Marketplace - Let's tour the growing AI economy

Episode Date: January 27, 2026

Artificial intelligence companies raised enormous amounts of money in 2025, and made major investments in development and infrastructure. What’s next? To understand more about the role AI could play in our futures, “Marketplace” host Kai Ryssdal and “Marketplace Tech” host Meghan McCarty Carino visited an AI company and an AI data center in Silicon Valley. In this episode, the next phase of AI innovation won’t come without obstacles.Every story has an economic angle. Want some in your inbox? Subscribe to our daily or weekly newsletter.Marketplace is more than a radio show. Check out our original reporting and financial literacy content at marketplace.org — and consider making an investment in our future.

Okay, so they call it artificial intelligence, but the AI economy is very, very real. From American public media, this is Marketplace. In Los Angeles, I'm Kai Rizzol. It is Tuesday, today January 27th. Good as always, to have you along, everybody. We're going to spend the program today recapping a quick trip I took a couple of weeks ago. Good morning. Hey, Kai. How are we?

How are we? Welcome to the Bay Area. Thank you very much. That's Marketplace's Megan McCarty Carino. You hear her here, and you hear her on Marketplace Tech. She was my ride at the airport up in San Jose. Tell me why I'm here. Well.

You made me get out of plane at like 8 o'clock this morning. You want to talk AI. You want to talk AI and money. You got to come to the Bay Area. Right? Fair enough. We started our series the other day, trying to figure out whether AI is getting too big to fail. Because there is massive spending happening, massive borrowing,

to finance the data center boom. I went to one in L.A. Megan talked to people about their not so good feelings about our AI future and the tradeoffs that come with it. But the thing is, the AI future is already happening in Silicon Valley. There's a lot going on here. There's data centers. There's energy.

I want to back up for a minute. What does it feel like here, AI-wise? I mean, all the rest of us hear about AI is, oh, my God, data centers and power and water and, you know, all those things. Chad GPT and blah, blah, blah. What does it feel like here in Silicon Valley? My answer might be a little counterintuitive. I mean, it's definitely like on everyone's, you know, the tip of everyone's tongue. You see it everywhere you go.

The billboards are everywhere. We passed a couple of those billboards literally just outside the airport for both biggies like Google Gemini and for smaller companies, neither Megan nor I had a ever heard of? In terms of like the boom on the ground, I grew up here, I've been through a lot of these cycles. Sorry, so let's back up. You grew up here, which I did not know. I probably should have, but I did. So you were here, not that you're necessarily old enough to remember this, you were here in the dot-com box. Oh, I remember it. I remember it well, yeah. I mean, that was probably the most dramatic time, you know, it was just sort of like overnight, you know, houses in my parents' neighborhood had luxury vehicles in front of them, you know, went from kind of a normal middle class neighborhood.

And then there's just millionaires everywhere, you know, BIPOs are popping off. So is it that yet for AI here? That's what I would say there's kind of a disconnect. So the local economy here is actually not as thriving as it has been in previous booms. And even just before the pandemic, there are certain metrics that you can look at, like the number of tech jobs in the Bay Area is still lower than. it was before the pandemic. You know, we hear about layoffs all the time. The same companies that are spending these billions of dollars on AI are doing layoffs all

the time. You know, we just heard about meta just announced some more layoffs. And so I've heard from a number of people that the job market in tech here hasn't felt this brutal since the dot-com bust. That's really interesting, right? Because during the dot-com bust, it was obviously bad here. I was here as well. And so, you know, I've got some history, but not like you.

You do. But the dot-com bust being bad makes sense, but we're in the middle of the bubble of AI, man. This is the boom time. Right. Where are we going? What are we doing? We should say, by the way, you're not driving as we're talking. Other people are driving.

