NICK THOMPSON: If you look at the amount of money that is being spent, the expectation of profits that companies have from AI based on what they’re spending to buy Nvidia chips, right? And if you take that number, and then you subtract the amount of money companies are getting, there’s like a $500 billion gap. There are not enough people buying OpenAI subscriptions and using Microsoft AI products.
It’s not even clear. Like, Facebook is buying tons of Nvidia chips. Right? Tons. How are they monetizing that? Not really clear. So if companies aren’t able to make any money off of their Nvidia investments, then eventually, they stop buying as many chips.
Now, my hypothesis is that that is not going to happen, that we are going to find enormous benefits from this, but there is definitely a scenario where they’re in trouble.
BOB SAFIAN: That’s Nick Thompson, CEO of The Atlantic, and former editor-in-chief of Wired. Nick has been a longtime source of insight for me about the tech world, and as we near the end of 2024, I wanted to talk with him about where Big Tech stands right now. Together we explore the dominance of Nvidia and Apple, the cult of personality around Elon Musk, Sam Altman, and Mark Zuckerberg, and what to keep our eyes on for 2025. Nick also shares what it’s like running The Atlantic during a time of heightened media mistrust. Nick doesn’t hold anything back. So let’s get to it. I’m Bob Safian, and this is Rapid Response.
[THEME MUSIC]
SAFIAN: I’m Bob Safian, and I’m here with Nick Thompson, CEO of The Atlantic. Nick, thanks for joining us.
THOMPSON: Bob, I’m delighted to join you.
The evolution of Big Tech’s trillion-dollar club
SAFIAN: The Atlantic covers so many things, but for the purposes of today’s discussion, I want to focus on big tech.
The value of tech businesses has become extraordinary, like the trillion-dollar club. I mean, I can’t believe we’re even using that figure, and at the same time, it’s become almost commonplace. Is it warranted? Like, is the consolidation in Big Tech a good thing, a bad thing, a bubble?
THOMPSON: It’s definitely not a good thing, right? It would be better for the world if there was more competition, but the way the tech industry works, there are network effects, so there’s inevitable consolidation. And then there are all kinds of societal forces that, once you have a big, successful company, it’s easier to get regulations written in your way.
It’s easier to buy competitors. It’s easier to grow than it is for a start-up to grow. So, it is sort of a natural effect of the way capitalism works, that you’re going to end up with some giant companies. Is it a bubble? It’s interesting. If you look at the trillion-dollar club, Apple is definitely going to be there, right? In three years, right? There’s no scenario, I think, where Apple’s market cap collapses. Nvidia, on the other hand, could be worth much less. AI could turn out to be far less valuable than we think, and it could be removed from the club. The same thing could happen with Tesla. It could be outcompeted in multiple ways, right?
So, of the companies that are either in the trillion-dollar club or have reached that valuation, there’s some variability. I would say that the expected value over a five-year period is roughly the same. It’s just with Apple, you have like a higher floor, lower ceiling. With Nvidia, you have a lower floor, higher ceiling.
Nvidia’s strategic importance and challenges
SAFIAN: I mean, Nvidia has kind of gone from an almost anonymous company to like an “it” company because of its stock price soaring. Some people say it’s the most important company in the world right now based on what it is. Where are you on Nvidia?
THOMPSON: I mean, it might be the most important company in the world. No, I mean, it is certainly one of the most interesting companies in the world because you’re completely right. Like, in the last few years, they made a number of bets that were brilliant, right? They bet on GPUs. They understood where AI was going, and then they developed all of this market power by being the one supplier of AI chips, which they’ve been able to use in all kinds of canny ways to shape the industry, shape their financial fortunes. They’re masterful. Now, what will be interesting is if markets don’t develop the way Nvidia’s investors expect and the way AI enthusiasts expect, then things could get rocky, but they’re pretty smart over there.
SAFIAN: They’ve also been beneficiaries in some ways of the rest of the chip business sort of missing this turn, right? I mean, Intel seemingly pushed Pat Gelsinger out as CEO. It’s almost like the biggest threat to Nvidia is not from others in their own business, but like from China blocking them from getting the parts they need or something.
