193 00:10:55,760 --> 00:11:00,000 Speaker 5: a relatively small number of people. But there's, you know, 194 00:11:00,160 --> 00:11:03,559 Speaker 5: a large enough amount of circulation between the various companies 195 00:11:03,559 --> 00:11:06,440 Speaker 5: that we can assume that the capabilities to do that exist, 196 00:11:06,880 --> 00:11:08,080 Speaker 5: not just at asan tropic. 197 00:11:09,240 --> 00:11:12,199 Speaker 2: Joelle Pinot of Cohre, thank you very much for your time. 198 00:11:12,400 --> 00:11:14,800 Speaker 2: What is our top story today? Now coming up, in 199 00:11:14,960 --> 00:11:17,840 Speaker 2: Vidia heads back to the bomb market after a five 200 00:11:17,920 --> 00:11:20,440 Speaker 2: year absence. Why is the AI giant looking to raise 201 00:11:20,559 --> 00:11:23,920 Speaker 2: billions of dollars while it's sitting on a mountain of cash. 202 00:11:24,080 --> 00:11:26,560 Speaker 2: That's what the stock picture looks like. Be right back. 203 00:11:26,600 --> 00:11:35,000 Speaker 2: This is Blimberg Tech. Today's big number twenty billion dollars. 204 00:11:35,040 --> 00:11:37,400 Speaker 2: In Vidia is looking to raise at least that in 205 00:11:37,440 --> 00:11:40,560 Speaker 2: its first bond sales since twenty twenty one. According to 206 00:11:40,600 --> 00:11:44,480 Speaker 2: sources seven trenches maturities from two to thirty years. The 207 00:11:44,559 --> 00:11:48,800 Speaker 2: proceeds will be used for general corporate purposes, including refinancing 208 00:11:48,920 --> 00:11:51,880 Speaker 2: existing debt as in Vidia joins other AI leaders like 209 00:11:51,920 --> 00:11:55,440 Speaker 2: Amazon and Alphabet in raising large amounts of capital to 210 00:11:55,520 --> 00:11:58,920 Speaker 2: fund AI related infrastructure. But why does a company so 211 00:11:59,080 --> 00:12:03,520 Speaker 2: flush with cash need to borrow more? Bloomberg Intelligence reckons 212 00:12:03,559 --> 00:12:07,040 Speaker 2: that it's to focus on funding partners, b I Senior 213 00:12:07,080 --> 00:12:09,840 Speaker 2: Credit and List. Robert Schiffman joins us, Now, when you 214 00:12:09,920 --> 00:12:11,920 Speaker 2: publish the react, that's the bit that jumps out of 215 00:12:11,960 --> 00:12:15,600 Speaker 2: me off the page, right, Why would they do that? Well, 216 00:12:15,720 --> 00:12:18,760 Speaker 2: they have a track record of investing in the kinds 217 00:12:18,760 --> 00:12:22,520 Speaker 2: of startups that rely on their technology. Extrapolate on that thesis. 218 00:12:22,760 --> 00:12:25,800 Speaker 6: Yeah, well, listen, you can never be too rich. It's 219 00:12:25,840 --> 00:12:28,960 Speaker 6: a good problem to have. Quite frankly, in Video does 220 00:12:29,080 --> 00:12:31,200 Speaker 6: have a cash problem. They have too much of it. 221 00:12:31,440 --> 00:12:33,600 Speaker 6: So people are scratching their heads this morning thinking why 222 00:12:33,600 --> 00:12:36,880 Speaker 6: are they raising more? Listen, you take money down when 223 00:12:36,880 --> 00:12:40,720 Speaker 6: it's available. There is so much money chasing after this 224 00:12:40,840 --> 00:12:42,839 Speaker 6: AI build out. They're going to put in the highest 225 00:12:42,880 --> 00:12:45,280 Speaker 6: quality names that they can, and I think in Video 226 00:12:45,360 --> 00:12:49,400 Speaker 6: is just taking opportunity here to basically borrow at effectively 227 00:12:49,440 --> 00:12:52,040 Speaker 6: a zero rate, at least from a whack perspective, to 228 00:12:52,040 --> 00:12:54,640 Speaker 6: give them more financial flexibility. I think they're likely to 229 00:12:54,640 --> 00:12:58,640 Speaker 6: pour money, probably back into street strategic partners like open Ai, 230 00:12:58,920 --> 00:13:01,960 Speaker 6: like Anthropic, and who knows, maybe a SpaceX, which is 231 00:13:02,040 --> 00:13:04,040 Speaker 6: likely to do their own bond deal might come to 232 00:13:04,559 --> 00:13:07,120 Speaker 6: in Vidia for some financing down the line as well. 233 00:13:08,120 --> 00:13:10,640 Speaker 2: There's like the smart cost of capital angle to this. 234 00:13:10,840 --> 00:13:15,600 Speaker 2: The shareholder return story has changed recently with Nvidia. You know, 235 00:13:16,400 --> 00:13:18,680 Speaker 2: I'm a student of your work, right, It's not that 236 00:13:18,880 --> 00:13:21,880 Speaker 2: unusual for companies flush or cash to just say, like, okay, 237 00:13:22,080 --> 00:13:24,360 Speaker 2: this is an interesting thing for the finance or with 238 00:13:24,440 --> 00:13:26,959 Speaker 2: the treasury team to do. Explain the kind of nitty 239 00:13:26,960 --> 00:13:27,560 Speaker 2: gritty of that. 240 00:13:27,880 --> 00:13:30,560 Speaker 6: Well, listen, you know, the weighted average cost of capital 241 00:13:30,600 --> 00:13:32,960 Speaker 6: is what they're looking at. The weighted average cost of 242 00:13:32,960 --> 00:13:35,680 Speaker 6: debt when you have a five trillion dollar equity market 243 00:13:35,760 --> 00:13:39,760 Speaker 6: cap is basically zero. So you could take money, borrow 244 00:13:39,800 --> 00:13:42,920 Speaker 6: it from from bondholders, and use it for equity or 245 00:13:42,960 --> 00:13:46,680 Speaker 6: equity style investments at a little or no cost. I 246 00:13:46,679 --> 00:13:47,959 Speaker 6: think you're going to see more of this. You sell 247 00:13:48,000 --> 00:13:50,920 Speaker 6: a lot of this from Apple over the years. 248 00:13:50,440 --> 00:13:52,000 Speaker 2: We're not used to. 249 00:13:52,000 --> 00:13:53,880 Speaker 6: It now though, because most of these other companies have 250 00:13:53,960 --> 00:13:56,520 Speaker 6: now turned free cashual negative and they need it for 251 00:13:56,600 --> 00:14:00,000 Speaker 6: funding for their AI investments. That's not what Nvidia needs. 252 00:14:00,040 --> 00:14:02,040 Speaker 6: And Video is probably going to generate a couple hundred 253 00:14:02,080 --> 00:14:04,560 Speaker 6: billion dollars of free cashrow over the next year, so 254 00:14:04,600 --> 00:14:07,240 Speaker 6: they're swimming in the money. But again, when people are 255 00:14:07,240 --> 00:14:09,319 Speaker 6: willing to give it away for next to nothing, you're 256 00:14:09,360 --> 00:14:11,360 Speaker 6: going to take it and you're going to give yourself 257 00:14:11,360 --> 00:14:12,520 Speaker 6: that financial flexibility. 258 00:14:13,600 --> 00:14:16,280 Speaker 2: Robert Schiffman, who leads our credit research of Bloomberg Intelligence, 259 00:14:16,320 --> 00:14:19,280 Speaker 2: thank you so much. Let's turn to SpaceX, who shares 260 00:14:19,360 --> 00:14:22,920 Speaker 2: our extending gains for their second day of trading following 261 00:14:23,200 --> 00:14:28,040 Speaker 2: the company's blockbuster public market's debut Friday. The trading offers 262 00:14:28,080 --> 00:14:32,360 Speaker 2: this other glimpse into investors demand for what was one 263 00:14:32,400 --> 00:14:34,400 Speaker 2: of the world's most private companies. And then they did 264 00:14:34,400 --> 00:14:37,120 Speaker 2: the biggest IPO in history. There are some mechanics to 265 00:14:37,160 --> 00:14:40,080 Speaker 2: what's happening in the market next joining us as Stephanie Aliaga, 266 00:14:40,200 --> 00:14:43,800 Speaker 2: JP Morgan Asset Management, Global market strategist. You know, the 267 00:14:43,800 --> 00:14:46,800 Speaker 2: big anxiety in the days leading up to the IPO 268 00:14:47,560 --> 00:14:51,880 Speaker 2: was this great rebalancing of the passive funds. Because nasdak 269 00:14:52,160 --> 00:14:57,400 Speaker 2: Russell Indicas fast tracks SpaceX's inclusion, it's day two. Shares 270 00:14:57,400 --> 00:15:00,560 Speaker 2: are still higher. But how much does the market need 271 00:15:00,640 --> 00:15:04,280 Speaker 2: time to digest before it has to undertake that rebalancing. 