On with Kara Swisher - Can Smaller AI Startups Compete in the AI Race?

Episode Date: May 26, 2025

Last year, AI and machine learning startups raked in about half of all VC funding in North America. And so far this year, AI is still leading the pack. But a huge chunk of the money in the Q1 — $40 billion — went to one player: OpenAI. So is there still room for smaller, more focused startups in the AI gold rush? Or will it be a case of “winner takes all?” In this live conversation at the Johns Hopkins University Bloomberg Center's inaugural Discovery Series, Kara speaks with Gary Rivlin, author of “AI Valley: Microsoft, Google and the Trillion-Dollar Race to Cash In on Artificial Intelligence,” and Christy Wyskiel, senior advisor to the president of Johns Hopkins University for innovation and entrepreneurship and the executive director of Johns Hopkins Technology Ventures. The three discuss the impact of government cuts on AI research, how small AI startups can compete with the tech giants, and how AI could revolutionize health care.  This interview was recorded on April 28, 2025. Questions? Comments? Email us at [e-mail] or find us on Instagram, TikTok, and Bluesky @onwithkaraswisher. Learn more about your ad choices. Visit podcastchoices.com/adchoices

I just heard a really good joke, I have to tell it. TechPro stands for technically broken. You like that, Gary? I do. Hi, everyone from New York Magazine and the Vox Media Podcast Network. This is On with Kara Swisher and I'm Kara Swisher. As we talked about on the show last week, the AI boom hasn't slowed down. According to Pitchbook, AI and machine learning startups took home about half of the venture

capital dollars in North America last year. Globally, it was about a third. It add up to more than $131 billion. So far this year, AI is still the top sector for venture funding. But a huge chunk of the VC funding in the first quarter of 2025, $40 billion went to open AI. So the question is, is there still room for smaller, more focused startups in this AI goal rush, or is everyone just placing their bets on the big guys? My guests today are two people who've been following the story on the ground, so to speak.

Gary Rivlin is a Pulitzer Prize-winning investigative reporter who has been writing about technology since the mid-1990s. He has written 11 books, including his latest, AI Valley, Microsoft, Google, and the Trillion Dollar Race to cash in on artificial intelligence. It follows the rise of inflection AI and why, despite deep-pocketed funding, much of the team, including leadership, was eventually assimilated into Microsoft. My second guest, Christy Weisgel, is the senior advisor to the president of Johns Hopkins University for innovation and entrepreneurship.

She oversees Johns Hopkins University for Innovation and Entrepreneurship. She oversees Johns Hopkins Technology Ventures, where she helps students and researchers launch their academic discoveries to the private sector through industry partnerships, technology licensing, and startup company Incubation. Our live interview was recorded on April 28th at the Johns Hopkins University Bloomberg Center as part of their inaugural Discovery Series. It was a great conversation and says a lot about the state of the industry today and where it's going tomorrow. Have a listen. The Tribeca Festival is back June 4th through 15th, and it's packed with camp-miss experiences.

Catch Sandra Oh on a live podcast recording of The Interview from The New York Times. Cheer on track-and-field superstar Allison Felix in the documentary She Runs the World. Or catch My Mom Jane, Mariska Hargitay's moving documentary feature directorial debut about her mother, Hollywood icon Jane Mansfield. There's something for everyone. Get your tickets now at trybeckafilm.com. The new BMO VI Porter MasterCard is your ticket to more. More perks.

More points. More flights. More of all the things you want in a travel rewards card. And then some. Get your ticket to more with the new BMO VI Porter MasterCard and get up to $2,400 in value in your first 13 months. Terms and conditions apply. Visit bmo.com slash VI Porter to learn more. Truck month is on at Chevrolet. Get 0% financing for up to 72 months on a 2025 Silverado 1500

Custom Blackout or Custom Trail Boss. With Custom Trail Boss's available, class-exclusive Duramax 3.0L diesel engine and Z71 Off-Road Package with a 2-inch factory suspension lift, you get both on-road confidence and off-road capability. Dirt road ahead? Let's go! confidence and off-road capability. Dirt road ahead? Let's go! Truck month is awesome! Ask your Chevrolet dealer for details. It is all.

Christy, I just met you, but Gary and I go back. Before we had kids and everything else, when at the dawn of the internet age, he wrote a book. The Plot to Get Bill Gates. Plot to Get Bill Gates. We were on a book tour with Poe Bronson, who wrote Nudists on the Late Shift, and I had

the paper-wrack version of AOL.com, and it was literally the dawn of the information age. And John Karp, who's now running Simon & Schuster at the time, had this genius idea to send us on a tour around the country called the Bleeding Edge Tour. The Silicon Valley Bleeding Edge Tour. Right. And so I was there to say this was a great moment and huge fortunes would be created, it turns out I was right.