THOMPSON: I think the biggest blocker is just like there are not effective business applications with AI, right? If you look at the amount of money that is being spent, the expectation of profits that companies have from AI based on what they’re spending to buy Nvidia chips, right? And if you take that number, and then you subtract the amount of money companies are getting, there’s like a $500 billion gap, right? So they’re spending as though the total revenue is 600 billion, and they’re bringing in 100 billion, right? There’s not enough people buying OpenAI subscriptions and using Microsoft AI products.
It’s not even clear. Like, Facebook is buying tons of Nvidia chips. Right? Tons. How are they monetizing that? Not really clear. Like, they’ve built a chat product into WhatsApp. They’ve built it into Instagram. There’s some value. They’ve built Llama, which is wonderful, but they don’t really monetize Llama. So if companies aren’t able to make any money off of their Nvidia investments, then eventually they stop buying as many chips.
Now, my hypothesis is that that is not going to happen, that we are going to find enormous benefits from this, that Facebook will figure out how to monetize, as it always figures out how to monetize things, and Nvidia will find a way, but there is definitely a scenario where they’re in trouble.
The sustainable lead of OpenAI
SAFIAN: You mentioned OpenAI. They recently released their AI video model, Sora. They seem to continually kind of lead development, even though others, like Google, are spending a ton, investing in their own models. Is OpenAI still special?
THOMPSON: OpenAI made this amazing bet five years ago, right? Six years ago when they saw the promise of transformers, and they understood scaling laws, they put more and more money into compute data, and they bet in a way that nobody else did, and that gave them a large lead, right?
And the question is, can that lead be sustained even though most of the original employees who were most important to it have left, many of whom are off building competitive companies, right? Dario Amodei has built Anthropic, which is a competitive company. Ilya Sutskever is building a company that may be competitive. There’s still Sam Altman, there’s still some engineers, but it is a huge question of whether they can continue to have their special sauce. They do have a big advantage having been first to market, right? People talk about ChatGPT the way they talk about Band-Aids, right? It has this incredible brand. It has massive user adoption. They do have huge press benefits. Will they still have a technical lead in two years over Google? I don’t know. Every six months the gap between OpenAI and the rest of the AI field narrows.
SAFIAN: And Sam Altman, who you mentioned, this turnover in the OpenAI leadership team and the sort of incredible governance drama a year ago where he was pushed out and they got his way back. What’s the rap on Altman these days? Because in some places he’s almost statesman-like in the way he acts, but there’s also this kind of wariness, right?
THOMPSON: Yeah, I mean, he’s clearly, he’s clearly utterly brilliant, right? But he doesn’t have the reverence and respect that Steve Jobs had. There’s not the same level of trust in his decision-making. He doesn’t have the same loyalty. You look at the people who worked with Jobs like they would have died for him. And if you look at the people who work with Altman, right, one of them tried to fire him. He’s a very complicated guy with a very complicated reputation.
SAFIAN: And the sort of the question about what OpenAI’s status is going to be, like this move between not-for-profit or for-profit or whatever the quasi-status that the company has, just sort of complicates it because no one really quite knows what his motivation is in all this.
THOMPSON: Yeah, I mean, I think that that conflict ends with them gradually over time becoming a fully profitable company. They’re just going to become like every other company. And the idealism with which it was founded, I think a lot of it has dissipated.
“We’re building AI for humanity, we’re going to give it away for free,” but they really just run like every other company we know, many of which have good value statements, and the employees are very good-hearted people who are fully intent on helping the world, but I don’t think the period in which OpenAI was different from the rest of the companies in Silicon Valley is closing.
Idealism versus profit in Silicon Valley
SAFIAN: The whole idea of idealism in Silicon Valley, is that less real or less important than it used to be with all these companies at these enormous valuations?