272 00:15:04,840 --> 00:15:07,480 Speaker 7: It's a great question, and that's exactly the point. You know, 273 00:15:07,600 --> 00:15:10,000 Speaker 7: this issue of all of this issue and sitting the 274 00:15:10,040 --> 00:15:13,040 Speaker 7: market doesn't happen all at once. You know, relative to 275 00:15:13,120 --> 00:15:16,680 Speaker 7: the size of how much supply of SpaceX there will 276 00:15:16,720 --> 00:15:18,360 Speaker 7: be in the market right now, we're just looking at 277 00:15:18,400 --> 00:15:20,960 Speaker 7: a small fraction of it. Right four and a half 278 00:15:20,960 --> 00:15:23,960 Speaker 7: percent was that first day's volume, and that's going. 279 00:15:23,880 --> 00:15:25,040 Speaker 2: To ramp up over time. 280 00:15:25,160 --> 00:15:27,320 Speaker 7: We think that six month mark is really going to 281 00:15:27,360 --> 00:15:30,680 Speaker 7: be the point where you might see more volatility just 282 00:15:30,680 --> 00:15:33,920 Speaker 7: to give in a greater supply in the markets, and 283 00:15:33,960 --> 00:15:36,600 Speaker 7: then that increasing over time gradually over the over the 284 00:15:36,600 --> 00:15:38,680 Speaker 7: course of the year, particularly given the S and P 285 00:15:39,200 --> 00:15:42,800 Speaker 7: taking a more temperate pace there, but given that longer 286 00:15:42,880 --> 00:15:45,040 Speaker 7: term horizon, I mean, that's key here. You know, this 287 00:15:45,080 --> 00:15:49,560 Speaker 7: is a market where we are looking at a large 288 00:15:50,080 --> 00:15:53,240 Speaker 7: bump in terms of IPO issuance this year. This year 289 00:15:53,320 --> 00:15:56,800 Speaker 7: projected around two hundred and sixty billion dollars, but for 290 00:15:56,880 --> 00:16:02,280 Speaker 7: a market the size of sixty five trillion dollars that 291 00:16:02,440 --> 00:16:05,680 Speaker 7: is less than one percent of the overall market cap 292 00:16:05,760 --> 00:16:08,240 Speaker 7: of the market, a market that is also fifty five 293 00:16:08,320 --> 00:16:11,680 Speaker 7: percent greater than it was in twenty twenty one. So 294 00:16:12,120 --> 00:16:14,720 Speaker 7: we think so long as the market continues to trunk 295 00:16:14,720 --> 00:16:17,960 Speaker 7: along here with robust fundamentals, we just had a really 296 00:16:18,000 --> 00:16:22,160 Speaker 7: solid earning season. This AI capex boom continues to funnel 297 00:16:22,400 --> 00:16:26,120 Speaker 7: earnings growth and revenue growth across the market, and households 298 00:16:26,120 --> 00:16:29,760 Speaker 7: remain in pretty strong buyers retail activity is strong. We 299 00:16:29,880 --> 00:16:32,520 Speaker 7: think that this is a market that can digest this 300 00:16:32,640 --> 00:16:35,120 Speaker 7: increase in issuance and its issuance that we have really 301 00:16:35,160 --> 00:16:37,880 Speaker 7: been strapped for for a number of years. 302 00:16:38,080 --> 00:16:39,840 Speaker 2: I want to go back to something we show them the 303 00:16:39,840 --> 00:16:41,640 Speaker 2: screen a minute ago, and it's basically more of an 304 00:16:41,720 --> 00:16:44,720 Speaker 2: fyi that in the end, they sold more than the 305 00:16:44,720 --> 00:16:47,400 Speaker 2: five hundred and fifty five point six million shares and 306 00:16:47,480 --> 00:16:49,920 Speaker 2: raised more than seventy five billion dollars. Right, this is 307 00:16:50,200 --> 00:16:53,640 Speaker 2: basically an instant exercising of the green shoe, eighty three 308 00:16:53,640 --> 00:16:56,800 Speaker 2: point three million shares raising proceeds of eighty five point 309 00:16:56,840 --> 00:16:59,960 Speaker 2: seven billion. This has been a story that started to 310 00:17:00,000 --> 00:17:03,440 Speaker 2: about a race to IPO, SpaceX and now likely anthropic 311 00:17:03,480 --> 00:17:06,720 Speaker 2: and open AI, and it quickly changed to a story 312 00:17:06,760 --> 00:17:10,320 Speaker 2: about race to capital. Alphabet came in massive equity offering 313 00:17:10,480 --> 00:17:14,480 Speaker 2: in Nvidia, tapping the corporate bond market. Do you see 314 00:17:14,480 --> 00:17:15,400 Speaker 2: a race to capital? 315 00:17:16,440 --> 00:17:21,320 Speaker 7: I think the economics around the AI boom continue to mount. Now, 316 00:17:21,359 --> 00:17:24,240 Speaker 7: obviously these companies are all in different pieces of this 317 00:17:24,359 --> 00:17:27,359 Speaker 7: value chain in are raising capital variety of different reasons, 318 00:17:27,680 --> 00:17:29,560 Speaker 7: but it is something that investors are going to have 319 00:17:29,640 --> 00:17:32,040 Speaker 7: to get used to. I think, you know, we as 320 00:17:32,080 --> 00:17:35,440 Speaker 7: great as the balance sheets of some of these companies are. 321 00:17:36,040 --> 00:17:38,560 Speaker 7: This AI wave is just getting to be that much 322 00:17:38,600 --> 00:17:43,080 Speaker 7: more expensive, and all of the revenues around AI will 323 00:17:43,080 --> 00:17:46,360 Speaker 7: eventually come, but this is going to be a multi 324 00:17:46,520 --> 00:17:50,280 Speaker 7: year phenomenon, and right now we are ramping up to 325 00:17:50,440 --> 00:17:54,359 Speaker 7: spend significantly more around AI infrastructure and all of these 326 00:17:54,400 --> 00:17:57,159 Speaker 7: different pieces of the value chain. I think that points 327 00:17:57,200 --> 00:18:00,200 Speaker 7: to the need for selectivity. Not all that glitters is gold. 328 00:18:00,240 --> 00:18:03,480 Speaker 7: These companies do vary in their financials. We think at large, 329 00:18:03,800 --> 00:18:06,760 Speaker 7: the fundamentals of this market and of the companies you know, 330 00:18:06,840 --> 00:18:10,960 Speaker 7: spending the most and raising issuance in various markets is 331 00:18:11,000 --> 00:18:13,880 Speaker 7: still robust, but you want to make sure in your 332 00:18:13,920 --> 00:18:17,960 Speaker 7: portfolios that you have some proper diversification, just given that 333 00:18:18,080 --> 00:18:20,760 Speaker 7: you know, the leverage and the economics around this AI 334 00:18:20,800 --> 00:18:22,840 Speaker 7: wave are definitely beginning to diverge. 