He was there to be sort of the contrarian, right? Sure, I'd point out that a lot of fortunes were lost and then eventually a few people made big fortunes. Right, that's correct. And then Poe Bronson was there to be super earnest. And we had the best time and we went all around the country talking about this new internet. And it was so weird to think about what's happened since.

In any case, we were young then, so naive, but not really, he was never naive. He had sort of a pain in the ass everywhere we went. But anyway, it was in a good way. But the feeling is mutual, and I think that's what counted. Yes, you're doubtful, yes, exactly. So anyway, you've come out with a book about the current makings of money around the AI sector.

You're one of many. I've just been delivered three or four, mostly about open AI, actually. Yours is called AI Valley, Microsoft, Google, and the Trillion Dollar Race to Cash In you're one of many. I've just been delivered three or four, mostly about open AI actually. Yours is called AI Valley, Microsoft, Google, and the trillion-dollar race to cash in on artificial intelligence. Well timed. Christy, you oversee Johns Hopkins technology ventures, which means you're responsible for shepherding university research from the lab to the private sector, including in AI. Both of you had a front row seat to the growing sector and a

broadcast of characters, researchers, founders, entrepreneurs, incubators, VCs, and also these big tech billionaires. A lot of drama. So to start off, can you describe the interplay between these characters and their relationships, this new field? Where's the most tension, affinity? Who do you relate to the most?

Why don't you start, Gary? So, you know, it's interesting. I've always loved startups, venture capital startups, the gambling, that whole ecosystem. So in the end of 2022, I said, wow, this AI thing's gonna be a big deal. So I wanna like, what startups are figuring out how to cash in on this moment?

So I wanted to go find what's the next Google? What's the next Facebook? And unfortunately, what I found is the next Google is probably Google when it comes to AI. The next Facebook or Meta is gonna be Meta. That there's still all this opportunity to create nice, in quotes, modest businesses, tens of millions, hundreds of millions of dollars.

But the next Google, those foundational companies that will be eventually worth a trillion dollars, AI is so expensive that there's very few startups that really can afford to be playing, you know, basic stuff like text to video, text to audio, large language models. I really fear is going to be the stuff of the large corporate, the large big tech. Yes, which is something I've talked about a lot. This is a sphere of mine is that this is not the same thing that when you and I were first, these were a lot of startups you never heard of, you never heard of Uber, you never heard of Instagram.

They were real ground breakers. In this case, it's too costly and also these companies have so much sway over compute that it's hard. Chris, you can say. And the data. And the data. They have access to the data, which is the new gold. Right.

Yeah, well, I love talking about startups. I was an investor and entrepreneur for a long time and we've spent out about 10 new companies out of Hopkins Research every year. But I think what people don't necessarily realize is that the foundations of the tech industry actually come from a lot of the work that happens at universities that's not acknowledged. Think about how many of you actually got here by typing something into Google Maps or Apple Maps today or your Uber or Lyft or whatever. That foundational research of GPS, people don't necessarily realize or acknowledge. That actually came from Hopkins researchers understanding Doppler satellite data 50 years

ago and eventually in the 80s was allowed in the public domain. And here we are using it every day. And so the origins of so many of these tech companies come from what we're doing. Before. Before. Before, meaning the early tech, yes, Stanford or whatever, but it's less and less so. That's been... I disagree.

I think if you think about where so much of the technology continues to come from, it's not like when OpenAI was founded 10 years ago that Sam and Ilan woke up and magically thought of AI. That was over decades and decades of research. And I think you're continuing to see that the types of faculty and students that are learning and training at places like Hopkins, but all over, they are bringing those nest technologies to market,

whatever the next GPS is in the next decade. I think what you were trying to say is what's scary to so many of us, what must be scary to Silicon Valley, even if the leaders are being quiet about it, is cutting off basic research to universities. Research for foundational exploratory research is so short-sighted. That's our magic formula in this country. That's our secret sauce. You know, AI in the 50s, the 60s, the 70s, the 80s, it was government-funded research.

