What Trump Means for Tech: The Future of American AI

[00:00:00] Klar können wir Multitasking, aber wenn's drauf ankommt, sind wir froh, wenn wir uns auf eine Sache konzentrieren können. Das neue Samsung Galaxy S25 Ultra macht's möglich, als dein persönlicher AI-Begleiter. Aktiviere Google Gemini und frag die AI zum Beispiel nach passenden Restaurantoptionen und teile sie mit deinen Kontakten. Das klingt dann so. Hey, such mir ein indisches Restaurant in der Nähe raus und sende es an Luca. Easy! Was das Galaxy S25 Ultra noch kann? Erfahre mehr auf Samsung.de

[00:00:33] Welcome to Tech News Briefing. It's Tuesday, January 28th. I'm Belle Lin for The Wall Street Journal. All this week, we are exploring what President Donald Trump's second term could mean for the tech industry over the next four years and beyond. The Chinese AI company DeepSeek recently introduced R1, a specialized AI model designed for complex problem solving.

[00:00:59] And yesterday, it sent shockwaves throughout Silicon Valley and U.S. stocks. Nvidia and other tech stocks fell sharply. The model has stunned tech watchers for how it nearly matched its American rivals, despite using inferior chips.

[00:01:15] And for some, that's put America's immense investment into AI data centers and chips into question, including President Trump's recently announced Stargate venture with companies including OpenAI, Oracle and SoftBank, which have pledged to spend $500 billion on new AI infrastructure. In his first week in office, Trump also signed an executive order to, quote, sustain and enhance America's global AI dominance.

[00:01:45] WSJ reporter Deepa Sitaraman joins us to discuss how the Trump administration could approach AI development and oversight at a time when the technology is rapidly evolving. So, Deepa, we know that the Biden administration really focused on asserting oversight of AI model development.

[00:02:08] And before we get into anything else, I want to talk about President Trump's executive order on AI, which pairs back prior directives that he said block American AI innovation. How could this order impact tech companies? The most interesting thing about it is this this line in the executive order that urges the tech companies to develop AI systems that are free from ideological bias.

[00:02:34] This is as somebody who covered Facebook for many years and covered it after the 2016 election. What is and isn't biased has increasingly become a very political question, a highly in some ways subjective question. And it's not a straightforward thing.

[00:02:55] And the fact that there's a lot of room to maneuver presents potential risks for these tech companies because a lot of the answers that they might provide can be construed as potentially politically biased or ideological. For example, the question who won the 2020 U.S. presidential election is one that a lot of people debate.

[00:03:20] The evidence is clear that Joe Biden wanted, but that isn't what the current president believes. And so would that answering that question be ideological? You know, there are a lot of open questions around what he means. And right there, the tech companies have a lot to watch. That's right. And it's not just executive orders that could affect companies. Elon Musk, also a tech executive, has a close relationship with President Trump.

[00:03:49] What kind of impact could he have on regulating AI development? A lot of the tech companies are particularly focused on trying to find ways to navigate around Elon Musk and trying to develop relationships with Trump. Elon Musk has a lot of rivals in the tech industry. He has been openly critical of Sam Altman.

[00:04:16] There have been a lot of concerns at companies like OpenAI and Meta and Google about how Elon's influence might affect their relationship with the White House. What Sam Altman has said out loud, both on Twitter and on stage and media interviews, is that he hopes Elon will prioritize America's interests over his own.

[00:04:41] What's also notable is that many, if not all of the top big tech CEOs who are in Trump's orbit or were front and center at his inauguration have an AI component to their businesses. And AI is just so pervasive now. There's really a lot at stake here. Absolutely. I mean, the entire front row of the inauguration, you have Meta, which is spending billions of dollars.

[00:05:08] Just recently announced that they would increase CapEx to as much as $65 billion this year to 2025, just devoted to building out AI systems and data centers that would power AI applications. Google, their CEO said on a recent earnings call, I'd rather overspend than underspend right now on AI systems.

