WEBVTT - Meta Funds Gas Plants to Power Mega Louisiana Data Center 0:00:02.520 --> 0:00:13.520 Bloomberg Audio Studios, podcasts, radio news. Bloomberg Tech is alive 0:00:13.560 --> 0:00:17.360 from coast to coast, with Caroline Hyde in New York 0:00:17.640 --> 0:00:23.160 and Edva Low in San Francisco. 0:00:23.440 --> 0:00:24.840 This is Bloomberg Tech coming up. 0:00:24.840 --> 0:00:28.160 The NASDAQ one hundred falls into correction territory is big 0:00:28.200 --> 0:00:32.440 tech stocks keep falling? The Iran war rattling investor confidence. 0:00:32.680 --> 0:00:35.720 Plus Meta is set to fund the construction of seven 0:00:35.800 --> 0:00:39.320 new natural gas fired energy plants to feel its Hyperian 0:00:39.400 --> 0:00:42.800 data center in Louisiana, and Anthropic is said to be 0:00:42.840 --> 0:00:45.800 looking to IPO as soon as October. This is the 0:00:45.840 --> 0:00:49.720 AI company wins a court order blocking a Trump ban 0:00:50.120 --> 0:00:54.360 on government use of its AI tool. Welcome to the program, 0:00:54.480 --> 0:00:58.760 Happy Friday, but less happy in financial markets where technology 0:00:58.800 --> 0:01:02.560 stocks have entered correction territory. Look at how the nas 0:01:02.640 --> 0:01:05.280 that one hundred is traded over the last six months 0:01:05.360 --> 0:01:07.959 or so, and you can see that as we kind 0:01:07.959 --> 0:01:10.160 of got into the beginning of twenty twenty six, we've 0:01:10.160 --> 0:01:12.119 gone from those highs down to. 0:01:12.600 --> 0:01:14.920 A bit of a trough that's. 0:01:14.720 --> 0:01:17.840 Been driven by partly the war in Iran and geopolitics, 0:01:17.880 --> 0:01:21.840 but also a very deep assessment of AI spending. Capsule 0:01:21.880 --> 0:01:25.320 expenditures and cash flows relating to AI. Case in point 0:01:25.360 --> 0:01:29.440 and a case study is Microsoft. Microsoft is on track 0:01:29.680 --> 0:01:32.039 to have its worst quarter since two thousand and eight, 0:01:32.360 --> 0:01:35.040 the worst quarter since the financial crisis, and a big 0:01:35.080 --> 0:01:38.600 part of that, including the activity we're seeing in this session, 0:01:39.040 --> 0:01:42.680 is the commitment to AI instructure, build out, the financing 0:01:42.720 --> 0:01:45.760 of it, and what's underpinning investor confidence right now. 0:01:45.880 --> 0:01:47.200 Then there's the newsflow in the moment. 0:01:47.280 --> 0:01:50.680 Shares of Meta are also down in this Friday session, 0:01:50.880 --> 0:01:52.520 and honestly, they're having a bit of a tough week, 0:01:52.560 --> 0:01:55.200 on track for the biggest drop since October. The social 0:01:55.240 --> 0:01:58.840 media giant is set to funder construction of seven new 0:01:58.920 --> 0:02:03.000 natural gas fired energy plants to fuel its Hyperion data. 0:02:02.760 --> 0:02:04.480 Center, which is in Louisiana. 0:02:04.560 --> 0:02:08.240 The project is expected to deliver five point two gigawatts 0:02:08.240 --> 0:02:12.119 of electricity, which is a lot. Bloomberg's Ridley Griffin joins 0:02:12.200 --> 0:02:15.240 us on set. This is a big project, So you know, 0:02:15.320 --> 0:02:19.480 you kind of talk about seven natural gas flowered sources 0:02:19.520 --> 0:02:22.240 of energy for a specific project, but give us the 0:02:22.280 --> 0:02:25.360 sum total of the reporting and a bit more about Hyperium. 0:02:25.520 --> 0:02:25.720 Yeah. 0:02:25.760 --> 0:02:28.880 So Hyperion is the crown jewel of Meta's data center. 0:02:28.919 --> 0:02:29.160 Fleet. 0:02:29.320 --> 0:02:33.800 They have in development more than thirty data centers. 0:02:33.440 --> 0:02:34.720 And this is the biggest. 0:02:34.760 --> 0:02:37.840 It's in rural Louisiana, And with today's announcement, we are 0:02:37.840 --> 0:02:40.880 actually going to see a total of ten natural gas 0:02:40.919 --> 0:02:45.600 plants serving this one single data center. It's a massive 0:02:45.680 --> 0:02:49.960 amount of energy from fossil fuels and it is just 0:02:50.400 --> 0:02:54.720 another push forward in this expansion of a major project. 