And you know, my big fear is, you fear is the Chinese have said by 2030, they want to be dominant in AI. And if we're cutting off this basic research, I think we're giving them a big advantage. Yeah. Christy? Look, I agree. I think that, to quote my former colleague David Singer,

who just had an article this week. In The New York Times. In The New York Times said, drawing a causal link between federal investment in basic science research and the rise of VC industry is about as difficult as reading a map. And so those connections are happening then, now, and in the future. And whether it's the AI winter that you described in your book from the 70s and 80s, when governments stopped funding research into AI, the research stopped, the same thing could happen now.

And it's just really important for us to acknowledge that. So talk a little bit, Chris, you play actually some of these roles, as research stopped, the same thing could happen now. And it's just really important for us to acknowledge that. So talk a little bit, Chris, you play actually some of these roles. As you said, investment analysts in the 2010s, you co-founded your own biotech startups. And 12 years ago, you started this. Now a lot of universities have these. What was it like to start that a dozen years ago? Because Stanford was already pretty active. Certain universities had already moved in that way.

Stanford probably is the most famous, as Gary knows. Yeah, Stanford and MIT, a number of universities have a 50-year head start, basically, to what this is. When the Dean of Engineering at Stanford in the 70s looked out at Palo Alto and said, there's a field, we should probably put a research park there. And some of the first inhabitants of that research park were Mr. Hewlett and Mr. Packard, right? They have the head start. Hopkins has always been around the foundational basic research, but I was

hired to be a bit of a disruptor 12 years ago to come in and think about how could we commercialize the incredible things that come out of Johns Hopkins. So that's the mission. I wake up every day and do that. So, Gary, in your book, you write about the early iterations. You and I have talked about this many times, the Silicon Valley. Obviously, Stanford is the unique role in that area, as you just noted. Google began also as a graduate student project there. Talk about sort of early research and incubator programs. Right, so, you know, dating back to the 50s, 60s,

it was the students on the campuses, it was the professors who were giving rise to these companies. So in the 1980s, there were all these companies blooming around Stanford, MIT, a few other universities that were doing cutting edge AI research. And you know, actually a few of them went public, they didn't last. You know, there was a second AI winter after that.

And that one was a little bit different because in the 80s, it was companies and it was venture capitalists that temporarily got into it. They were burned and then it really wouldn't be until 2015, 2018, where the pioneering VCs were finally daring to put money into AI back then. It would have been autonomous vehicles. But you know, I mean, that's the Silicon Valley way. That's what I love about Silicon Valley.

You say Stanford, to me, it's just a startup machine. The whole place is almost built to create startups. You've got Stanford, which gives rise. You got UC Berkeley, which gives rise. But the venture capitalists out there and there's this whole ecosystem. That's startups to venture capitalists, to wealth to the triangle.

Right, right. And you know, there's almost like this mindset this mindset like why aren't you starting a company they're hiring people from a Google or Meta. And still, you think still today that's the case. Yeah I mean yeah. From an AI it certainly is, it's in San Francisco now. Right so one big difference between the 90s when we were you know starting off reporting and today is the center of gravity has moved from Silicon Valley to San Francisco. It's not a coincidence, OpenAI, Anthropic, a lot of the other big AI companies are in

San Francisco. The venture capitalists used to famously be in all in Menlo Park. They now have outposts in San Francisco. And again, I don't want to give the wrong idea. There's going to be a gazillion startups that have nice returns for venture capitalists, nice returns for venture capital, nice returns for their founders. It's more that stuff that's the real innovation, the stuff underneath.

So there'll be apps on top of these foundational models, the basic stuff, but I really do fear that maybe OpenAI will break in, maybe Anthropic will break in, that's Claude, the LLM. But I actually think it's a pretty good chance, given how much money they still have to raise long before they'll ever show profit, that they'll just get bought up by a giant. And by the way, what's interesting is like, Google could lose, like Gemini there,

Chatbot doesn't win, but they've invested billions of dollars in Anthropic. Microsoft and their co-pilot could lose, but they put what, 15 billion or so into OpenAI. So even if they lose, they win. Correct. We'll be back in a minute. When does fast grocery delivery through Instacart matter most? When your famous grainy mustard

potato salad isn't so famous without the grainy mustard. When the barbecue's lit, but there's nothing to grill. When the in-laws decide that, actually, they will stay for dinner. Instacart has all your groceries covered this summer, so download the app and get delivery in as fast as 60 minutes. Plus, enjoy zero dollar delivery fees on your first three orders. Service fees exclusions and terms apply. Instacart, groceries that over-deliver.