335 00:18:23,720 --> 00:18:26,320 Speaker 2: Stephanie, we've just got thirty seconds. But how closely are 336 00:18:26,359 --> 00:18:29,760 Speaker 2: you tracking private markets because there are competing stories right 337 00:18:29,840 --> 00:18:32,760 Speaker 2: There is also the staying private for longer because you 338 00:18:32,840 --> 00:18:34,840 Speaker 2: can like, how much do you track that? 339 00:18:35,480 --> 00:18:37,639 Speaker 7: Yeah, well, we think that that theme is still very 340 00:18:37,760 --> 00:18:40,040 Speaker 7: much alive and well, I mean, you are going to 341 00:18:40,040 --> 00:18:43,720 Speaker 7: get likely some big, high profile AI names, but the 342 00:18:43,760 --> 00:18:47,000 Speaker 7: IPO markets are still going to be very selective. And 343 00:18:47,280 --> 00:18:50,960 Speaker 7: a lot of the companies building in this AI race, 344 00:18:51,000 --> 00:18:55,800 Speaker 7: whether it is robotics or even semis, but also particularly 345 00:18:55,880 --> 00:19:00,560 Speaker 7: the application layer and building AI for healthcare, AI for logistics, 346 00:19:00,920 --> 00:19:03,240 Speaker 7: all of those companies are in private markets right now. 347 00:19:03,480 --> 00:19:05,760 Speaker 7: And of course not all that glitters is gold. I mean, 348 00:19:05,880 --> 00:19:08,800 Speaker 7: market returns there can be quite volatile too, But when 349 00:19:08,840 --> 00:19:11,439 Speaker 7: these companies eventually reach public markets, a lot of that 350 00:19:11,480 --> 00:19:14,040 Speaker 7: growth has already happened, and that is a reason for 351 00:19:14,080 --> 00:19:16,840 Speaker 7: investors to make sure they're looking beyond just public markets 352 00:19:16,840 --> 00:19:17,760 Speaker 7: and their portfolios. 353 00:19:18,280 --> 00:19:21,960 Speaker 2: Stephanie eliagro JP Organizet Management. Thank you very much. A 354 00:19:22,000 --> 00:19:24,439 Speaker 2: quick update on the Anthropic story. A US official has 355 00:19:24,560 --> 00:19:27,680 Speaker 2: just confirmed that unpropping executives will be meeting at the 356 00:19:27,680 --> 00:19:31,000 Speaker 2: Commerce Department today. Will follow the story as it develops 357 00:19:31,000 --> 00:19:33,359 Speaker 2: and bring you the latest. Now coming up, Fox has 358 00:19:33,400 --> 00:19:36,240 Speaker 2: agreed to buy device and streaming company Roku in a 359 00:19:36,320 --> 00:19:38,600 Speaker 2: deal that would create a new TV juggernaut. We've got 360 00:19:38,640 --> 00:19:45,320 Speaker 2: those details. Next, this is Bloomberg Tech. Fox has agreed 361 00:19:45,359 --> 00:19:47,840 Speaker 2: by Roku in a deal that values the streamer at 362 00:19:47,840 --> 00:19:50,760 Speaker 2: twenty two billion dollars including debt. The tie up will 363 00:19:50,760 --> 00:19:53,960 Speaker 2: create the third largest player in the US television market 364 00:19:53,960 --> 00:19:57,560 Speaker 2: by viewership. Bloomberg Deals. Michelle Davis joins us with the deats. 365 00:19:58,160 --> 00:20:00,920 Speaker 2: Interesting one. I've got a Roku TV somewhere in my house. 366 00:20:01,240 --> 00:20:04,399 Speaker 2: What's this about. What's the strategic play from Fox? Yeah, 367 00:20:04,440 --> 00:20:04,840 Speaker 2: this is a. 368 00:20:04,760 --> 00:20:08,640 Speaker 8: Really interesting deal because Roku is seen as kind of 369 00:20:08,680 --> 00:20:12,280 Speaker 8: the last big aggregator out there that everyone was wondering 370 00:20:12,359 --> 00:20:15,760 Speaker 8: when it would get taken over. The founder for years 371 00:20:16,600 --> 00:20:19,040 Speaker 8: has controlled the company and there's always a question as 372 00:20:19,040 --> 00:20:21,840 Speaker 8: to does it make more sense to aggregate content and 373 00:20:21,880 --> 00:20:24,679 Speaker 8: be the platform on top of which you know, content streams, 374 00:20:24,760 --> 00:20:26,760 Speaker 8: or does it make sense to be a really big 375 00:20:26,800 --> 00:20:29,880 Speaker 8: streamer that generates most of its revenue that way? And 376 00:20:30,480 --> 00:20:33,440 Speaker 8: by buying Roku, Fox is basically saying we want to 377 00:20:33,440 --> 00:20:36,479 Speaker 8: make sure we are that neutral third party platform. Roku 378 00:20:36,800 --> 00:20:39,399 Speaker 8: makes most of its money on advertising, which is an 379 00:20:39,440 --> 00:20:42,760 Speaker 8: area that Fox had kind of fallen behind its competitors 380 00:20:42,800 --> 00:20:46,600 Speaker 8: on in this you know, new streaming central environment, and 381 00:20:46,720 --> 00:20:49,119 Speaker 8: so that's a lot of what the value is going 382 00:20:49,200 --> 00:20:51,800 Speaker 8: to be coming from here for them twenty. 383 00:20:51,640 --> 00:20:54,560 Speaker 2: First century folks down significantly take me into the terms 384 00:20:54,640 --> 00:20:55,960 Speaker 2: like where we're at in the process. 385 00:20:56,480 --> 00:21:00,040 Speaker 8: So the deal, it's a cash and stock deal that 386 00:21:00,320 --> 00:21:02,800 Speaker 8: the shares are down on the news of course, because 387 00:21:02,840 --> 00:21:05,480 Speaker 8: you know, investors don't always love to see a stock deal. 388 00:21:05,920 --> 00:21:08,199 Speaker 8: At this point, the deal values rocud about twenty two 389 00:21:08,240 --> 00:21:11,720 Speaker 8: billion dollars enterprise value, so that's including debt. Of course, 390 00:21:11,760 --> 00:21:16,280 Speaker 8: if Fox shares continue to decline up until the deal closes, 391 00:21:16,560 --> 00:21:19,479 Speaker 8: that will mean that some of the overall value for 392 00:21:19,600 --> 00:21:22,000 Speaker 8: the deal will be a little bit lower than twenty 393 00:21:22,000 --> 00:21:24,640 Speaker 8: two billion dollars. But this really caps off a multi 394 00:21:24,800 --> 00:21:30,720 Speaker 8: month process where Roku had been exploring strategic alternatives. The founder, 395 00:21:30,760 --> 00:21:32,439 Speaker 8: Anthony Wood as if we were just talking about had 396 00:21:32,480 --> 00:21:36,080 Speaker 8: been thinking about, you know, what's next, succession planning and 397 00:21:36,160 --> 00:21:39,320 Speaker 8: Roku and Fox actually have a long history. Fox had 398 00:21:39,359 --> 00:21:43,000 Speaker 8: invested in Roku years ago before the IPO. They sold 399 00:21:43,000 --> 00:21:45,320 Speaker 8: out of their stake at the time that they bought 400 00:21:45,480 --> 00:21:49,080 Speaker 8: to be but by buying Roku, they're kind of the 401 00:21:49,080 --> 00:21:51,800 Speaker 8: two companies are coming back together. The founder will have 402 00:21:51,920 --> 00:21:55,360 Speaker 8: a role on the board of the combined company. 403 00:21:56,160 --> 00:21:58,680 Speaker 2: In books, Michelle Davis, thank you very much. In other 404 00:21:58,800 --> 00:22:00,920 Speaker 2: m and A and US sales Force has agreed to 405 00:22:00,960 --> 00:22:04,159 Speaker 2: by aifirm, Finn for three point six billion dollars. Finn's 406 00:22:04,200 --> 00:22:09,119 Speaker 2: customer service tools will complement Salesforce's agentic capabilities. According to 407 00:22:09,160 --> 00:22:11,600 Speaker 2: the company, AI is a key part of the software 408 00:22:11,600 --> 00:22:15,480 Speaker 2: maker's growth strategy as it works to boost investor confidence. 409 00:22:15,600 --> 00:22:19,000 Speaker 2: Salesforce shares upper percentage point right. M and A has 410 00:22:19,040 --> 00:22:21,679 Speaker 2: been at the heart of the Salesforce story for a 411 00:22:21,680 --> 00:22:24,560 Speaker 2: little while. Now we'll keep tracking it. Coming up. City's 412 00:22:24,560 --> 00:22:27,840 Speaker 2: head of pan Asia Internet Research, Alisha Yapp, joined us 413 00:22:27,840 --> 00:22:31,119 Speaker 2: to break down where China's tech market stands in a 414 00:22:31,160 --> 00:22:35,720 Speaker 2: big global race. It is halftime. People will be right back. 415 00:22:36,359 --> 00:22:47,520 Speaker 2: This is Bloomberg Tech Welcome back to Bloomberg Tech. This 416 00:22:47,560 --> 00:22:49,639 Speaker 2: is what the technology sector looks like. There is a 417 00:22:49,640 --> 00:22:53,200 Speaker 2: big focus on the outcome of what's happening between the 418 00:22:53,320 --> 00:22:55,959 Speaker 2: United States in Iran and some sort of deal. We'll 419 00:22:55,960 --> 00:22:57,919 Speaker 2: get at the details on that in just a moment. 