0:02:54.840 --> 0:02:57.760 There isn't any sort of like dollar figure attached to this, 0:02:57.919 --> 0:03:00.240 you know, it's the gigawatt figure. And you know, over 0:03:00.240 --> 0:03:01.880 the course of the war in Iran in particular, we've 0:03:01.880 --> 0:03:04.440 been very focused on the movement in oil prices, but 0:03:04.560 --> 0:03:08.120 actually natural gas is much more relevant to how a 0:03:08.200 --> 0:03:11.120 data center is powered. Explain Meta's footprint here, I guess 0:03:11.200 --> 0:03:14.760 relative to the other big data center operators around the world. 0:03:14.880 --> 0:03:18.200 Yeah, I mean, this is a huge increase just from 0:03:18.200 --> 0:03:21.000 Meta's own fleet when it was five gigawats alone, that 0:03:21.040 --> 0:03:25.160 was about a thirty three x increase from the old 0:03:25.320 --> 0:03:29.800 data center size. This is much bigger than other competitors. 0:03:30.080 --> 0:03:32.520 But I want to say five gigawa is just to 0:03:32.560 --> 0:03:36.480 support the compute. This entire facility is right now expected 0:03:36.720 --> 0:03:39.840 to have seven point five gigawats powering it because it's 0:03:39.880 --> 0:03:41.600 not just the gps that need the power, but the 0:03:41.600 --> 0:03:42.320 broader facility. 0:03:42.360 --> 0:03:45.600 It's an interesting part of your reporting about Hyperia and 0:03:45.680 --> 0:03:48.800 Louisiana in particular, which is META saying, look, where are 0:03:48.960 --> 0:03:51.160 going to cover the cost of the electricity one way 0:03:51.240 --> 0:03:53.440 or the other, one reason being they don't want to 0:03:53.480 --> 0:03:56.280 pass it on to the general population of that zip code. 0:03:56.440 --> 0:03:59.120 That's been a tension in the Great AI buildout. 0:03:59.160 --> 0:04:03.960 Absolutely recently, President Donald Trump, you know, actually require that 0:04:04.240 --> 0:04:07.880 big tech companies pledge to pay those costs. This is 0:04:07.880 --> 0:04:09.880 something META has been saying for quite some time, but 0:04:09.920 --> 0:04:12.160 they don't want to ruffle feathers in a political moment 0:04:12.160 --> 0:04:14.680 that is already bringing a great amount of scrutiny to 0:04:14.680 --> 0:04:15.760 these data centers the. 0:04:15.640 --> 0:04:18.920 Most Riley Griffin, thank you very much. Wall Street's under 0:04:18.920 --> 0:04:22.040 pressure with the tech heavy Nazak one hundred slipping into 0:04:22.040 --> 0:04:24.400 correction territory. We talked about that at the top of 0:04:24.440 --> 0:04:28.800 the show. Rising oil prices, renewed inflation concerns tied to 0:04:28.960 --> 0:04:32.360 escalating tensions in Iran are weighing heavily on the sector. 0:04:32.400 --> 0:04:36.400 It's pressuring valuations and the outlook for big tech this 0:04:36.520 --> 0:04:39.080 is what markets look right now in the moment this Friday. 0:04:39.160 --> 0:04:43.640 Let's speak to Natalie Galla Ball, principal economist and director 0:04:44.200 --> 0:04:46.880 at Board. There is a lot going on in the 0:04:46.880 --> 0:04:49.919 world right now. You know, we focused on the correction 0:04:50.000 --> 0:04:52.440 in the Nasdaq one hundred as being a consequence of 0:04:52.720 --> 0:04:54.880 what's been happening around the world. Actually a lot of 0:04:54.920 --> 0:04:57.520 that still relates to what's happening in Ai. 0:04:58.200 --> 0:04:59.279 But in the here and now. 0:04:59.160 --> 0:05:02.000 This Friday, the war in Iran drives markets, and it 0:05:02.080 --> 0:05:05.479 drives the sentiment of technology investors. Where are we at 0:05:05.560 --> 0:05:07.960 in this global economy? 0:05:08.080 --> 0:05:10.479 You know, from a broad point of view, what we 0:05:10.560 --> 0:05:14.000 have to sort of acknowledge is that economic volatility is 0:05:14.080 --> 0:05:18.640 no longer episodic. It is an embedded part of the 0:05:18.720 --> 0:05:21.680 economic narrative. And we're seeing that in oil prices in 0:05:21.720 --> 0:05:26.880 real time. Monday alone, Brent crude fell ten fourteen percent 0:05:26.920 --> 0:05:29.680 in the matter of five hours. I mean it's back 0:05:29.800 --> 0:05:30.960 up as of today. 