Okay, so the other day I was in Portland, Oregon on vacation, I'm hanging out, feeling really good about myself, and I walk past this store where a guy is offering free skincare samples. I say, sure, I'll take one. Then he literally grabs me by the hand and pulls me into his store.

And suddenly he's putting like this goop under my eyes. And I'm not a big skincare girly, but I dabble. And I'm telling you that this stuff was magic. Like I have the beginnings of crow's feet, but I'm looking in the mirror and they're gone. And he tells me, this stuff costs $1,300, but it is so worth it because you won't need Botox

for another three years. How old are you, Amanda? I'm 28 years old. Ha ha ha ha! Coming up on Today Explained... The pressure to fix your face. Tell them when we drop. Weekdays wherever you get your podcasts.

Have you ever scrolled through Instagram in the summer and wondered how everyone you know is suddenly in Europe? You've probably thought to yourself, how can they actually afford that trip? Well, summer's almost here and on this week's episode of Net Worth and Chill, I'm diving into practical ways to save on summer travel. Flight prices change constantly, literally day to day, and often those changes in price can fluctuate based on your location, the strength of your country's currency, and how popular a destination is in your region. So how do we see what that flight price would be in another place, we are using a VPN.

Plus, I'm answering some of your burning travel questions from finding better flight deals to making the most of your travel credit cards. We're breaking down the travel industry's best kept secrets. Listen wherever you get your podcasts or watch on youtube.com slash your rich BFF. So Christy, part of your mission is to make Baltimore itself a tech hub, like Silicon Valley or Boston. It's really Silicon Valley and everybody else goes down pretty far. As Gary writes in the book, Silicon Valley is both a place and idea.

Baltimore was one of the 31 federal tech hubs. It was skipped over in the first two funding rounds. Who knows what's going to happen now? But talk about this idea, you know, when they had Silicon Beach, they had Silicon Prairie, they had Silicon Haller. I'd always, I would be like, no, this is not happening in any of these places. So talk a little bit about trying to do that in a place, because place does matter in an, even in digital world, place does matter. Yeah, no, I think it does.

And look, at Johns Hopkins, one of the things that we are known for is biomedical research. A lot of the basic foundation models, you can look back to Bert Vogelstein, who several decades ago discovered the genetic basis for cancer. And since then, we have built on all sorts of therapies and diagnosis. What is the future of drug discovery? It's going to be AI. And we're making a huge bet, as you probably know, Kara, in AI at Johns Hopkins, the State of Science Institute, over 100 hundred faculty, hundreds of students, who is going

to train the people that are going to work at those companies? It's going to be places like Hopkins. And so I think the types of companies that we spin out are those that are good for the world. We're not creating the seventh dog walking app. We're actually, you know, creating... Why not? Right. I mean, are the first six not good enough? I mean, really, let's- No. Let's-

Healthcare is really, oddly enough, my next book's all about this. But the AI's impact on healthcare is obviously one of the biggest possibilities. My friend Alec Ross, who wrote a book, Industries of the Future, said that the last trillion dollar industry was created with using zeros and ones, and the next trillion dollar industry will be with the four letters of DNA. And there's no reason why Johns Hopkins and Baltimore will not dominate that. We already are. We already have companies spinning out, raising billions of dollars.

So according to PitchBook, just for a little fact check here though, venture capital investment in AI startups was up more than 50% last year. Over $131 billion feels fublicious. In the first quarter of the year, they say the AI sector raised $73.1 billion, but spoiler alert, $40 billion of that went to open AI. Both of you talk about this. You raised about $4.4 billion, correct, Christine? Venture capital.

The total company is out of coming up. So Gary, first, can you talk a little bit about this fact that it's sort of coalesced around really one company? Right. Well, a few. So XAI, that's Elon Musk's company there out there trying to raise, I think, 20 billion. So 40 billion is the largest raise ever in venture history. 20 billion would be the second largest. You know, we covered the dot-com boom and bust,

and we're seeing some of the same things. So Ilya Tsetskever, famous co-founder of OpenAI, famous for initiating a coup that lasted five days at OpenAI. It's called a coupette. So, you know, after he left, started his own company, Safe Superintelligence, and, you know, after he left, started his own company, Safe Superintelligence, and, you know, they raised $2 billion at, you know, a paper worth of $30 billion. They don't have a product yet.

Right. But that's venture capital, right? They slap down bets really fast. So, you know, some company writes a memo. Two or three promising founders write a memo. A week later, they've raised tens of millions of dollars at a hundred million valuation. It's really just rolling the dice, but that's the venture capital business. Most of these will not work, but a few of them will.