THOMPSON: Silicon Valley has always been a mix of idealism, true idealism, right? You look at the founding DNA of Silicon Valley; there are true idealists who come out of the counterculture, who come out of the Berkeley free speech movement, who come from sort of the Stewart Brand, Whole Earth Catalog, open source. There’s a real desire to have an independent internet that is different from all the other structures in the world. So that strand exists. And, in fact, you can still see that strand. To a certain degree, it exists. It’s interesting. The Web3 world, to me, is both the purest idealists and some of the most grotesque grifters.
So you can see the strand of idealism, and that idealism exists at all of these big companies. It certainly exists to some degree in early Facebook, right? Where it’s about connecting because of a genuine belief that connecting people makes the world better. Now it so happens connecting people also makes you billionaires, which turns out to be very beneficial as well. Google starts with a certain amount of idealism. Over time, as idealistic employees leave, as money becomes more important, as people start to compete about the size of their yachts, the idealism dissipates. And I think you’re seeing that in AI. The folks who went into AI very early on saw, in a way that much of society didn’t see, how powerful this would be. They really cared about making sure they did it right. And then over time, some of them just got excited by the money. And then also, all these other people came in because of business opportunities, money, stock grants. The most interesting company to me on this question, the one to really watch, is Anthropic.
Because it was 100 percent founded because a group within OpenAI didn’t believe that Sam Altman and the other leaders of OpenAI were taking safety seriously enough, and that some of the idealism had dissipated. And so Anthropic was set up in a very different way, and it has continued to hold to those principles in a way that has been impressive. And so I am extremely interested in whether Anthropic goes through a big shift.
Mark Zuckerberg’s resilience and long-term vision
SAFIAN: Another complex persona you alluded to, Mark Zuckerberg, shortlisted for Time’s Person of the Year last week. You ran a WIRED cover story in 2018 with a sort of beaten and bruised Zuckerberg on the cover. I’m curious how you see Zuckerberg today, how you sort of explain why Time would consider him a Person of the Year candidate?
THOMPSON: He’s a remarkable figure. He keeps making the right long-term decisions, and his ability to do that over and over is astonishing. He does make lots of bad decisions. We were critical of, if you look at those stories that I wrote with Fred Vogelstein at WIRED, two very long investigative pieces and a number of other pieces, it was a factual analysis of just what went wrong. But there was never a sense that, like when you would talk to people, even people who are very critical of Mark Zuckerberg, even people who felt that he was motivated by the wrong things, everybody saw that he kind of figured things out.
He was able to understand where the mobile web was going before everybody else. He was able to understand the way that social architecture works. He may probably have made a big mistake on VR, but we don’t know for sure. And then with AI, it’s astonishing that Facebook somehow has been able to build the primary open source platform, which kind of defies all logic, all that you would think.
My guess is that they’re able to figure out a really smart way to make that a core part of their company.
It’s amazing because he, I don’t know how old he is, maybe early 40s, but he’s gone through a whole number of reinventions. He does some stupid billionaire stuff like the cage fight threat with Elon. There’s stuff where you’re just like, “Oh, you don’t need to do that. You don’t need to play those games,” but I’m impressed that he’s still there and still working and still making decisions. He hasn’t stepped away, hasn’t gone through some kind of midlife crisis, given up on the company, and decided he’s just going to disappear on his boat.
Elon Musk’s unexpected paths
SAFIAN: You never know where these characters are going to go in the same way that Elon Musk has become a character. He is going places that maybe you and I wouldn’t have predicted.
THOMPSON: I definitely wouldn’t have predicted where Elon Musk has gone. He’s built two companies that are incredibly important to humankind’s future. He’s seen things out in the distance that nobody else has seen.
And then, he’s become a professional shit poster in certain ways that just seem utterly beneath him and utterly toxic for this country. To me, the most remarkable thing about Musk is his ability to turn losses into victories. If you look at Twitter, it seemed like he had spent $40 billion and wasted it, right?
And certainly, the value of Twitter is massively less than when he bought it. But his ability to use it to help get Trump elected, which has then massively helped all of his other businesses. Like, Elon has made hundreds of billions of dollars off of his Twitter acquisition. His ability to…
SAFIAN: But not through Twitter.
THOMPSON: Not Twitter, right?