420 00:22:58,040 --> 00:23:01,320 Speaker 2: SpaceX is trading for just the second day as a 421 00:23:01,320 --> 00:23:04,440 Speaker 2: public company, up eight percent, one hundred and seventy three 422 00:23:04,480 --> 00:23:07,480 Speaker 2: dollars fifty five cents a share and rising. We'll keep 423 00:23:07,520 --> 00:23:09,720 Speaker 2: tracking that story. There's a lot of factors going into 424 00:23:09,760 --> 00:23:13,400 Speaker 2: the biggest IPO in history. Let's get out to Geneva. 425 00:23:13,440 --> 00:23:17,159 Speaker 2: Global leaders are arriving in France for the G seven summit, 426 00:23:17,320 --> 00:23:22,040 Speaker 2: and this time the usual geoplogical guest list has tech additions. 427 00:23:22,080 --> 00:23:26,639 Speaker 2: The chief executives of major AI companies Open Ai Andthropic 428 00:23:26,920 --> 00:23:30,080 Speaker 2: Google are all slated to attend the gathering commes just 429 00:23:30,160 --> 00:23:34,480 Speaker 2: days after an unprecedented White House order forced Anthropic to 430 00:23:34,560 --> 00:23:38,399 Speaker 2: block foreign access to its most advanced AI platforms. But 431 00:23:38,480 --> 00:23:41,040 Speaker 2: first on the agenda the US and Iran. The two 432 00:23:41,080 --> 00:23:44,520 Speaker 2: countries reached an interim deal to reopen the Strait of 433 00:23:44,560 --> 00:23:49,680 Speaker 2: Hormuz bliomers. Tyler Kendall joins US Now live from Geneva Tyler. 434 00:23:52,600 --> 00:23:54,479 Speaker 9: Hey, and well, at this point, we're waiting for some 435 00:23:54,560 --> 00:23:57,320 Speaker 9: firmer details from either side about what is in this 436 00:23:57,440 --> 00:24:00,480 Speaker 9: potential memorandum of understanding that we did here from the 437 00:24:00,520 --> 00:24:03,520 Speaker 9: Vice President Jadie Vanced earlier today, who said that they 438 00:24:03,560 --> 00:24:06,320 Speaker 9: are going to try to get that text out before 439 00:24:06,400 --> 00:24:11,120 Speaker 9: this potential signing ceremony happening here in Switzerland on Friday. 440 00:24:11,320 --> 00:24:13,679 Speaker 9: So far, we know that the main pillars of the 441 00:24:13,760 --> 00:24:15,800 Speaker 9: deal are what we've been tracking for weeks, which of 442 00:24:15,800 --> 00:24:19,120 Speaker 9: course includes reopening the strait of her moves. But there 443 00:24:19,160 --> 00:24:22,199 Speaker 9: is still some big questions here about how Iran feels 444 00:24:22,280 --> 00:24:25,880 Speaker 9: its role is going to be connected to this going forward, 445 00:24:25,960 --> 00:24:29,400 Speaker 9: after Iranian state media earlier today seem to indicate that 446 00:24:29,440 --> 00:24:32,920 Speaker 9: Iran wants to continue to impose fees and have some 447 00:24:33,000 --> 00:24:36,320 Speaker 9: sort of grip over the street even after a deal 448 00:24:36,480 --> 00:24:39,240 Speaker 9: is signed. We also have questions about what economic relief 449 00:24:39,600 --> 00:24:42,560 Speaker 9: will look like and ultimately what sort of curves Iran 450 00:24:42,640 --> 00:24:45,399 Speaker 9: will agree to in this deal. Though it is our 451 00:24:45,480 --> 00:24:47,960 Speaker 9: understanding that this is going to be a phased agreement 452 00:24:48,000 --> 00:24:51,439 Speaker 9: where those negotiations come later, and while the conflict in 453 00:24:51,480 --> 00:24:55,560 Speaker 9: Iran is dominating the conversation already here at the G seven. 454 00:24:55,640 --> 00:24:58,680 Speaker 9: As you mentioned, President Trump has some really important meetings 455 00:24:58,960 --> 00:25:01,320 Speaker 9: scheduled for later in the week on Wednesday, where we 456 00:25:01,359 --> 00:25:03,440 Speaker 9: will see this base to based meeting with G seven 457 00:25:03,560 --> 00:25:06,800 Speaker 9: leaders and these big tech CEOs, and that meeting is 458 00:25:06,840 --> 00:25:09,399 Speaker 9: going to be just as much about the potential opportunities 459 00:25:09,440 --> 00:25:12,320 Speaker 9: around AI as well as those potential risks because ed 460 00:25:12,400 --> 00:25:16,119 Speaker 9: Bloomberg News has now confirmed that senior staffers for Anthropic 461 00:25:16,240 --> 00:25:19,159 Speaker 9: are meeting in Washington today with members of the Commerce 462 00:25:19,200 --> 00:25:22,800 Speaker 9: Department after this unprecedented move by the administration to threaten 463 00:25:22,920 --> 00:25:23,920 Speaker 9: that export band. 464 00:25:24,880 --> 00:25:27,880 Speaker 2: Grin Books, Tyler Kendall, n Geneva, thank you very much. 465 00:25:28,280 --> 00:25:31,520 Speaker 2: There's also a massive ways of news across China's tech 466 00:25:31,600 --> 00:25:35,639 Speaker 2: landscape today. Social media powerhouse Jaohongshu is reportedly gearing up 467 00:25:35,680 --> 00:25:39,040 Speaker 2: for a blockbuster Hong Kong Ipo this month, while shares 468 00:25:39,040 --> 00:25:42,520 Speaker 2: of AI model maker GPU's skyrocketed following a major price 469 00:25:42,600 --> 00:25:46,000 Speaker 2: target hike from JP Morgan. But despite all this heavy 470 00:25:46,080 --> 00:25:50,400 Speaker 2: corporate activity, City Group warns that global investors are still 471 00:25:50,440 --> 00:25:54,720 Speaker 2: treating China tech as quote a source of funds, with 472 00:25:54,800 --> 00:25:58,720 Speaker 2: Wall Street dumping local stocks to fund the global AI 473 00:25:58,800 --> 00:26:02,200 Speaker 2: hardware tradeing US now as Alicia, Yeah, head of Pan 474 00:26:02,280 --> 00:26:05,280 Speaker 2: Asia Internet Research at City Group, And let's start with 475 00:26:05,320 --> 00:26:08,080 Speaker 2: that thesis. I find it so interesting. There's a big 476 00:26:08,160 --> 00:26:14,200 Speaker 2: distinction between newsflow and activity China's tech sector and deployment 477 00:26:14,240 --> 00:26:18,199 Speaker 2: of capital or finding capital to deploy elsewhere. Explain what 478 00:26:18,280 --> 00:26:18,919 Speaker 2: you're arguing. 479 00:26:19,840 --> 00:26:22,880 Speaker 10: Yeah, So, I think the reason why we mentioned it 480 00:26:22,920 --> 00:26:26,760 Speaker 10: is because I think the investor interests, especially for those 481 00:26:27,359 --> 00:26:30,840 Speaker 10: you know from the US and Europe so outside of Asia, 482 00:26:30,920 --> 00:26:34,119 Speaker 10: and even for those that we speak to in Asia 483 00:26:34,160 --> 00:26:38,080 Speaker 10: they're based in Asia. The there's so, I mean, people 484 00:26:38,160 --> 00:26:41,399 Speaker 10: have interests, but then somehow we know that they've been 485 00:26:41,560 --> 00:26:47,840 Speaker 10: underway uh China Internet specifically or even too many that 486 00:26:48,119 --> 00:26:51,080 Speaker 10: you know investors that share with us, they're actually uh 487 00:26:51,359 --> 00:26:54,800 Speaker 10: using you know, internet as a source of fun obviously 488 00:26:54,840 --> 00:26:57,760 Speaker 10: to to find the hardware trade in Korea, in Taiwan 489 00:26:57,800 --> 00:27:03,320 Speaker 10: and Japan, partly because I despite Internet, for example, Ali 490 00:27:03,400 --> 00:27:07,320 Speaker 10: Baba by doing their heart full stack of ai uh. 491 00:27:07,320 --> 00:27:10,240 Speaker 10: But then because of their core business are not really 492 00:27:10,280 --> 00:27:12,760 Speaker 10: doing as well are you know, they've been you know, 493 00:27:13,200 --> 00:27:16,720 Speaker 10: investing aggressively on food delivery for example, they've been dragging 494 00:27:16,760 --> 00:27:19,320 Speaker 10: the profit. So a lot of investors give us uh, 495 00:27:19,440 --> 00:27:22,280 Speaker 10: you know, the feedback is that why should I border 496 00:27:22,840 --> 00:27:25,760 Speaker 10: so much of that while I have to deal with 497 00:27:26,320 --> 00:27:29,119 Speaker 10: the core business is not a pure play and I 498 00:27:29,200 --> 00:27:32,000 Speaker 10: mus as well just play directly into some of the 499 00:27:32,040 --> 00:27:35,720 Speaker 10: hardware ings. So that's why this has been you know, happening. 