0:05:31.279 --> 0:05:31.479 Now. 0:05:31.520 --> 0:05:34.200 If you are at all in an industry that has 0:05:34.320 --> 0:05:39.600 energy sensitivity, this environment feels almost impossible to plan in. 0:05:41.200 --> 0:05:43.279 One industry that is sensitive. 0:05:43.400 --> 0:05:46.599 That Riley Griffin was just talking about it Meta committing 0:05:46.640 --> 0:05:52.000 to invest in seven LNG powered plants for one specific 0:05:52.080 --> 0:05:55.000 data center site. You've been looking at the data of 0:05:55.080 --> 0:05:57.720 how the war in Iran is impacting energ What are 0:05:57.720 --> 0:05:58.200 you seeing? 0:05:59.440 --> 0:06:02.720 Yeah, I mean think just relative to pre conflict, LNG 0:06:02.880 --> 0:06:07.280 prices have swung sixty to eighty percent. Even if part 0:06:07.279 --> 0:06:10.200 of the impact is transitory. Say we get a best 0:06:10.200 --> 0:06:13.719 case scenario and the Strait of Hormuz reopens tomorrow, we 0:06:13.760 --> 0:06:16.360 still have a structural component we have to account for. 0:06:16.720 --> 0:06:19.200 What really comes to mind is the fact that seventeen 0:06:19.240 --> 0:06:22.400 percent of Qatar's LNG production is offline for three to 0:06:22.480 --> 0:06:25.640 five years. So this really has an implication on the 0:06:25.720 --> 0:06:27.919 cost side of the balance sheet for these companies. 0:06:29.520 --> 0:06:32.440 The other side of the conflict in Iran, and something 0:06:32.480 --> 0:06:36.320 specific to that region is the production of helium and 0:06:36.480 --> 0:06:42.840 sulfur based compounds critical components or materials. 0:06:42.440 --> 0:06:43.440 In chip fabrication. 0:06:43.640 --> 0:06:48.200 Both they provide stabilization in the etching process, in the 0:06:48.200 --> 0:06:51.839 deposition process. A lot of that happens in Katar. Again, 0:06:51.960 --> 0:06:55.279 are using data that suggests some kind of long term 0:06:55.279 --> 0:06:58.520 impact or indeed, in the short term a crisis of 0:06:58.560 --> 0:06:59.839 supply in those markets. 0:07:01.240 --> 0:07:05.440 Yes, absolutely, so. Let's take helium for example, thirty four 0:07:05.520 --> 0:07:10.120 percent of global supply comes from one facility in Qatar 0:07:10.440 --> 0:07:13.840 that has been affected by the strikes. It's anticipated to 0:07:13.880 --> 0:07:17.240 be offline in some regard for four to six months. 0:07:17.640 --> 0:07:21.960 Industry estimates are that we have about three months of inventory. Right, 0:07:22.000 --> 0:07:24.360 So you do some back of the Napkin math and 0:07:24.440 --> 0:07:27.440 you realize that the question really isn't will there be 0:07:27.480 --> 0:07:30.320 an impact? There will be an impact? When will it 0:07:30.440 --> 0:07:33.360 hit very likely Q two and to what extent? That 0:07:33.440 --> 0:07:37.080 really depends on the continued path that this conflict takes 0:07:37.080 --> 0:07:41.080 and if there is future escalation or hopefully de escalation. 0:07:42.440 --> 0:07:46.000 Away from I guess the short term in the war 0:07:46.080 --> 0:07:50.760 in Iran, there probably still is a debate around the 0:07:50.800 --> 0:07:54.280 health of the global economy, which in the technology context 0:07:54.320 --> 0:07:57.920 has just been underpinned by massive commitments to spend. The 0:07:57.960 --> 0:08:01.760 impact on raw materials, labor, How close are you tracking 0:08:01.840 --> 0:08:06.000 right now the consistency of capital expenditures that relate to 0:08:06.600 --> 0:08:08.320 data center in particular. 