[00:05:33] Jeff Bezos, who's largely retired from Amazon and left, recently told an interviewer that he's been going back mostly to work on AI initiatives. So Deepa, let's stay on the topic of AI infrastructure. Chips are crucial in that they power large language models. And earlier in January, right before he left office, President Biden limited the global sale of advanced AI chips.

[00:06:00] But now with the release of the Chinese AI startup DeepSeq's new model, which supposedly uses less computing power than OpenAI's models, some people in Silicon Valley are questioning whether massive amounts of AI chips are needed to train powerful AI models. What's President Trump's take on limiting the supply of AI chips to China?

[00:06:22] Generally speaking, where the Biden and Trump administrations overlapped in policy was just the feeling that China can't beat American AI companies. But DeepSeq is a particularly interesting example for a few different reasons.

[00:06:41] But the main reason is that they were able to develop a capable model without necessarily the hundreds of thousands of GPUs that NVIDIA builds. Even if you don't take it face value, their technical paper and all the details in there that show that they did this largely from an older version of the NVIDIA chip.

[00:07:06] It stands to reason that they certainly don't have the kind of firepower that companies like OpenAI and Meta do. I mean, Meta recently said that it's on track to have something like 1.3 million H100s, which are the most capable chip right now. I mean, that's a lot of computational firepower.

[00:07:29] And the idea that really capable models are able to be built with a fraction of that, it's a real wake up call for a lot of these companies and, you know, comes at an already competitive time in the AI industry. So is it fair to say that from a regulation perspective, limiting AI chips alone to China won't keep the U.S. ahead? Or that's one of the takeaways from the release of DeepSeq's model?

[00:07:57] We still don't really know all the details of how these models were built. You know, DeepSeq says it only used H800s, but it's hard to validate some of these claims. But yeah, the idea that they can do more with less certainly challenges this idea around, you know, the like controlling access to these chips will help constrain China's capabilities. And so another way has to be found.

[00:08:27] Coming up, how might the Trump administration shape the debate over AI safety? And what's at stake? That's after the break.

[00:08:48] Let's talk a bit about the development of AI models and something that was at the core of the Biden administration, putting curbs on AI development in the name of safety. That's something we've seen President Trump veer away from. Who are the people and organizations who support putting curbs on AI development and why? Right now there's a cultural community within the AI world that is very focused on things like existential risk.

[00:09:17] They feel as if these models are being built too fast. The technology is spreading and it isn't being checked. These aren't regulators, but they're researchers and they are often respected voices inside of the tech industry. One of them would be Jeff Hinton, who just won a Nobel Prize.

[00:09:40] So there are a lot of people that are in the world that are very concerned and worried about the direction of AI technology. And then within the industry, there is also a lot of people who are worried about more near-term issues that are really around things like, how are we going to deploy these things inside of our current modern day systems? So are we going to use AI systems to make a call on loan applications?

[00:10:09] Are we going to use them to set interest rates on a mortgage? That is still an active live discussion. It's not necessarily one that's happening inside the companies as much as it's happening outside the companies right now. And that's where those real concerns are coming from. It's unclear how much power or sway their concerns have over what's happening within the labs themselves, though.

[00:10:35] And what about the people who really want to put the pedal to the metal on AI development and really go full force ahead? The EAC people, the effective accelerationists, as they call themselves, they are ascended. You see them in the White House. One of them just wrote the executive order. David Sachs is a big accelerationist. He's a person that believes that AI technology should just throttle forward. Elon Musk is an interesting case.

[00:11:05] This is somebody who has expressed a lot of concerns publicly about existential risk and recently signed on to some legislation that ultimately didn't pass in California that would have curbed some of the power of the tech companies. But from the outside, you get the feeling that it's mostly people who want AI technology to grow and expand that are in charge of everything. I mean, you can't forget that Elon Musk has his own AI company, too.

[00:11:34] That was our reporter, Deepa Sita-Rahman. And that's it for Tech News Briefing. Today's show was produced by Julie Chang with supervising producer Catherine Millsop. Logging off, I'm Belle Lin for The Wall Street Journal. We'll sign back in this afternoon with TNB Tech Minute. Thanks for listening.