500 00:27:35,800 --> 00:27:38,320 Speaker 10: And we've seen the last few weeks, I mean sorry, 501 00:27:38,320 --> 00:27:41,520 Speaker 10: the last few months. You know, China internet sector has 502 00:27:41,560 --> 00:27:46,200 Speaker 10: not been performing despite the valuation is actually getting quite attractive. 503 00:27:46,280 --> 00:27:48,679 Speaker 2: Right, Alicia. The story of the first half of this 504 00:27:48,840 --> 00:27:53,560 Speaker 2: year was Chinese internet giants and frontier labs moving very 505 00:27:53,640 --> 00:27:57,840 Speaker 2: quickly to try to close the gap with American technology companies. 506 00:27:58,240 --> 00:28:02,320 Speaker 2: Does your research show that that gap is closing, Yes. 507 00:28:02,119 --> 00:28:06,000 Speaker 10: I think indeed, especially for example, take you know Ali 508 00:28:06,040 --> 00:28:10,320 Speaker 10: Baba Kuan model, you know since three point five and 509 00:28:10,359 --> 00:28:12,680 Speaker 10: then three point six, three point seven. Now we're actually 510 00:28:12,720 --> 00:28:15,760 Speaker 10: seeing there's a huge step up in terms of the 511 00:28:15,840 --> 00:28:19,800 Speaker 10: model intelligence. And then same thing for you know the 512 00:28:19,840 --> 00:28:23,280 Speaker 10: independent lab for example Minimax they recover and also I 513 00:28:23,280 --> 00:28:27,240 Speaker 10: think ten Cent the latest model are and three that 514 00:28:27,320 --> 00:28:30,160 Speaker 10: they released at the end of April. This are also 515 00:28:30,520 --> 00:28:33,760 Speaker 10: increasingly closing the gap with the frontier model. 516 00:28:35,200 --> 00:28:38,480 Speaker 2: The story and the data that we track in the 517 00:28:38,600 --> 00:28:42,360 Speaker 2: US is capital expenditures on the hyperscale side, and then 518 00:28:42,520 --> 00:28:45,960 Speaker 2: how real is the agentic AI deployment on the enterprise side. 519 00:28:46,720 --> 00:28:49,840 Speaker 2: Take that story to China, how is that playing out 520 00:28:49,880 --> 00:28:51,120 Speaker 2: in that market? 521 00:28:52,240 --> 00:28:55,000 Speaker 10: So I think from the Kapex point of view, and 522 00:28:55,120 --> 00:28:58,040 Speaker 10: think many of us aware that I think China side 523 00:28:58,240 --> 00:29:00,800 Speaker 10: is not spending as you know, as high as what 524 00:29:00,880 --> 00:29:04,840 Speaker 10: the US hyperskill is indicating. Partly, I think there's a 525 00:29:04,840 --> 00:29:08,120 Speaker 10: lot of cost components. It's cheaper right in China. Partly 526 00:29:08,160 --> 00:29:11,480 Speaker 10: I think there's also some of the assisting capacities that's 527 00:29:11,520 --> 00:29:14,640 Speaker 10: already built up the last few years is actually able 528 00:29:14,680 --> 00:29:17,360 Speaker 10: to support that. And then in terms of the agendic 529 00:29:17,600 --> 00:29:21,160 Speaker 10: right the future agenda workflow, we think that you know, 530 00:29:21,560 --> 00:29:24,120 Speaker 10: you know, Ali Baba's trying in terms of the coin 531 00:29:24,480 --> 00:29:28,040 Speaker 10: versus the assisting you know application for e somebod the 532 00:29:28,080 --> 00:29:30,960 Speaker 10: Talbo or the AMAP. But then I think over time, 533 00:29:31,120 --> 00:29:34,240 Speaker 10: we do hope that Tencent at the end of the 534 00:29:34,320 --> 00:29:37,959 Speaker 10: day could actally come up with their AI agents insidech 535 00:29:37,960 --> 00:29:40,640 Speaker 10: Ad which they are obviously testing on it, and then 536 00:29:40,680 --> 00:29:44,240 Speaker 10: with the next versions of the model if it's intelligent enough, 537 00:29:44,400 --> 00:29:46,200 Speaker 10: because I think they're going to train a lot of 538 00:29:46,240 --> 00:29:49,680 Speaker 10: their proprietary data into the next versions of the model 539 00:29:49,920 --> 00:29:52,960 Speaker 10: and then hopefully that could actually you know, ensure the 540 00:29:53,040 --> 00:29:56,360 Speaker 10: AI agents also working within the Vita ecosystem. 541 00:29:56,880 --> 00:30:00,000 Speaker 2: Another big story overnight from China was a group based 542 00:30:00,320 --> 00:30:03,880 Speaker 2: planning to overhaul the Alipaid super app, but to bring 543 00:30:03,920 --> 00:30:07,400 Speaker 2: in an agentic AI user face you brilliantly set out 544 00:30:07,480 --> 00:30:10,560 Speaker 2: kind of what's happened one aged twenty twenty six, what 545 00:30:10,640 --> 00:30:13,600 Speaker 2: does city think happens in China's tech sector in the 546 00:30:13,640 --> 00:30:14,640 Speaker 2: second half of this year. 547 00:30:16,120 --> 00:30:19,320 Speaker 10: So we hope that for example, you know, company like 548 00:30:19,400 --> 00:30:22,160 Speaker 10: ten Cent or even you know and to someone sense 549 00:30:22,240 --> 00:30:25,120 Speaker 10: some of this super app that if they're actually able 550 00:30:25,240 --> 00:30:29,560 Speaker 10: to bring up the agendic workflow working within the super app, 551 00:30:29,960 --> 00:30:32,760 Speaker 10: which allow a lot of the enterprise that have mini 552 00:30:32,840 --> 00:30:37,640 Speaker 10: programs for example, they're able to use agent within that ecosystem, 553 00:30:38,120 --> 00:30:41,280 Speaker 10: then most of these enterprises would be able to stay 554 00:30:41,320 --> 00:30:44,240 Speaker 10: within that ego system in a way. Then hopefully you know, 555 00:30:44,320 --> 00:30:47,000 Speaker 10: things could take off, especially for the super app environment, 556 00:30:47,040 --> 00:30:51,080 Speaker 10: that they can become the AI power ecosystem that we 557 00:30:51,160 --> 00:30:54,360 Speaker 10: are looking for, and hopefully that could help to re 558 00:30:54,560 --> 00:30:58,160 Speaker 10: rate you know, the leaders like Tencent or Ali Baba. 559 00:30:58,240 --> 00:31:00,960 Speaker 10: Then in that sense to help the entire sector to 560 00:31:01,040 --> 00:31:04,560 Speaker 10: get the fund flow rotation back. Yeah. 561 00:31:04,720 --> 00:31:07,720 Speaker 2: Alicia Yapp, head of pan Asia Internet research at City Group, 562 00:31:07,760 --> 00:31:10,520 Speaker 2: big focus on China tech right now, appreciated. Thank you. 563 00:31:10,520 --> 00:31:14,840 Speaker 2: Now coming up, should we be sticking data centers in 564 00:31:14,920 --> 00:31:19,080 Speaker 2: Earth orbit or should they be under the ocean. We're 565 00:31:19,080 --> 00:31:21,080 Speaker 2: going to speak with Mike Schrope for a founding partner 566 00:31:21,360 --> 00:31:25,760 Speaker 2: of gigasale caps. He might recognize the name former meta cto. 567 00:31:25,880 --> 00:31:28,400 Speaker 2: He has a thesis. This is what the Nazak one 568 00:31:28,480 --> 00:31:32,440 Speaker 2: hundred looks like. We're back nearer toward session highs and 569 00:31:32,480 --> 00:31:36,280 Speaker 2: again of almost three percent, a lot of momentum. The 570 00:31:36,280 --> 00:31:38,480 Speaker 2: index is near record high. Is a big part of 571 00:31:38,520 --> 00:31:41,240 Speaker 2: it is kind of feel good from the SpaceX ipo Friday, 572 00:31:41,440 --> 00:31:44,040 Speaker 2: but also of course what was outlined by Tyler from 573 00:31:44,120 --> 00:31:47,960 Speaker 2: Geneva about a US Iran deal. This is blimberg tech 574 00:31:52,920 --> 00:31:55,800 Speaker 2: check back in on SpaceX shares hire second day of 575 00:31:55,840 --> 00:32:00,240 Speaker 2: trading after a blockbuster public debut Friday. That performs help 576 00:32:00,280 --> 00:32:03,400 Speaker 2: booth confidence in a broader AI rally and it marks 577 00:32:03,400 --> 00:32:06,600 Speaker 2: the start of a shift from two decades of shares 578 00:32:06,680 --> 00:32:10,360 Speaker 2: disappearing from public markets to this new era of stock 579 00:32:10,400 --> 00:32:13,560 Speaker 2: abundance as the likes of Open Ai and Anthropic also 580 00:32:13,600 --> 00:32:17,960 Speaker 2: prepare for IPOs and other companies sell new shares. Bloombos Common, 581 00:32:18,000 --> 00:32:21,080 Speaker 2: Rhinokey has been all over it, and we expect in 582 00:32:21,120 --> 00:32:24,000 Speaker 2: the next days, weeks, and months a bit of volatility 583 00:32:24,040 --> 00:32:26,480 Speaker 2: should we start, would like to hear and now I 584 00:32:26,520 --> 00:32:28,680 Speaker 2: hope you had some of a weekend to digest what 585 00:32:28,800 --> 00:32:32,360 Speaker 2: happened Friday. But right now that the net takeaway is 586 00:32:32,400 --> 00:32:34,000 Speaker 2: what cormen about this IPO. 587 00:32:34,720 --> 00:32:37,680 Speaker 11: The net takeaway is very positive. We see shares up 588 00:32:37,760 --> 00:32:39,880 Speaker 11: a second day here. It was up about eight percent 589 00:32:40,000 --> 00:32:43,320 Speaker 11: the last time I looked at my terminal, extending gains 590 00:32:43,360 --> 00:32:46,240 Speaker 11: from Friday's first day of trading. And this gives the 591 00:32:46,280 --> 00:32:49,320 Speaker 11: company a valuation of about two point two trillion dollars. 592 00:32:49,360 --> 00:32:52,720 Speaker 11: So it's already launched itself into one of the top companies, 593 00:32:52,760 --> 00:32:55,400 Speaker 11: the one of the biggest companies in the stock market. 594 00:32:55,560 --> 00:32:57,880 Speaker 2: So the takeaway was really positive. 595 00:32:57,920 --> 00:33:01,320 Speaker 11: You know, investors into the IPOH there was solid trading, 596 00:33:01,400 --> 00:33:06,040 Speaker 11: nothing went horribly wrong, and that's kind of given investors 597 00:33:06,200 --> 00:33:08,960 Speaker 11: the ability, one just to be very enthusiastic about the 598 00:33:09,320 --> 00:33:12,560 Speaker 11: AI trade. That's supported by what happened over the weekend 599 00:33:12,680 --> 00:33:16,200 Speaker 11: a potential piece deal with Iran, and investors can hopefully 600 00:33:16,240 --> 00:33:19,080 Speaker 11: start to shift their focus towards a FED meeting on Wednesday. 601 00:33:20,040 --> 00:33:21,560 Speaker 2: Com and you and the team are out with the 602 00:33:21,560 --> 00:33:23,840 Speaker 2: big take. And basically there's going to be a lot 603 00:33:23,880 --> 00:33:27,320 Speaker 2: of new equity that hits the market and it's going 604 00:33:27,400 --> 00:33:30,280 Speaker 2: to change the story of like the last two decades. 605 00:33:30,720 --> 00:33:32,800 Speaker 2: Explain that to us, give us the key numbers that 606 00:33:32,800 --> 00:33:33,320 Speaker 2: we need to know. 607 00:33:34,200 --> 00:33:37,440 Speaker 11: Yeah, so this is a really huge deal. Basically, over 608 00:33:37,440 --> 00:33:41,080 Speaker 11: the last few decades, there's been you know, twelve trillion 609 00:33:41,320 --> 00:33:43,920 Speaker 11: shrink you know, equity taken out of the markets as 610 00:33:43,960 --> 00:33:46,840 Speaker 11: company buy backstock, which is a key way that they 611 00:33:46,880 --> 00:33:50,479 Speaker 11: once spend you know money down on their balance sheet 612 00:33:50,680 --> 00:33:53,800 Speaker 11: and you know, return value to shareholders. It's been a 613 00:33:53,880 --> 00:33:57,440 Speaker 11: major support of the stock market and especially these big companies. 614 00:33:57,760 --> 00:34:01,200 Speaker 11: So what's happening now. I mean, we see IPOs like 615 00:34:01,440 --> 00:34:05,880 Speaker 11: SpaceX's record IPO on Friday, as well as equity issuances 616 00:34:05,880 --> 00:34:10,840 Speaker 11: from companies like Alphabet, Meta and Oracle bringing equity back 617 00:34:10,960 --> 00:34:13,600 Speaker 11: into the market, so growing the total number of shares 618 00:34:13,640 --> 00:34:16,239 Speaker 11: that are available in a way that actually it can 619 00:34:16,280 --> 00:34:21,160 Speaker 11: be diluted for stockholders. So one point five trillion dollars 620 00:34:21,160 --> 00:34:23,480 Speaker 11: in shares coming back into the markets this year. That's 621 00:34:23,480 --> 00:34:26,319 Speaker 11: according to a JP Morgan analysis. And really what the 622 00:34:26,360 --> 00:34:28,920 Speaker 11: signal is these companies are saying, one, we can go 623 00:34:28,960 --> 00:34:34,000 Speaker 11: to the equity markets. There's investor, you know, desire demand 624 00:34:34,080 --> 00:34:36,160 Speaker 11: for our shares, and it's a good way to raise 625 00:34:36,280 --> 00:34:39,680 Speaker 11: money to spend on what we need to do to 626 00:34:39,800 --> 00:34:42,920 Speaker 11: build out more capacity, more data centers, to use AI 627 00:34:43,320 --> 00:34:45,040 Speaker 11: in a way that will hopefully be a creative to 628 00:34:45,120 --> 00:34:46,360 Speaker 11: earnings going forward. 629 00:34:47,160 --> 00:34:50,279 Speaker 2: The most common rhinicky thank you very much so. One 630 00:34:50,280 --> 00:34:53,000 Speaker 2: way that Moss games to drive revenues and profit is 631 00:34:53,040 --> 00:34:56,839 Speaker 2: to build out of all bit data centers, a complicated 632 00:34:57,320 --> 00:35:00,680 Speaker 2: and expensive ambition that our next guest says may only 633 00:35:01,000 --> 00:35:06,200 Speaker 2: be achievable by SpaceX. Instead, others should turn to submarine 634 00:35:06,280 --> 00:35:09,440 Speaker 2: data centers to save on things like energy costs. Microver 635 00:35:09,560 --> 00:35:12,160 Speaker 2: Hoped add tens of millions of square feet of cloud 636 00:35:12,160 --> 00:35:15,600 Speaker 2: computing capacity. As the former CTO at Meta, He's now 637 00:35:15,600 --> 00:35:19,040 Speaker 2: the founding partner of Gigascale Capital, a venture firm making 638 00:35:19,080 --> 00:35:21,640 Speaker 2: some early stage bets on those companies that are more 639 00:35:21,680 --> 00:35:25,719 Speaker 2: focused on climate impact. But you put yourself in one 640 00:35:25,719 --> 00:35:30,000 Speaker 2: bucket or another. I guess orbital data center or data 641 00:35:30,040 --> 00:35:34,840 Speaker 2: centers where the ocean is the playground. Present your thesis 642 00:35:34,880 --> 00:35:35,239 Speaker 2: to me. 643 00:35:36,440 --> 00:35:38,040 Speaker 12: Well, I don't think it's an either or. I think 644 00:35:38,040 --> 00:35:40,640 Speaker 12: it's a little bit of all of the above. For space, 645 00:35:41,080 --> 00:35:42,840 Speaker 12: it's very hard and expensive to get there, and I 646 00:35:42,880 --> 00:35:45,360 Speaker 12: think nobody but SpaceX can do it with their vertical 647 00:35:45,360 --> 00:35:49,439 Speaker 12: integration and launch costs at wholesale. Otherwise, economics are just tough. 