0:08:09.600 --> 0:08:11.920 I would say, right the question we have to ask 0:08:11.960 --> 0:08:16.880 ourselves is what is being impacted by this conflict. The 0:08:16.920 --> 0:08:19.440 cost side of the balance sheet is what's being impacted 0:08:19.440 --> 0:08:22.280 The cost to power these AI data centers, for example, 0:08:22.360 --> 0:08:26.520 the cost to produce these chips. Now, the second question 0:08:26.680 --> 0:08:30.240 is well these hyperscalers continue to meet demand. The answer, 0:08:30.240 --> 0:08:32.640 in my opinion, absolutely yes. So where do we see 0:08:32.640 --> 0:08:35.440 the impact? We see the impact on margin compression that 0:08:35.520 --> 0:08:37.840 seems like almost a given. And when do we see it? 0:08:37.880 --> 0:08:40.520 We see it in Q Two, Natalie. 0:08:41.240 --> 0:08:44.120 People at least on this program have been split on 0:08:44.760 --> 0:08:49.160 drawing parallels between this economic shock that is a war 0:08:49.200 --> 0:08:51.679 in Iran and some of what we saw in the 0:08:51.720 --> 0:08:55.040 early part of the COVID era in its impacts in 0:08:55.080 --> 0:09:00.400 supply chain shipping. For example, are you seeing any similarities 0:09:00.480 --> 0:09:02.680 or parallels? Well, can you tell me why this time? 0:09:02.720 --> 0:09:05.199 For the tech sector in particular, it is different. 0:09:06.920 --> 0:09:10.080 In one way, It's different because we aren't seeing a 0:09:10.120 --> 0:09:12.360 total shutdown of our supply chain network. 0:09:12.440 --> 0:09:12.600 Right. 0:09:12.679 --> 0:09:15.640 We are seeing a very large impact in the Strait 0:09:15.640 --> 0:09:18.760 of Hormuse. We have an ongoing impact in the Red Sea. 0:09:18.920 --> 0:09:21.160 But the good news is right, we can still get 0:09:21.200 --> 0:09:21.880 goods from. 0:09:21.880 --> 0:09:22.440 A to B. 0:09:22.640 --> 0:09:25.080 The bad news is that to do that we add 0:09:25.080 --> 0:09:28.439 thirty five hundred to four thousand nautical miles and about 0:09:28.480 --> 0:09:32.199 one to two point five million dollars in excess fuel costs, 0:09:32.200 --> 0:09:35.280 depending on vessel size and so we can continue to 0:09:35.320 --> 0:09:37.520 meet demand. It's just the fact that the supply chain 0:09:38.840 --> 0:09:41.640 environment is quite stressed and we're very likely to see 0:09:41.679 --> 0:09:45.480 an elevated cost implication, whereas in COVID right even the 0:09:45.520 --> 0:09:48.720 ability to meet demand was quite questionable due to the 0:09:48.720 --> 0:09:49.720 broad based shutdown. 0:09:50.840 --> 0:09:51.880 Not a Gallagher of Board. 0:09:51.880 --> 0:09:53.440 It's great to have you back on the show, and 0:09:53.480 --> 0:09:57.080 it's difficult to look at the impact of what's happening 0:09:57.360 --> 0:10:00.920 from the technology sector to the technology sector with everything 0:10:00.960 --> 0:10:02.719 happening with the war in Iran coming up on the 0:10:02.760 --> 0:10:06.319 show and Thropic as an October IPO, we have the 0:10:06.320 --> 0:10:07.640 Bloomberg reporting next. 0:10:08.040 --> 0:10:09.480 This is Bloomberg Tech. 0:10:23.840 --> 0:10:26.719 And Thropic, the maker of the popular Claude chatbot. It 0:10:26.760 --> 0:10:29.600 is said to be looking to IPO as soon as October. 0:10:29.640 --> 0:10:32.760 That's according to sources. Bloomberg's Bailey Lipshaltz, part of the 0:10:32.800 --> 0:10:36.400 team that broke the story. It's one of the three 0:10:36.840 --> 0:10:42.200 that we've been waiting for, SpaceX, Open AI and Anthropic. 0:10:42.240 --> 0:10:44.080 But I found this interesting, like give us some of 0:10:44.080 --> 0:10:47.320 the detail what we're hearing from sources. October is an 0:10:47.320 --> 0:10:48.320 interesting timeline. 0:10:49.160 --> 0:10:50.640 October is an interesting timeline. 