No, I would not be able to engage in such complex idea. So where are we going? What are we doing? Well, we're headed to an AI startup called Tiny Fish AI, AI, which we thought was. an appropriate place to check out. You know. There are literally thousands of AI startups all over this country. According to Pitchbook and the National Venture Capital Association,

AI startups got $222 billion in venture capital investment in 2025. That is 65% of all U.S. VC dollars that year. Tiny Fish, the company we're visiting, got about $47 million worth of that. And yes, that is small potatoes compared to the big food. like Anthropic and Open AI. But for artificial intelligence to pay off in the way that investors are betting it will, it's going to take a whole ecosystem of companies using it. And that's why we came here to a kind of suburban office park in Los Altos.

It looks a little bit like a Chipotle, though, with those umbrellas and stuff. Indeed. That's what I'm saying. We walked into an office with a fish tank in the lobby, and we met Sudizh Nair, the co-founder and CEO of Tiny Fish. I'm Kai, how are you? I'm Kai Sudish. Nice to see you.

Nice to see you. I'm welcome to time. Megan, very good to meet you in person. The company is about a year and a half old and trying to fill a very particular niche in the AI economy. And the problem is with that. AI is becoming more and more powerful and value driven. The chasm between the large and the small companies are expanding.

We are intentionally trying to see if it is possible for us to bridge that. So that's what we are doing. What specifically are you doing? Actually, I was thinking of showing you if you're interested. Well, yeah, maybe that's good way. Yeah, absolutely. Let's go show. Show us. Software is really hard to do on the radio, but we're going to try.

Sudish walked us past a dozen or so people coding away on their computers. It has a very tech-start-y vibe. And he grabbed the laptop with a hotel booking site pulled up. Google Hotels is our customer. So far, we've been talking about the physical infrastructure behind AI, data centers and networking technology and power plants. This company makes software infrastructure for sites like Google Hotel, and Expedia, which pull live pricing information from hotel websites.

Marriottes and Hiltons and Hyattes, they are so technically built that they will have the stack to optimize for all of this. Now think about tens of thousands of these small properties. All the tiny fish, if you will, the small hotels that don't have as much technological know-how as the Hilton's and the Hyattes of the world. They don't have APIs. They don't have access. And the more AI gets integrated into the web, the harder it's going to be for those little fish to be seen when you go searching for a hotel. I'll just run this. Okay. This is a very small hotel in mountains of Japan, and it only has eight rooms.

It's owned by a family. Availability and pricing information for places like that sometimes doesn't show up on the big hotel aggregator sites because their websites just aren't sophisticated enough. Tiny Fish is trying to solve that problem with AI software that can pull information from anywhere, even from a PDF on the underpowered website of a tiny Japanese hotel. This means that this hotel can show up on Google and become just like a Hyatt or Hilton. The bet that investors are making in startups like Tiny Fish and thousands of others is that even if you don't necessarily know you're using AI when you're booking a hostel hotel are making a doctor's appointment. AI software is going to be working for you behind the

scenes. There are so many, you know, small and large and medium-sized startups out there that are all kind of feeling their way through this AI boom. Clearly, there are productivity gains to be had. There are, you know, new conveniences and how we use the, the internet and, you know, use different tools. I know that some of those companies are maybe not going to make it through. You know, it's, Really important question. I'm going to be totally honest about the fact of, I think about this all the time. The thing about startups, you never know when the fog will lift. It is a question of grit in the sense that you are going to get punched.

You will feel like, you know, there is no future. What do you do after that? Like there is, it's constant course correction. I've been talking to a lot of people in areas of the country where these hyperscale data centers are being built. And I hear from a lot of people, you know, wariness about the tradeoffs that are implicit in putting these kinds of facilities in their community. The, you know, electricity prices are going up, but they're worried about water. And they say, what is the benefit to me? I don't use these tools. Maybe, you know, who knows they may not travel a lot, they may not use a tool like

yours. What would you say to someone like that who's worried about the tradeoffs of this technology? There is, there's also, usually data centers come together where three things are coming together. The first one is electricity. Second is fiber optics. And third is water. So smaller communities are going to be really affected by this because this is like in sort of invisible fracking. The value that we are actually going to share will reach them. But it is going to be, for example, if we find a new cancer drug.