And when they do, they'll show extraordinary returns. So Christy, you said with these investments you're making, you co-founded Upsurge Baltimore Network of Entrepreneurs and Startups to facilitate the funders. Talk about that environment, because a lot of people definitely feel that there's an over investment in infrastructure, an over investment in the valuations. And this isn't like the internet because it was just an over investment in companies. It wasn't infrastructure. It wasn't the costs were quite low actually.

Here they're very high with whether it's energy or infrastructure or whatever. So the bits are slightly bigger in that regard, not just the money, the figures. So talk a little bit about that. Yeah. I mean, 80% of what we do is in the life sciences. Again, there's a heavy overlay now with drug discovery and AI on that, but it's been a bit of a biotech winter since 2021. And so it's a little bit different and it's also not necessarily Kara about the cult of

personality. It's much more about the research and what's behind it. And I would say those valuations and the money flowing there are much more down to earth. It reminds me a little bit, you had Josh Johnson on your podcast. I'm a huge fan. The famous AI researcher and comedian. Exactly. But he actually made a really, I thought it was really poignant about the idea that so much of our world today, the foundation of that was people who will never know their names, will have forgotten some of the things that we take for granted.

And I feel like that's so much of the research that happens at places like Hopkins are not necessarily going to get the same hype of a big AI startup in California, but there's the level of impact that one would have. So you can get ROI and you can bring great things to market, but it doesn't necessarily have to raise 30 billion and be on the front page to do that. Or have to deal with egos and things like that. Hopkins is a very low ego place. I get to deal with some of the smartest, most earnest folks in the world and at the end

of the day, what they care about is the impact, not their name in the headlines. Silicon Valley's got you covered for that. So still, still, after all this time. Especially. Especially, really. It's gotten worse, hasn't it? It's gotten much worse. Yeah. It really has.

Yeah, yeah. Every human growth hormone they take, it's gotten worse. Anyway. They take other stuff too, I've read. I've read, yeah. No, they say it. It doesn't matter. So I want to get deeper into AI, specifically the intersection of life sciences and this.

Gary, you point out in your book, healthcare is the area, the techno-optimists, and that's the term they use for people who are pro AI versus they're gonna kill us tomorrow. They often use the proof point to bolster their position that AI will change lives for the better. In a recent 60 Minutes interview, Google DeepMind CEO and Nobel laureate Demis Hababas said he believes AI has the ability to cure all diseases within the next,

this is something they go on. This is the problem. Okay, all right, okay, wait, okay. Okay, sorry. Do you think, talk about this, go ahead, go ahead, go ahead. No, no, it is. I'm sorry, allow you is something they go on. This is the problem. Okay, all right, okay, wait, okay. Okay, sorry. Do you think, talk about this, go ahead, go ahead, rant away, I'm sorry, allow you to rant. No, I mean, I would say the same about the early internet, like it will do extraordinary things.

I am totally with both you that AI and healthcare is gonna be extraordinary for humanity. The connections it's gonna make, the vaccines, the treatments, all of that. But why do you have to say it's going to wipe out cancer in 10 years? Which I have heard. Well, why do you have to say it's going to wipe? It won't.

I mean, it'll be- It's sort of like saying self-driving is here today, 10 years ago. Someone said that, I'm not going to say. No, but that one I believed. I really thought- He's been recently unemployed from Doge, but go ahead. They're coming. No, but it's like, this is where Silicon Valley, the tech bros get themselves in trouble.

They're overhyping. And like, AI has a problem. First off, they had really bad timing. AI hit at the end of 2022 when distrust among the public in big tech was at its peak. You know, the terms I like to use, there's the do-mers who are convinced that laser-eyed robots are going to take over and subjugate humanity. There's the zoomers who want no speed bumps, just let us do our thing, China, China, China

kind of thing. Meaning they're going to beat us. And you know, I'm kind of more with, there's a Reid Hoffmanism, a bloomer, that I think I could do extraordinary things, but let's be deliberate about it. We're way ahead of where the public is. Pew did a poll last year, the majority of people are fearful of AI. And here we're talking about a technology where humans are going to lose their APEC status.