0:10:50.679 --> 0:10:53.079 It also would be right before midterm elections, but it 0:10:53.160 --> 0:10:56.280 certainly is one of those IPO windows that we typically 0:10:56.280 --> 0:10:59.440 see getting hit. We're looking at, you know, normally April 0:10:59.600 --> 0:11:01.839 May and then post labor to our top of mind. 0:11:02.040 --> 0:11:03.479 The thing that's top of mind. 0:11:03.200 --> 0:11:06.520 For Anthropic and Opening Eye from our sourcing is that 0:11:06.559 --> 0:11:08.079 it's a bit of a race to try to go 0:11:08.160 --> 0:11:11.480 public first. Both are ambitiously trying to push ahead and 0:11:11.520 --> 0:11:14.400 forge ahead. We've seen and we've reported a number of updates, 0:11:14.440 --> 0:11:16.800 whether it's talking to bankers, talking to lawyers. 0:11:17.000 --> 0:11:17.760 We've heard that the. 0:11:17.679 --> 0:11:20.679 Big three in Goldman, Sachs, JP, Morgan, and Morgan Stanley 0:11:20.840 --> 0:11:24.720 are closely circling both companies with the expectation that one 0:11:24.800 --> 0:11:27.520 or two of them, hopefully according to Sourcing, will be 0:11:27.800 --> 0:11:29.440 potentially going public later this year. 0:11:30.160 --> 0:11:32.800 As a team, we've been super focused on the SpaceX 0:11:32.840 --> 0:11:37.839 IPO frankly, but Mthhropic. You know, the reported numbers are big, 0:11:37.960 --> 0:11:39.200 they're worth paying attention to. 0:11:39.280 --> 0:11:40.959 What do we know about their goals here? 0:11:41.920 --> 0:11:45.480 Yeah, so the information reported that they could raise sixty 0:11:45.520 --> 0:11:47.760 billion dollars. We haven't put out a number on what 0:11:47.840 --> 0:11:50.640 they would want to raising an IPO. The big focus 0:11:50.679 --> 0:11:53.760 has been what scale can they get to and what 0:11:53.920 --> 0:11:56.560 will investors be willing to underwrite. The big topic for 0:11:56.600 --> 0:11:58.880 both Anthropic and open AI is these are companies that 0:11:58.960 --> 0:12:01.320 spend a lot of money, and so when you talk 0:12:01.400 --> 0:12:04.160 to investors, the big question is how can you solve 0:12:04.280 --> 0:12:07.440 for that need to send and what does that ultimately 0:12:07.440 --> 0:12:07.760 look like? 0:12:07.760 --> 0:12:09.200 From evaluation perspective. 0:12:09.720 --> 0:12:12.280 Tens of billions of dollars is probably easy to raise, 0:12:12.320 --> 0:12:14.640 just given some of the rounds we've seen take place 0:12:14.640 --> 0:12:18.560 in private funding. But the big question will be how 0:12:18.600 --> 0:12:21.360 can the buyside come up with that capital? If we 0:12:21.400 --> 0:12:24.480 do see SpaceX potentially raising seventy five billion and then 0:12:24.559 --> 0:12:27.479 just a number of months later another fifty. 0:12:28.080 --> 0:12:32.800 We're waiting on a SpaceX confidential filing perspectus, but this 0:12:33.120 --> 0:12:37.160 reporting that we put out yesterday that suggests they're still 0:12:37.200 --> 0:12:39.440 working on this. What's happening in the month of April, 0:12:39.520 --> 0:12:41.440 for example, Yeah. 0:12:41.280 --> 0:12:43.800 So our reporting was that they've been working to and 0:12:43.840 --> 0:12:46.640 telling investors to get ready for meetings on the other 0:12:46.679 --> 0:12:49.840 side of that Easter holiday. So April sixth, that week 0:12:49.880 --> 0:12:52.480 and the week following is when investors are being told basically, 0:12:52.480 --> 0:12:56.000 clear your schedules and be ready to interact with SpaceX management. 0:12:56.280 --> 0:12:59.360 The big thing is that confidential filing from our sourcing 0:12:59.360 --> 0:13:01.440 and our undersea and it needs to be submitted to 0:13:01.480 --> 0:13:06.120 the SEC. Before those more fullsome conversations around valuations and 0:13:06.200 --> 0:13:09.360 around ultimate IPO sides can start to take place. So 0:13:09.559 --> 0:13:12.680 the big thing is if and when that confidential filing happens. 0:13:12.679 --> 0:13:14.800 We had reported it could take place as soon as March. 