Which, fine. A new cancer drug would be great, but the trade-off between a new medicine that saves lives years from now and the very real cost of your power bill going up right now is a tough sell. But you see, like, it is the same investment funds, you know, sovereign wealth funds and others who have used to invest in tobacco and gasoline. They're the same money coming into AI. So this is not a noble dollar, but we have amazing PR because we control most of the distribution of information now. So, you know, we have the ability to paint a picture that an oil company never had.

You say that like it's not a bad thing. It's a thing. I don't know that or good, but it is what it is. We have so much data and then distribution that press the button, 300 million people are immediately learning that. Think what we could do with that as an industry. Good and bad. That industry is building a whole supply chain of infrastructure,

both hardware and software, trying to change the way this economy works. Coming up. There is a kind of insatiable demand for data center capacity wherever they can get it. That's after the break. But first, a Google AI summary of search results for The Numbers says this. The segment is best known for its use of specific musical cues to signal how the markets performed that day. So let's hit it. Industrial's down 408 today.

810% closed at 49,03. The NASDAQ gained 200. $415 points, about 910% 23,817. The S&P 500 picked up 28 points. That is 410% 69 and 78. There, that is a record high, by the by.

So a couple of prominent AI companies on Wall Street, shall we? InVito, which designs chips that power AI, picked up 1.1% alphabet with its Google Gemini product, rose 410% Apple, which calls its AI effort Apple Intelligence. Should have done better, honestly.

Packed on 1.1% Microsoft with its co-pilot thing, which always pops up on your outlook when you don't really want it, lifted 2.2% on the day. A lot of businesses and organizations surveyed by McKinsey recently said they are still in the experimentation phase with AI, but 62% said their organizations are at least kind of goofing around with AI agents to see what is possible. 32% expect workforces to shrink as AI adoption quickens. 43% expect no change. Today, by the by, interest announced Pinterest, rather, announced job cuts,

about 15% of its staff. What did they cite as a reason? Say it with me now. AI, Pinterest down 9.5%. That Google AI summary of the numbers also says, and this is a quote, I know I'm quoting AI. I appreciate the irony. Depending on the news cycle, the segment may also include 10-year treasury bond yields. Do not hallucinate on this one, AI. We almost always include the tenure, which today, prices fell, yield rose, 4.23%. you are listening to Marketplace, which is made by humans for humans. This is Marketplace. I'm Kai Risdon. After Tiny Fish, Megan and I pulled up and parked outside a new and huge white building. It looked pretty deserted, no parked cars, just a single security guy who, after the obligatory

ID check, directed us into the lobby. Hi, Colin. How are you? Megan, good to meet you. Kai, Rosdahl, nice to see you. How are you? Pleasure to meet you. Thanks for coming. Yeah, you've been. Thanks for having us. We're really happy y'all could make it and we could host you. Yeah, we're excited to check it out.

Let's first things first. Tell us who you are and where we are. Great. So Colin McLean, Chief Revenue Officer, Digital Realty. Digital Realty has more than 300 data centers worldwide. It's one of the biggest data center providers out there. We actually passed another huge digital realty building while we were driving in just two blocks away. The company is more than 20 years old, which is another way of saying it's been building and leasing data center space for way longer than this current AI business. Boom. And we are in... Oh, excuse me. We are in San Jose at 641 Walsh, which is our flagship new data center that we're opening here in the heart of Silicon Valley. All right. I feel like you guys are usually cage you about giving out addresses of data centers.