They're going to be these agents, AI is gonna be our personal assistants, these AI agents, which means they're gonna know everything about us and privacy issues. Like, we're getting way, way ahead. And I think the Zoomers are making a strategic error by saying like there should be no regulation. We should just like do everything we can to accelerate,

to quote our vice president, stop with the hand-wringing and let's just start winning. And I'm just scared they're going to get way ahead of where the public is. It is interesting because the ones who are terrified are also not so much fun at a dinner party. I was at one dinner party with a bunch of them and they were like, you need to stop Sam Altman because the human race is, you need to stop him now. Kara, you're the only one. And I'm like, that's the plot of The Terminator and I am not her. Like I have no...

It was like they literally recounted the plot of The Terminator to me and I had to kill him, which I didn't. I think our problem is that Hollywood is defining what we should be scared of. Right. Whereas there's much more line of sight things we should... AI and surveillance, AI and surveillance, AI and warfare, AI manipulating people, autonomous AI, this stuff isn't ready, humans still have to be in the loop.

That's the stuff I wish we were debating. That's the stuff I wish we were, you know, worried about and not like the robots taking over. Many are. But Christy, life sciences, obviously, as John's talking about, superpower. But a lot of the things, I'd like to hear about them because they are very AI-assisted tools for early detection, disease prediction. When I was with Vinod Khosy, he was talking about drug interaction is something that kills a lot of people, the wrong drug. Obviously, diagnostic tools for respiratory inches, cancers. Where do

you see the most potential? One very specific area, I can think of several, but I think within cancer diagnosis, there are 30 million CT scans done on the abdomen every single year. Imagine a world where you could just with a simple algorithm detect pancreatic cancer, right? And we have researchers working on that to basically say there could be otherwise undetected pancreatic cancer. And I don't know how much you all know about that. But generally, by the time it's detected, it's a death sentence.

And so the idea of early detection, which again is something that a number of our researchers have been working on because cancer develops over decades, not overnight. And so the sooner you can find it, the sooner you can treat it. Another application that we're all parents up here that maybe you would have appreciation of, we have a wonderful woman who's an ER doc who basically just saw parents exasperated having to bring their kids in for a strep throat test, right? So she figured out that- I did that yesterday, but my wife did it. I did not do it. My wife did it. She totally gets credit.

So think about a world where you take your phone camera, take a picture of the throat and it says strep throat or not and you get the prescription. You don't have to sit at urgent care on a Friday night. You don't have to come into the ER and wait 12 hours on a Sunday, but you basically get treated right then. So both of those applications, one is great quality of life. One is literally life changing, life saving. In both of those cases, those are real and happening. We'll be back at McDonald's. With juicy 100% Canadian-raised seasoned chicken, shredded lettuce, crispy jalapenos, and that completely craveable hot honey sauce, it's a sweet heat repeat you don't want to miss.

Get your Hot Honey McCrispy today. Available for a limited time, only at McDonald's. OpenAI spent the last few years turning chat GPT into one of the most important and popular products on the internet. Johnny Ive spent the last several decades building products at Apple that became truly iconic like the iPhone. Now those two are teaming up to work on something. We don't know much, but it's gonna be some kind of AI gadget and they think it's gonna be a really big deal.

This week on the Vergecast, we talk about what John E.I.V. and OpenAI might be up to, plus everything that happened at Google I.O., the Developer Conference, and all of the other news in the AI and gadget world, because there is just so much of it. All that on the Vergecasts, wherever you get podcasts.

When comedian Chris Gethard was growing up, he went to a place called Action Park. It was one of the first water parks in the country, and it was built by a man who had no experience building theme parks. Some people called him a berserk Willy Wonka. Anyone who went to Action Park understood you could get really messed up going there. Not only did we know that, it was a huge part of the appeal. I'm Phoebe Judge. Listen to our latest episode, Action Park on criminal, wherever you get your podcasts. So one of the, these R&D heavy startups aren't coming just out of universities, Demis have

office, as I just mentioned, co-founded Isomorphic Labs. It's a division of Google spun out of DeepMind to focus on drug discovery, for example. They not only have Google money, they recently raised outside VC money. This kind of internal R&D used to come from universities. Gary, what impact does it have? And Christy, how do you build them into sustainable businesses if they're destined to be bought out by tech companies

or big pharma as acquihires? What does that look like? First you, Gary. Well, I mean, corporations have been investing in cutting edge research, Microsoft research, Google has a big research arm, IBM for decades and decades. So that's always been going on. It is interesting. Some of the biggest venture capitalists in Silicon Valley right now are the tech companies. You know, Microsoft in 2023 was the number one VC just by dollars given.