He’s lost a lot of money through Twitter. So anybody who gave him money for Twitter regrets that. However, he’s been able to turn the platform into a sort of a tool for self-promotion and mobilization, and it’s hugely rebounded in his favor. So his ability to turn a losing hand into a winning hand… Gotta admire that.
SAFIAN: Characters like Musk, Zuckerberg, and Altman are remarkable. I love how Nick talks about them in the context of 2024’s historic tech boom, with an eye toward what happens from here. After the break, Nick talks about Apple, TikTok, and what it’s like to run The Atlantic amid growing mistrust for media. Stick around.
[AD BREAK]
SAFIAN: Before the break, Nick Thompson, CEO of The Atlantic, talked about how and why Big Tech is thriving, and analyzed the impact of Elon Musk, Sam Altman, and Mark Zuckerberg. Now, he talks about Apple and TikTok, his prediction for the transition from search engines to answer engines, plus how AI is poised to reshape journalism. Let’s dive back in.
Apple’s future amid challenges and opportunities
Apple had an interesting year.
They made a big bet on the Apple Vision Pro, which didn’t really pay off. Apple intelligence is kind of a clever playoff of AI, but not yet a breakthrough experience.
Do you see more trouble or more opportunity ahead for Apple? Because it feels like a particularly tricky moment.
THOMPSON: It’s a tricky moment, but I mean, like you, people have been saying Apple has lost its steam since six months after Tim Cook took over. So, Vision Pro, okay, maybe it didn’t hit the mark, but they have so many products. They have an ecosystem people are so locked into.
I think on-device AI is actually going to really matter. And I think that we’re going to want on-device AI that really works in part for energy costs. You’re not going to want to be constantly calling servers with privacy concerns.
I think Apple will figure out on-device AI, whether it’s two years from now, but, you know, AI will end up being extremely valuable to Apple.
The complexities of regulating TikTok
SAFIAN: You recently called the potential ban on TikTok a slippery slope for First Amendment rights.
THOMPSON: I have kind of non-conventional views on it in which, like, I don’t like the existence of TikTok partly because, as a father of three children, you can see how toxic it is on young people. I see the dangers.
And I have lots of friends in security who are all very convinced that TikTok can and will be weaponized by China in a potential moment of conflict, right? If there is a conflict over Taiwan, the smartest people I know who know the most about national security absolutely are convinced that TikTok will be weaponized to change American public opinion on.
In fact, there are real privacy risks.
All of that said, I don’t like the idea of banning it. It just does not seem like something a country that values the First Amendment should do. I understand that there’s a complicated question of whether technically the existence of a platform counts as free speech. Right? Is it the individual? Is it the platform? There is an interesting legal argument over whether the First Amendment applies. I am just saying, as someone who loves speech, who loves more of it, the notion that we would go and shut down an entire platform where people exercise their free speech, is quite unsettling.
In the United States, well, you try to do counter-programming. You try to have speech of another sort. Maybe you have people go on TikTok and try to present an alternative view. Maybe you talk more about the biases in the algorithm. I don’t think you shut the whole thing down.
I think the United States would be in a better position with no TikTok. I do not like the logic for it.
Navigating media trust and business sustainability at The Atlantic
SAFIAN: I want to ask you about your business at The Atlantic. You’re operating and running this business in the midst of growing distrust of the media. I’m curious what it’s like for you to run The Atlantic in this climate and how you navigate and think about that.
THOMPSON: Well, that’s a very hard thing and is one of the most unfortunate facts about media today.
It does hang over everything. So why is there all this distrust, right? There’s all this distrust because, in part, there’s been a huge campaign by extremely rich people, including Elon Musk and Donald Trump, to discredit the media because they didn’t like coverage of themselves. It has happened because of the disintermediation of the media, it has happened because of mistakes that media has made, there’s all kinds of reasons, but it has certainly happened.
What we have to do is make sure that we are as trusted as possible, work as hard as we can to make sure every fact is as accurate as possible, that every story, in a big sense, is as fair as possible, that we’re reaching as many people as possible, and it’s an editorial matter.