0:13:14.840 --> 0:13:16.920 We're just a few days from the end of the month, 0:13:17.200 --> 0:13:19.839 and then that can tee up some of those what 0:13:19.880 --> 0:13:23.199 we call testing the waters, just formal conversations and engagement 0:13:23.400 --> 0:13:26.160 with investors, just maybe providing a bit more detail on 0:13:26.200 --> 0:13:28.320 a bit more context about some of the numbers that 0:13:28.360 --> 0:13:29.880 could have been submitted to the SEC. 0:13:30.640 --> 0:13:34.319 Bloomberg's baby Lipschaltz, Thank you very much. Now coming up, 0:13:34.760 --> 0:13:38.400 President Trump pronounces a new Council of Advisors on Science 0:13:38.440 --> 0:13:41.160 and Technology. We speak with David Sachs, who has tapped 0:13:41.240 --> 0:13:42.600 us up as co chair. 0:13:43.080 --> 0:13:45.440 That's next. This is Bloomberg Tech. 0:13:54.640 --> 0:13:58.520 President Trump has tapped tech industry titans, including Mark Zuckerberg, 0:13:58.600 --> 0:14:02.480 Larry Ellison, and Jens to a new presidential council that 0:14:02.520 --> 0:14:05.960 will focus on AI policy and other science related issues. 0:14:06.000 --> 0:14:09.520 Co sharing that council with David Sachs, whose role as 0:14:09.520 --> 0:14:12.120 White House cryptos are has now ended. You spoke with 0:14:12.160 --> 0:14:15.240 me yesterday about the group's role in getting federal AI 0:14:15.360 --> 0:14:16.120 rules passed. 0:14:16.160 --> 0:14:16.679 Take a listen. 0:14:17.840 --> 0:14:20.400 Last week we released a National AI Framework, and the 0:14:20.480 --> 0:14:23.600 idea is to create one rule book for AI in 0:14:23.760 --> 0:14:26.680 the US. The problem that we're seeing right now is 0:14:26.680 --> 0:14:30.120 that you've got fifty different states regulating this in fifty 0:14:30.160 --> 0:14:32.960 different ways, and it's creating a patchwork of regulation that's 0:14:32.960 --> 0:14:35.200 difficult for our innovators to comply with. So what the 0:14:35.200 --> 0:14:38.320 President has called for is one rule book. What we 0:14:38.360 --> 0:14:41.200 did is and I worked with Michael Kratzios at OSTP 0:14:41.600 --> 0:14:43.680 and other folks of the White House. We looked at 0:14:43.720 --> 0:14:45.400 all the different things that the states we're doing, and 0:14:45.440 --> 0:14:47.680 we try to find some common denominators, and then we 0:14:47.720 --> 0:14:50.360 published that in a set of principles that we're calling 0:14:50.400 --> 0:14:53.440 the National AI Framework, and we're calling on Congress to 0:14:53.480 --> 0:14:57.320 act on that framework. The framework contains things like child safety. 0:14:57.360 --> 0:15:00.280 That's a very salient issue right now. How do you 0:15:00.280 --> 0:15:03.640 protect kids online? So we want to take care of that. 0:15:04.320 --> 0:15:06.560 There's things like the Raye Pair Protection Pledge, which the 0:15:06.560 --> 0:15:09.360 President's already announced. We want to make sure that these 0:15:09.360 --> 0:15:12.800 new AI data centers don't increase the cost of electricity 0:15:13.120 --> 0:15:15.320 for residential consumers. At the same time, we want to 0:15:15.360 --> 0:15:17.760 make it easier for those AI companies to bring their 0:15:17.800 --> 0:15:19.840 own power. So this is I think a much better 0:15:19.840 --> 0:15:23.280 approach than effectively the ban on new data centers that 0:15:23.360 --> 0:15:27.520 you're seeing from Bernie Sanders and others. There's principles and 0:15:27.560 --> 0:15:30.920 provisions related to content creators and how do you protect 0:15:32.120 --> 0:15:36.280 their new content while allowing for AI models to be trained. 0:15:36.680 --> 0:15:40.480 So there's a lot of different areas here that we've covered, 0:15:40.760 --> 0:15:42.480 and I think you're seeing a very good reception to 0:15:42.520 --> 0:15:46.120 this from Capitol Hill. Even the criticism I thought from 0:15:46.240 --> 0:15:49.280 Democrats was pretty muted. There's some Democrats we've already reached 0:15:49.280 --> 0:15:51.040 out to us. They want to work with us to 0:15:51.080 --> 0:15:53.280 see if we can do something in a biparisan way. 0:15:53.800 --> 0:15:55.440 And so I think there's actually a very good chance 0:15:55.440 --> 0:15:58.200 that you'll see Congress now act on the streamwork and 0:15:58.200 --> 0:16:02.640 we'll get some meaningful AI legislation again, the one rule 0:16:02.680 --> 0:16:04.760 book that President Trump's talking about in the next few. 0:16:04.560 --> 0:16:06.040 Months for November. 