NVIDIA, they invest in like 35 or so startups. So there is this, some of it is just so damn expensive. Like, you know, we're talking about a $40 billion raise. A large venture capital fund in Silicon Valley is $1 billion. That's a tiny fraction of what they need to raise. And so, you know, what we're seeing is the large corporations, big tech companies, Amazon, Apple, Google, are taking over that role. And, Chrissy, what does that have, this idea that how do you make a healthy AI industry

if there's consolidation in this way or if the research comes out of here because they weren't going to let it out. You want diversification, presumably, to be healthy. Yeah, absolutely. Look, one of the cool things about my job is I get to be a bridge between these basic science researchers and industry and be that front door for Johns Hopkins. And so we have a lot of examples where we took things and worked really collaboratively with it.

There are actually a lot of good players out there. We have a drug that is for men's health. It's called Pilarify. It's for diagnosing prostate cancer. And one of our clinicians described it as if you wanted to treat prostate cancer before this incredible technology came out, it was like watching black and white TV in the 1950s and it's now in Technicolor. And we licensed to a company in Boston called Lanthias. They sell it. we get royalties on it, so everybody wins. The patient wins, the company gets the growth and the ROI. We're rewarded with those royalties as something that we discovered.

So I think just being that natural bridge over and over, that does require care though, that we get credit for what we've done, right? So a lot of working with industry is maybe acknowledging that there were people that came before, there were people that built this foundation. And what does that look like? And how can we be more collaborative? But the idea of these acquihires, there was obviously Mustafa Suleiman with Inflection. There were many AI companies started now.

Microsoft essentially bought that company, although it wasn't looked at that way. They hired everybody. They hired virtually every employee there. And then gave $650 million as a licensing deal to like, you know, make everyone whole, make everyone happy. Yeah. You know, my big concern on this is, you know, this idea of consolidation of power in just a few big tech companies.

The Silicon Valley way, we've seen this forever, Kara, is that let's get, you know, a few smart guys and there are always guys in a room and we'll figure out AI is different. It's, I mean, first off, it's not just computer scientists, it's math, it's linguists, but where are the sociologists and philosophers? Where is there a diverse group of people? Like this stuff is powerful, it's reflecting humanity. We're gonna increasingly end up relying on it,

and everything for education, social life, business and all. And so that's my biggest worry. That same mentality, Sam Altman and a few smart guys are gonna show up at his mansion in San Francisco and they'll figure it out because they're smart. Yes, they're smart, but it's more than just being smart. Two more things before we have to go.

The costs are so high, but Deep Sea had scared the crap out of everyone. This is a Chinese AM model trained for less than $6 million. That's questionable if that was actually. The costs are so high, but Deepsea had scared the crap out of everyone. This is a Chinese AM model trained for less than $6 million. That's questionable if that was actually, but they called it a Sputnik moment. If there were more efficient models, Christy, with more chips, bring down the overall price of computing power, how would it impact small startups like you invest in? Well, I think there are a lot of parallels here to even the tech industry today because what, 15, 20 years ago, you had to buy big servers and big

computers and have all this infrastructure and now it's accessible. So I think in the same way these tools make the playing field a lot more level and I think we'll be surprised with the level of disruption really across the board, not just in pure tech, but whether it's energy, whether it's life sciences, whether it's adjacence. I think life sciences, whether it's adjacent, I think there's a lot of room for innovation. And I think it's great that the playing field will be much more level. So I want to finish up with two things. One is this threat to research that we talked at the top. I'm going to ask you, Gary, I'd be happy to put you on the spot,

Christy. The federal government is one of the biggest funders of university R&D. A lot of Johns Hopkins money comes from the National Institute of Health, for example, about $1 billion in fiscal 2024, more than any other institution in the US. A lot of the money has been canceled, not just at Johns Hopkins, but elsewhere is in limbo. Why don't you talk about this, Gary? You can talk about it, Christy, but what is the impact right now on research at places like Johns Hopkins if the Trump administration doesn't reverse course on this? So it was the way it happened.

To me, it's like they bent over and unplugged the machine. And just so there's all this wasted stuff. And what really scares me, I was out in Silicon Valley a few weeks ago talking to people at Stanford. And like, it's not like you could just, oh, okay, this was this four year period. Now let's just revive it.