And then on the business side, we have to go out and find the largest group of readers we can and to get them into direct relationships with us as opposed to relationships through intermediaries, particularly tech platforms or walled gardens.
SAFIAN: One of the things that I always struggled with as an editorial person, which I know is sort of where your soul is, is wanting to get your stories out, your content out to as many people as possible. But sometimes the best business model is a high price to a fewer, smaller, more targeted group of people. How do you think about balancing those motivations and those business realities?
THOMPSON: I have a view, which is probably not the one that maximizes economic return, but it happens to work, which is editorial chooses what they’re going to write about and how they’re going to write about it, and they do that without input. And then once that’s done, we choose on the business side how best to make a business model around it.
Sometimes it does involve trade-offs. So my view is like, edit should write whatever stories they want to write, however they want to do it. Then we should go out and fight for every possible reader in every possible way, right? You should SEO optimize, figure out the exact right time to post it on social media. You should do all the things, like get it out on the subreddit and hope that somebody posts it there. Fight for every reader. Don’t just sit. I think that is one difference between the way I view it and the way some other people, particularly in elite media institutions, where they view that as not something you should spend time and effort on.
But I believe in fighting for every reader in every possible way. Also, though, I believe in finding a business model that is most likely to make the place financially sustainable and able to exist in the long run. Probably the key thing that we’ve done at The Atlantic, and also something we did when I worked at The New Yorker and when I worked at WIRED, is build a paywall.
There are trade-offs in a paywall because once you have a paywall, fewer people can read your stuff, and you accept that trade-off, because you have to make the place sustainable, and you can’t support the kind of journalism The Atlantic does without a subscriber model. It’s an either-or. You can either support this and make it sustaining in the long run with a paywall, or you cannot. And so we’ve chosen to do that.
Inside The Atlantic’s partnership with OpenAI
SAFIAN: I know The Atlantic has a partnership with OpenAI. Is that something that’s going to meaningfully benefit the business?
THOMPSON: No. I mean, they do pay us, but we have advertisers who will pay us more than AI pays this year. I mean, it’s a revenue stream. The most useful thing that will come of it is if, through our partnership, OpenAI builds a search product that drives readers and does over time, like Google built a search product that helps support all of our media businesses.
That’s going to be very hard to replicate as we move from search engines to answer engines. My default assumption is that three years from now, we’re navigating the web through answer engines and we’re not clicking out to external sites. If OpenAI ultimately is able to build an answer that also drives readership directly to products created by media organizations, that is much better for media.
So by far the most important thing will not be the check that they will write to us for licensing, but will be, in the long run, I hope, an answer engine that really drives readership.
That’s good for everybody, right? That is the outcome that is best for OpenAI. If search engines disappear and answer engines undercut the whole business of media and media organizations go away, well, then there’s far less quality information to feed into the answer engines. Google benefited because it drove a lot of readers.
We were able to drive businesses around those readers, and there was more high quality information that Google could index.
The hope with AI is that the same thing will happen. That is entirely unclear, though, and my worry is that we’re not heading in the right direction. So I’m doing what I can to try to push us there, but I don’t know whether that will succeed.
SAFIAN: Yeah, because otherwise, the economics of being a creator of this content sort of dissipates.
THOMPSON: Yeah. By far the best outcome, in my view, would be answer engines properly crediting the content that helps create the answers, right? You know, The Atlantic has partnered with a company called ProRata that is working on this. They’re trying to show, in any given answer, okay, this answer is 6 percent due to a Fast Company story and 5 percent from a Wall Street Journal story. And can then attribute revenue back to the publications on that. And then also, it has external links so that people can go and read the original source material, and then also takes into account journalistic values, doesn’t quote excessively, doesn’t mimic style, right? There are a whole lot of things that really matter to writers and journalists.
That would be the ideal answer. So it is paying people whose sources contributed to the answer. It is driving people to it, and it is respecting journalistic rules. Whether we’ll get there, who knows. The economics are not entirely aligned, but that’s the goal.
What is success for The Atlantic today?
SAFIAN: So how do you think about or measure success for your media company today?