648 00:35:49,719 --> 00:35:52,200 Speaker 12: You know, I can put a ton of mass in 649 00:35:52,239 --> 00:35:54,359 Speaker 12: the ocean one hundred times cheaper than I can put 650 00:35:54,360 --> 00:35:56,600 Speaker 12: it in orbit. So I think you've got a lot 651 00:35:56,600 --> 00:36:00,840 Speaker 12: of room to go in terms of using huge energy resources. 652 00:36:00,840 --> 00:36:03,080 Speaker 12: We've got ten terot lots of power in the Southern 653 00:36:03,080 --> 00:36:07,040 Speaker 12: Ocean that's completely untapped, unused. We've also got you know, 654 00:36:07,120 --> 00:36:08,960 Speaker 12: I built a lot on land, tens of millions of 655 00:36:09,000 --> 00:36:11,320 Speaker 12: square feet. I don't think we're anywhere close to done 656 00:36:11,520 --> 00:36:15,759 Speaker 12: using solar resources on land. Between solar and batteries, I 657 00:36:15,760 --> 00:36:18,160 Speaker 12: could put together a new power plant in twelve to 658 00:36:18,200 --> 00:36:21,360 Speaker 12: twenty four months, you know, in a pace that's faster 659 00:36:21,560 --> 00:36:24,480 Speaker 12: than gas or launching into orbit. So I think there's 660 00:36:24,520 --> 00:36:25,960 Speaker 12: a lot of different ways to do this. I think 661 00:36:26,000 --> 00:36:28,399 Speaker 12: SpaceX is going to do space based data centers. I'm 662 00:36:28,440 --> 00:36:31,920 Speaker 12: bullish on on land solar plus batteries plus next generation 663 00:36:32,480 --> 00:36:35,640 Speaker 12: and in the ocean with companies like Panhalasa. 664 00:36:35,760 --> 00:36:39,200 Speaker 2: I think we therefore need to kind of understand the 665 00:36:39,320 --> 00:36:42,880 Speaker 2: lasting and ripple effects of SpaceX's IPO, right, which is, 666 00:36:43,200 --> 00:36:46,000 Speaker 2: you know, the prospectus could not be more clear. Here 667 00:36:46,080 --> 00:36:49,600 Speaker 2: is the plan orbital data center. But when you're in 668 00:36:49,640 --> 00:36:53,480 Speaker 2: private markets putting bets on earlier stages, you know, if 669 00:36:53,480 --> 00:36:56,200 Speaker 2: someone comes to you with a deck and it says, 670 00:36:56,280 --> 00:36:59,200 Speaker 2: I too have a plan for orbital data center, do 671 00:36:59,239 --> 00:37:01,640 Speaker 2: you bother? So you just say, well, SpaceX has got 672 00:37:01,680 --> 00:37:05,600 Speaker 2: that covered. There isn't a need to fund this concept, 673 00:37:05,640 --> 00:37:07,799 Speaker 2: this idea, this plan elsewhere. 674 00:37:09,360 --> 00:37:11,080 Speaker 12: I mean, I'm in the business of looking at everything, 675 00:37:11,080 --> 00:37:13,319 Speaker 12: so I'll always take a meeting to understand, to see 676 00:37:13,320 --> 00:37:16,719 Speaker 12: if I've missed something, and I'll update my priors if 677 00:37:16,760 --> 00:37:19,320 Speaker 12: I see new information. Every time I've looked at the math, 678 00:37:19,400 --> 00:37:21,240 Speaker 12: the economics on space based data. 679 00:37:21,040 --> 00:37:23,360 Speaker 2: Centers, it's doable. 680 00:37:23,760 --> 00:37:27,400 Speaker 12: It just like it's extraordinarily expensive because you have to 681 00:37:27,440 --> 00:37:29,960 Speaker 12: pay for launch, you have to pay for radiation hardening, 682 00:37:30,000 --> 00:37:33,319 Speaker 12: you have to pay for radiators large enough to get 683 00:37:33,360 --> 00:37:36,000 Speaker 12: rid of the heat, because these servers produce a lot 684 00:37:36,040 --> 00:37:38,200 Speaker 12: of heat, and once you add up all those costs, 685 00:37:38,280 --> 00:37:41,000 Speaker 12: I just don't see how it's cost competitive with current 686 00:37:41,080 --> 00:37:43,640 Speaker 12: on land or current ocean. You'd have to assume a 687 00:37:43,719 --> 00:37:47,320 Speaker 12: dramatic increase in the price of building stuff on land 688 00:37:47,560 --> 00:37:50,120 Speaker 12: before you want to go into space. Again, unless you're 689 00:37:50,120 --> 00:37:52,880 Speaker 12: SpaceX and you get all of those things that wholesale costs, 690 00:37:53,040 --> 00:37:55,800 Speaker 12: in which case your costs are very different. So I'm 691 00:37:55,840 --> 00:37:59,120 Speaker 12: not wholly no, but I'm skeptical when people pitch me 692 00:37:59,160 --> 00:37:59,880 Speaker 12: on these ideas. 693 00:38:01,400 --> 00:38:04,759 Speaker 2: Mike, if we measure the economics of and we're just 694 00:38:04,800 --> 00:38:08,759 Speaker 2: showing what SpaceX presented the other night, the design the 695 00:38:08,800 --> 00:38:12,600 Speaker 2: economics of the space based data center is a dollar 696 00:38:12,640 --> 00:38:18,960 Speaker 2: per token basis, how does submarine data center compare on paper? 697 00:38:20,280 --> 00:38:21,680 Speaker 12: Well, again, I think you have to look at all 698 00:38:21,680 --> 00:38:24,719 Speaker 12: the assumptions into that, and you know, when I've done 699 00:38:24,719 --> 00:38:26,400 Speaker 12: the modeling in math, if you just sort of do 700 00:38:26,440 --> 00:38:28,840 Speaker 12: some basics on this, it is against about one hundred 701 00:38:28,880 --> 00:38:32,279 Speaker 12: times cheaper currently to put a ton of mass in 702 00:38:32,320 --> 00:38:34,319 Speaker 12: the ocean than it is to put that same ton 703 00:38:34,360 --> 00:38:37,600 Speaker 12: of mass in space. So I think you have a 704 00:38:37,680 --> 00:38:40,200 Speaker 12: huge cost advantage when you look at the ocean, you 705 00:38:40,280 --> 00:38:45,080 Speaker 12: have almost unlimited free cooling in cold seawater. You know, 706 00:38:45,120 --> 00:38:48,640 Speaker 12: in space you can radiate heat, but that requires building 707 00:38:48,719 --> 00:38:50,480 Speaker 12: radiators or rating into a vacuum. 708 00:38:50,480 --> 00:38:51,360 Speaker 2: It's expensive. 709 00:38:51,440 --> 00:38:54,160 Speaker 12: So again, I don't see how the math closes for 710 00:38:54,480 --> 00:38:58,200 Speaker 12: anyone who's but SpaceX, who again has everything at wholesale costs, 711 00:38:58,160 --> 00:39:01,520 Speaker 12: already launching large volumes of starling satellites. So I think 712 00:39:01,560 --> 00:39:04,440 Speaker 12: it's a great business for them. I'm just like not 713 00:39:04,600 --> 00:39:07,480 Speaker 12: clear on how anyone else can compete on costs. 714 00:39:07,960 --> 00:39:10,120 Speaker 2: So our starting assumption is right. Now the world is 715 00:39:10,239 --> 00:39:13,840 Speaker 2: compute constrained, right. You weren't Meta for a long time. 716 00:39:14,120 --> 00:39:16,600 Speaker 2: Meta is going to be one of the big capital 717 00:39:16,640 --> 00:39:19,880 Speaker 2: expenditure deployers. Do you see Meta being open to a 718 00:39:19,920 --> 00:39:23,240 Speaker 2: world of submarine data center as a solution or looking 719 00:39:23,280 --> 00:39:25,680 Speaker 2: at orbital data center as an alternative. 720 00:39:27,280 --> 00:39:30,200 Speaker 12: I think all of these providers, you know, will look 721 00:39:30,239 --> 00:39:34,480 Speaker 12: at these technologies when you show tokens being generated. 