0:16:06.120 --> 0:16:10.320 David a bipartisan AI framework before November. 0:16:10.720 --> 0:16:12.800 I think it could happen in the next few months. 0:16:12.840 --> 0:16:14.640 I mean, again, we've gotten a very good reception from 0:16:14.680 --> 0:16:17.800 Capitol Hill. This is an area where I think we're 0:16:17.920 --> 0:16:20.400 willing and happy to work with Democrats, and I think 0:16:20.400 --> 0:16:22.440 there's a lot of Democrats who would like to see 0:16:22.680 --> 0:16:25.680 a single national framework as well. I think they understand 0:16:25.720 --> 0:16:30.080 that it's not feasible or practical to have fifty states 0:16:30.120 --> 0:16:32.440 running in fifty different directions. And I think if we 0:16:32.480 --> 0:16:34.440 can work something out, I think we could get to 0:16:34.480 --> 0:16:36.640 the one rule book that the President's talking about. 0:16:37.000 --> 0:16:40.520 David, looking at the composition of the council, you and 0:16:40.600 --> 0:16:43.120 I have spoken in the past that the definition of 0:16:43.160 --> 0:16:47.520 a China Hawk, for example, and the context of exporting technology, 0:16:47.520 --> 0:16:51.600 it needs some discussion. But on the chip side, the 0:16:51.600 --> 0:16:54.240 infrastructure side, those are a group of CEOs that want 0:16:54.240 --> 0:16:57.040 to sell their technology around the world. Do you see 0:16:57.080 --> 0:17:02.400 that changing the President's current thinking on on exporting cutting 0:17:02.480 --> 0:17:05.000 edge or lead edge chips to China or more generally 0:17:05.040 --> 0:17:06.840 in different jurisdictions around the world. 0:17:08.080 --> 0:17:11.520 Well, I think this administration already has a pro export 0:17:11.960 --> 0:17:15.720 mindset because we understand that the way that American technology 0:17:15.800 --> 0:17:18.800 wins and dominates the globe is through market share. You 0:17:18.920 --> 0:17:21.560 want to have the most market share all over the world. 0:17:21.640 --> 0:17:24.760 You want the American textack to become the global standard. 0:17:25.080 --> 0:17:27.280 That's good for us economically, I think is also good 0:17:27.640 --> 0:17:30.720 in terms of national security and spreading our influence. Now, 0:17:30.720 --> 0:17:34.760 whenever you're talking about countries of concern, China, potential adversaries, 0:17:34.800 --> 0:17:37.760 you have to be more careful and needed more nuanced policy, 0:17:38.080 --> 0:17:40.320 and I think the administration has supported that. But I 0:17:40.359 --> 0:17:42.919 think that in general, we want the world to be 0:17:42.960 --> 0:17:45.840 a place where American technology wins, as it did with 0:17:45.920 --> 0:17:46.520 the Internet. 0:17:48.960 --> 0:17:51.600 That was David Sachs, co chair of the President's Counsel 0:17:51.640 --> 0:17:55.000 of Advisors on Science and Technology and partner at Craft 0:17:55.080 --> 0:17:58.679 Benches the Sex calls for clearer rules around artificial intelligence. 0:17:58.880 --> 0:18:02.000 There's also a growing push to modernize how those rules 0:18:02.040 --> 0:18:05.560 are written and implemented within governments. Joe Schidler, it's the 0:18:05.600 --> 0:18:08.960 CEO of Helios and AI platform designed to help track and 0:18:09.080 --> 0:18:12.560 navigate legislation in real time. Joe was a former advisor 0:18:12.600 --> 0:18:15.640 at the White House and US State Department under President 0:18:15.680 --> 0:18:18.520 Biden and joins US now. Just interest in your thoughts 0:18:18.520 --> 0:18:21.119 in the first instance, on what mister Sachs was saying. 