THOMPSON: Success for us will be increasing our reach, increasing our influence, publishing as much journalism into the world as possible, and being financially sustainable.
Our owner is Lorraine Powell Jobs. I should probably disclose that while talking about Apple, she was married to Steve Jobs. So she is well off, and whether The Atlantic makes money or not is not her principal source of income. However, she is a businesswoman, and she has mandated that we operate this place at a profit. She doesn’t want to fund it indefinitely. There was a period of losses that she covered. So, we worked extremely hard over the last few years to get this place to profitability, and we fully intend to stay there. Every decision we make will be directed towards doing the best and most important journalism that has the best effect on the world that we can do, and doing that in a way that is sustainable. We don’t want to ever go back to losses.
How Nick Thompson tracks his goals each year
SAFIAN: This time of year, a lot of people aim to set new goals to get themselves on track. I’m curious if you have any advice for those who are looking to create new habits for the new year.
THOMPSON: I have a thing I do each year where I have habits I want to change or shift, and then I keep track of how many days in a row I’ve done that. I have an Excel spreadsheet that tracks it for the whole year.
On, I have a pretty intense to-do list, which I think has been very helpful to me.
It’s basically structured with, on the left, what I’m going to do right now, and I try to only do like three things. So that will do this, and then we’ll do that. I don’t do this, and then the second category is stuff I’m going to do today. Like, so this stuff has to get done today. And the third category is stuff I’ve got to get done that doesn’t have to be today, but ideally, you’re getting rid of doing some of it. And then the next category is like, you know, the list of stuff for the year that I’m trying to pay attention to.
It helps me keep track of what’s going on in the day so you don’t end up losing a bunch of time because you look over and you’re like, “Oh wait, I gotta get all that done today.” Right? So it’s a useful mechanism.
What’s at stake for The Atlantic in 2025?
SAFIAN: Yeah, well, you got to find the system that works, the one that’s right for you. Are there things at stake as we look ahead to 2025 that are?
THOMPSON: Well, here’s the thing that’s at stake for us: I do not think that The Atlantic or really any other media company is positioned in a way that makes it fully resilient if AI continues on the trajectory that it is on right now.
If the improvement curve is the same as it has been in the last two years, there will be a complete reshaping of the internet and there will be a reshaping of the way knowledge, information, reporting, and storytelling work. So preparing The Atlantic to be in a completely resilient spot is very hard, and it’s a combination of figuring out how answer engines will work, figuring out how The Atlantic will exist in a web that has been turned upside down by AI, figuring out how AI can be used to help make the business side operate more efficiently, which we’re doing.
Figuring out also how to prevent AI from being used on the editorial side in a way that would reduce trust, because there’s no doubt that there can be a bunch of publications that use AI to report and tell stories in a way that blows trust, right? And so you can’t do that. We say with The Atlantic magazine, for humans by humans, everything is written by humans.
Well, also, I genuinely believe that it will be a useful tool for interrogating ideas, figuring out how to go through data sets. So the balance of how AI is used in the newsroom is extremely delicate and extremely important.
Then the way we build our business model to make it maximally resilient in a world in which there is extremely powerful AI. That is my New Year’s resolution, to try to get that more right.
SAFIAN: Well, Nick, as always, it’s great to talk with you.
THOMPSON: Oh, so much fun. That was amazing. I love chatting with you, Bob. It’s always wonderful to see you, always wonderful to chat with you, and it’s an honor to be on the podcast. Thank you.
SAFIAN: Like Nick’s approach to getting his runs in, I take a pragmatic outlook on AI and its implications on the work I do — namely the cross-section of journalism and podcasting. There’s a lot of speculation about how AI is poised to transform these mediums, but there’s only so much preparation we can do until the capabilities are there.
I’m Bob Safian, thanks for listening.
I tend to agree with Nick that the story points to a much larger issue for the future of technology and free speech here in the U.S., which itself could have real implications for the likes of Altman, Zuckerberg, and Musk.
The post Big Tech trends shaping 2025 appeared first on Masters of Scale.