722 00:39:35,000 --> 00:39:36,440 Speaker 2: You know, once I'm out a terminal, I. 723 00:39:36,360 --> 00:39:39,719 Speaker 12: Can say, here's a chat experience with a live webcam 724 00:39:39,800 --> 00:39:42,799 Speaker 12: of a little booie bobbing in this southern notion, or 725 00:39:42,960 --> 00:39:45,319 Speaker 12: you know, a feed from the live satellite and saying 726 00:39:45,320 --> 00:39:46,840 Speaker 12: this is where the tokens are getting generated. 727 00:39:47,040 --> 00:39:49,280 Speaker 2: I think everyone's going to perk up and add more attention. 728 00:39:49,719 --> 00:39:52,560 Speaker 12: Until then we're talking about theory because no one's still 729 00:39:52,719 --> 00:39:53,479 Speaker 12: either of these things. 730 00:39:53,520 --> 00:39:55,440 Speaker 2: But I think that you know, at the end of the. 731 00:39:55,520 --> 00:39:59,400 Speaker 12: Day, people want reliability, they want availability, and they want 732 00:40:00,040 --> 00:40:03,880 Speaker 12: okens now. So if you can provide people compute now, 733 00:40:04,200 --> 00:40:05,680 Speaker 12: then they're open to it. 734 00:40:07,040 --> 00:40:09,920 Speaker 2: And reminders of the audience that SpaceX said the earliest 735 00:40:09,920 --> 00:40:13,360 Speaker 2: deployment of orbital data center is twenty twenty eight. Mike Trope, 736 00:40:13,480 --> 00:40:16,279 Speaker 2: founding partner at Gigascale Capital, thank you so much and 737 00:40:16,360 --> 00:40:18,839 Speaker 2: as allys continue to be on SpaceX, we bring you 738 00:40:18,880 --> 00:40:22,200 Speaker 2: some of the biggest investors in the company. Tomorrow, Sequoia 739 00:40:22,280 --> 00:40:25,160 Speaker 2: Capital partner Sean maguire joins us a conversation you don't 740 00:40:25,160 --> 00:40:27,160 Speaker 2: want to miss now coming up, we're going to discuss 741 00:40:27,200 --> 00:40:30,319 Speaker 2: the newest update on the long awaited AI capabilities of 742 00:40:30,480 --> 00:40:38,799 Speaker 2: Apple's Siri paron. This is Bloomberg Tech. Apple's Serie AI 743 00:40:39,040 --> 00:40:41,680 Speaker 2: is catching up to its chatbot competitors, pulling the tech 744 00:40:41,680 --> 00:40:44,839 Speaker 2: company out of an AI slub two years after making 745 00:40:44,880 --> 00:40:48,759 Speaker 2: promises of AI functions. Series newest features center around it's 746 00:40:48,760 --> 00:40:53,600 Speaker 2: so called personal context and on screen awareness technology. Bloombers 747 00:40:53,640 --> 00:40:57,600 Speaker 2: Mark German got the access and joins us. Now this 748 00:40:57,640 --> 00:41:00,920 Speaker 2: is so interesting. So you've had time to to use 749 00:41:01,040 --> 00:41:04,400 Speaker 2: Apple's Serie AI in a way that you can explain 750 00:41:04,640 --> 00:41:07,320 Speaker 2: where we're at with it in power on Over the weekend, 751 00:41:07,640 --> 00:41:10,279 Speaker 2: you detailed the most important parts. What what's it been like? 752 00:41:11,680 --> 00:41:15,520 Speaker 13: Yeah, Seri AI, Apple's revamped Siri. You want to call 753 00:41:15,560 --> 00:41:17,919 Speaker 13: it Serie two point zero, you can do that. Here's 754 00:41:17,960 --> 00:41:21,040 Speaker 13: what I'll say. It fulfills the promise of Siri that 755 00:41:21,080 --> 00:41:23,719 Speaker 13: the company made fifteen years ago when it debuted the 756 00:41:23,719 --> 00:41:26,279 Speaker 13: first version as part of the iPhone for us in 757 00:41:26,400 --> 00:41:27,800 Speaker 13: two thy and eleven. 758 00:41:28,800 --> 00:41:29,480 Speaker 2: It's terrific. 759 00:41:29,800 --> 00:41:34,800 Speaker 13: It's on par probably with recent versions of Chat GPT, Google, 760 00:41:34,840 --> 00:41:38,279 Speaker 13: Gemini and Claude. It's not perfect. There's a lot it 761 00:41:38,320 --> 00:41:41,200 Speaker 13: can't do. You can't do in depth research like you 762 00:41:41,239 --> 00:41:44,080 Speaker 13: can do on chat GPT. You're not going to get 763 00:41:44,640 --> 00:41:48,319 Speaker 13: major PDF summaries of very long documents. You're not going 764 00:41:48,360 --> 00:41:51,960 Speaker 13: to get tax preparation, you're not going to get Excel 765 00:41:52,120 --> 00:41:56,080 Speaker 13: and powerpoints slide decreation. But for ninety five percent of 766 00:41:56,120 --> 00:42:00,080 Speaker 13: people they use chat GPT in Gemini for web AI, search, 767 00:42:00,120 --> 00:42:04,520 Speaker 13: for historical information, for asking questions about anything. It could 768 00:42:04,520 --> 00:42:07,120 Speaker 13: do all of that, plus actually do the things that 769 00:42:07,200 --> 00:42:10,200 Speaker 13: Seria has been supposed to do, like correctly player music 770 00:42:10,320 --> 00:42:11,880 Speaker 13: or correctly navigate your places. 771 00:42:12,640 --> 00:42:15,360 Speaker 2: This is a big win for Apple, just as people 772 00:42:15,480 --> 00:42:18,080 Speaker 2: use chat GBT for the most part. Right now, the 773 00:42:18,120 --> 00:42:20,640 Speaker 2: title of the Power on Apple's new series is just 774 00:42:21,000 --> 00:42:23,480 Speaker 2: good enough to ease AI crisis if I put the 775 00:42:23,520 --> 00:42:25,400 Speaker 2: emphasis in the right place quick. 776 00:42:26,239 --> 00:42:29,200 Speaker 14: It is just good enough because it's not exactly as 777 00:42:29,239 --> 00:42:33,160 Speaker 14: good as the competition, but for ninety five percent of people, 778 00:42:33,520 --> 00:42:36,000 Speaker 14: and to meet the needs of the majority of the 779 00:42:36,040 --> 00:42:39,719 Speaker 14: Apple userbase two billion plus strong, it does that. 780 00:42:41,000 --> 00:42:43,960 Speaker 2: Bloomberg's Mark Gunman a Power on every Sunday. You've got 781 00:42:43,960 --> 00:42:45,160 Speaker 2: to read it and then Mark will be on the 782 00:42:45,200 --> 00:42:47,239 Speaker 2: show give us the details that we need. Thank you 783 00:42:47,400 --> 00:42:49,600 Speaker 2: very much. A quick check on shares of SpaceX, so 784 00:42:49,640 --> 00:42:51,719 Speaker 2: the gains in the second day of trading now up 785 00:42:51,760 --> 00:42:55,240 Speaker 2: to ten percent, giving SpaceX as it stands a market 786 00:42:55,280 --> 00:42:58,680 Speaker 2: cap of around two zero point three to two trillion 787 00:42:58,800 --> 00:43:02,080 Speaker 2: dollars just day two as a public company trading in 788 00:43:02,120 --> 00:43:05,480 Speaker 2: the public markets. But we talked throughout the hour about 789 00:43:05,719 --> 00:43:08,080 Speaker 2: the lasting effect this will have in not just the 790 00:43:08,080 --> 00:43:10,720 Speaker 2: next few weeks, for the next few months and years. 791 00:43:11,320 --> 00:43:14,520 Speaker 2: That does it for this edition of Bloomberg Tech. Those 792 00:43:14,560 --> 00:43:17,960 Speaker 2: conversations really worth recapping. There's a lot happening in capital markets, 793 00:43:18,320 --> 00:43:22,600 Speaker 2: across public equity, across debt, and it's the biggest technology 794 00:43:22,640 --> 00:43:26,040 Speaker 2: companies in the world that are dominating that news flow 795 00:43:26,160 --> 00:43:28,279 Speaker 2: and that market move. Check out the podcast. You know 796 00:43:28,320 --> 00:43:30,359 Speaker 2: where to find it on all the Bloomberg platforms, as 797 00:43:30,400 --> 00:43:35,120 Speaker 2: well as on Apple, Spotify, and iHeart. From San Francisco, 798 00:43:35,480 --> 00:43:38,120 Speaker 2: Day one, This is Bloomberg Tech