0:18:21.560 --> 0:18:25.200 You know, the trajectory and timeline for a federal level 0:18:25.280 --> 0:18:26.320 set of rules on AI. 0:18:27.320 --> 0:18:29.720 Excuse to be with you, Ed, but yeah, I think 0:18:30.320 --> 0:18:33.680 what David had kind of stated yesterday makes a lot 0:18:33.680 --> 0:18:37.720 of sense. A patchwork of regulations across fifty states is 0:18:37.760 --> 0:18:42.560 not a sustainable structure in a global competitive technology race. 0:18:43.200 --> 0:18:46.880 I think as a builder and somebody who has sat 0:18:46.920 --> 0:18:50.400 inside of government, I can confidently say what we need 0:18:50.480 --> 0:18:55.360 right now as a single national framework that advances this 0:18:55.440 --> 0:18:58.240 critical technology and one of the most consequential times we're 0:18:58.280 --> 0:18:59.520 seeing in modern history. 0:19:00.440 --> 0:19:03.359 Joe, what every day Americans may be confused by is 0:19:03.920 --> 0:19:07.320 the distinction between either there not being enough regulation. In 0:19:07.359 --> 0:19:10.119 other words, we do not have a federal level framework, 0:19:10.640 --> 0:19:13.240 but there is a lot of state by state regulation. 0:19:13.320 --> 0:19:17.679 The point that mister Sachs was making, what is the 0:19:17.720 --> 0:19:22.480 real risk over regulation or under regulation in this context? 0:19:23.119 --> 0:19:26.600 Yeah, I think there is a fundamental mismatch between the 0:19:26.640 --> 0:19:31.000 pace of this technology and the rate at which we're 0:19:31.000 --> 0:19:36.280 trying to regulate this across different states, locality, cities. And 0:19:37.280 --> 0:19:40.560 the issue is this is still a very emerging technology 0:19:40.560 --> 0:19:44.359 in many respects. It's a horizontal technology that cross cuts 0:19:44.400 --> 0:19:50.040 across every sector of vertical industry. And you know, there's 0:19:50.119 --> 0:19:54.399 a question that I think is persisting throughout circles in 0:19:54.480 --> 0:19:57.280 DC in the valley of If not America. 0:19:57.240 --> 0:19:57.680 Then who. 0:19:58.720 --> 0:20:02.920 And we are at a moment right now where we're 0:20:02.960 --> 0:20:06.840 seeing massive productivity gains and key suckers of the economy 0:20:06.880 --> 0:20:11.000 to include government itself. And I think it's far too 0:20:11.040 --> 0:20:14.840 early to be taking an over regulatory approach to a 0:20:14.880 --> 0:20:16.879 technology we're still learning a great deal about. 0:20:18.359 --> 0:20:20.440 There's two specific areas I want to talk to you about. 0:20:20.520 --> 0:20:23.879 The first is the use of AI generally in national 0:20:23.920 --> 0:20:27.760 security and defense, and thropic is the biggest case study 0:20:27.800 --> 0:20:30.679 of late right And just having returned from DC, the 0:20:30.720 --> 0:20:32.800 position of many was it is not up to the 0:20:32.840 --> 0:20:37.320 companies how the government, particularly the defense or arms of 0:20:37.320 --> 0:20:39.160 government use the technology. 0:20:39.920 --> 0:20:42.160 Your viewpoint on that, please, yeah. 0:20:42.200 --> 0:20:45.200 I mean, I would say it's the prerogative of any 0:20:45.240 --> 0:20:48.760 American to express their First Amendment rights. And at the 0:20:48.800 --> 0:20:52.280 same time, we have systems to affect change. We live 0:20:52.320 --> 0:20:55.679 in a democratic, electoral society. If folks are unhappy with 0:20:56.920 --> 0:21:00.679 the position of any given administration or a government composition, 0:21:01.560 --> 0:21:05.200 there are plenty of tools at our disposal to affect change. 0:21:05.560 --> 0:21:07.960 What I would say is these technologies aren't anything new 0:21:08.040 --> 0:21:11.600 to the US military or defense circles. In general. We've 0:21:11.600 --> 0:21:17.040 been using machine learning and AI capabilities for decades, and 0:21:17.119 --> 0:21:19.399 for now it's starting to get a lot of attention 0:21:19.480 --> 0:21:24.320 in the media, particularly because of I think, adjacent stories that. 0:21:26.200 --> 0:21:27.480 Have a tendency to. 0:21:29.080 --> 0:21:32.800