I mean, that to me is the big worry, that a new administration would come in as more foresighted, like, you know, thinking like, this, we really need this. And it's not going to be so easy to just like, oh, okay, well, we lost a few years, but now we're going to go back right back to where we were. How do you deal with it? Look, I think there are just a couple things I would say here. I think what the NIH has done over time with very competitive grant funding universities and ultimately to the commercial sector, this incredible group of three sectors that came

together have lowered the deaths of heart disease and stroke by 75% in the last 40 years. When I was growing up, HIV and AIDS was a decedent, and now you can live a normal lifespan thanks to NIH research into universities and to the private sector. And so I think we want to continue that. My greatest fear is for this incredibly talented generation of scientists that are going to choose necessarily not to do that work. If their grant funding is cut off, if they're unable to continue their work, they'll go do something else. And what does that mean for the next decades of that type of progress? Irreparable harm? I don't know.

What they didn't invent, we don't know. Correct. All right. Last question. Speaking of chilling effect, Gary, at the end of the book, you write AI was of course a bubble, but bubbles are as much a part of the Valley's boom and bust economy as underaged cocksure founders and the VCs who fund them. Used the past tense.

So was it a bubble? Did it burst? Is there a possibility of another AI winter that some commentators have been warning about since last year? And Chrissy, how do you feel about the bubble and the winter? Yeah, no, I don't think the bubble has burst. There's still tons, you just mentioned a stat,

that 150 billion or whatever it is in venture capital going in AI. I mean, I think what we're gonna see is a natural cycle where a lot of these companies are gonna go out of business, but some will break through and will go on to have kind of long, great lives for a company. But it's mountains and valleys. It's like, it's gonna peak and then there'll be

a slower period. I don't agree with people in Silicon Valley who say like, oh, we have artificial general intelligence in the next year or two. I still think we're a breakthrough two away. They'll say they have AGI because their venture capitals need them to say that. And it's an amorphous definition kind of thing. But now I think we're still in the middle of a prolonged fruitful period for AI.

For the VCs to get their return on investment here, we need the capital markets to open up, which means we need certainty, which means we need great investments. And the moment that a number of these investments go bad, it's not going to be great. I've been part of four downturns in my career. And you never know exactly when the top is, but you've got to hope that that things go well, at least for a couple bellwathers. Are you worried about that? What would make you worried? I mean, if if companies aren't able to have exits, then it's hard for the VCs to tell their LPs, yeah, let's pour more money in. So last question, let's end on a positive note.

If you could do one thing to fundamentally secure innovation in the AI space for the next generation of entrepreneurs, what would it be? Look, I think there has to be a combination of thinking about the products, the returns, and also the policy. I mean, look at this center. This is where we can really be thoughtful. The companies don't necessarily have the best interest in humanity in mind. And so let's come together and figure out what the policy arguments are, what the future

of that investment would be. So I think having... What would be the key policy, you think, if you had to... I don't know. I'd have to ask our policy experts on that. I do think there just needs to be a few ground rules that the Biden administration, I thought, had a gentle policy like, hey, before you release these things, these powerful models, we're going to require you to test them and share the results with us.

And Silicon Valley really rebelled against that. Sam Altman was all for it until the Trump administration. Now he's all against it kind of thing. But I do think that we could put in some pretty modest regulations that make sure that the general public is there with these AI companies. I'm convinced there's something bad's gonna happen. I don't know what it is. A trillion dollars siphoned off

from the global economic system, money system, before a human even realized it. Whatever it is, and then people are gonna panic and hate AI. And that's what I really want to prevent because I'm with you, you know, education, healthcare, science, I think extraordinary things can happen with AI, and I'd hate for there to be another AI winter because of the carelessness. Are these the critical years right now? And who's the most important person in this? The most important person? I would say policymakers in DC. I don't know if it's a single person.

Yeah, I think these are the formative years. This is really going to shape people's idea of AI. And again, people are mistrustful of AI. And I'm scared that that's going to really slow things down, which would be a shame. Kristi, the most important person right now or company? Yeah, I don't know that there's any one, but I think just keeping an eye on the promise and the peril. All right.

On that note, thank you too. Thank you. Thank you. On with Kara Swisher is produced by Christian Caster-Russell, Kateri Okum, Dave Shaw, Megan Burney, Megan Cunane, and Kailin Lynch. Nishat Kherwa is Vox Media's executive producer of podcasts. Special thanks to Eric Litke. Our engineers are Rick Kwan and Fernando Arruda, and our theme music is by Trackademics.

If you're already following the show, you are keeping an eye on the promise and peril of AI. If not, you may be technically broken. Go wherever you listen to podcasts, search for On with Kara Swisher and hit follow. Thanks for listening to On with Kara Swisher from New York Magazine, the Vox Media Podcast Network and us will be back on Thursday with more.