Well, I mean, we're really proud of this data center. This data center, though, kind of isn't one yet. Digital Realty broke ground back in 2021. And to the untrained eye, it does look about ready to to go. The whole place is 430,000 square feet, four stories high, with an identical data hall on each floor still empty for now. All right, let's go. This place is so new, man. The whiteboard markers aren't even unwrauded. That's right. That's right. Come on through. This place is a lot like the Los Angeles Data Center we were in yesterday, just quieter without the hum of 48 megawatts plugged into it. Colin led us down a hall with industrial pipes and huge swaths of cables running all along the ceiling.

What you'll see is we've got some more work to do on mechanical and the electrical side. It looks like it could be a storage warehouse, honestly. A massive room with gray slab floors, pretty bare bones everything, where cages holding servers are eventually going to go. I'm just doing a little math here. So there's a 430,000 square foot. So give or take this is 100,000 square feet this space, cluster minus? Now a big empty room.

How long until it's literally humming with activity? I think the go live on this is 2027. And you're waiting on what? It's a mix of things. Obviously, we have to work very closely with the municipalities and the utility grid when power is delivered. So our job is to make sure we work with SVP, Silicon Valley Power in this case, to make sure we deploy the infrastructure, the timing, exact the right time where the power is delivered.

And then we turn that up based on when that is available. So you can't do anything until you get electricity in here, right? It's a challenge if you don't have power for a data center. So yeah. So we have to wait for the timing of that. It's very much not a demand topic. There is a kind of insatiable demand for data center capacity, wherever they can get it.

So is the availability of power an issue anywhere else besides here? I think power, it will be the issue moving forward as it relates to data centers. Power is already the issue. One study from Carnegie Mellon and North Carolina State found that nationwide energy costs could rise 8% by 2030 because of data centers. And in some parts of the country, like Virginia, which is the AI data center hub, that cost increase could be more than 25%. So what's a bigger concern? for the data center business right now, the idea of, you know, say a bubble pops and demand decreases, or just the challenge of meeting the demand that you have now with the constraints in electric transmission?

Yeah, that's a good question. I mean, I think those are both challenges and opportunities, quite honestly. I mean, at the moment, demand is one where it's pretty frothy. We do see a continued trend of more consumption data as becoming more and more valuable. It's almost like the new oil, if you think about it, in terms of value proposition. We've been really consistent in saying that, you know, we see a real home in need for AI, but we also see a real home in need for other work streams that are going to come as a core part of our client's requirements. So you're maybe a couple, maybe a few years from this sort of being operational, but at what point do you start leasing to customers?

Sure. And also, what's the sales pitch? Sorry. Just, you know, give us a sales pitch. I guess, sure. No problem. So depending on the client that we're talking to, the conversations can be anywhere between, you know, 12 or 14 months out in advance of actually securing capacity. Part of that's going to be context of, you know, what they're actually looking at to do in the facility.

We can support the high-density-oriented use cases or AI, or we can support, you know, large capacity chunks of more traditional-oriented deployments. So why digital realty? So digital realty is the largest data center provider in the world. we feel like our strongest position is that we have the ability to support the entire portfolio of requirements. So if you need a cage or you need a data hall or you need the entire building, we feel like we're uniquely positioned to support those requirements, whatever they may be. This is kind of weedy, but how much is it going to cost me? How do you, is it's not, is it like a square foot thing?

Is it a per megawatt thing? It's per KW. So we charge you a rate, yeah, per kilowatt. So based on how much power you consume in a data hall. Power really is the currency. No joke, right? That's right. No joke.

Legitimately. I have to say, though, after being at the core site data center in L.A. the other day, the one in the old post office building, all these rooms kind of look the same to me. This is called a Meet Me Room where actually the magic happens. Sorry, Meet Me as in. Meet Me Room. Meet Me Room. MMR is what we call it.

Yeah. Looking at that pipe over there, that, I don't know what, a 10-inch pipe that says chilled water return, there are people who are going to hear this who are going to hear data centers, man. they're using all the electricity, they're using all the water. Sure. Yeah, no, we understand, and we want to be good partners to the community. We feel like the data centers add quite a bit to the digital economy and the local economy. That's just not in form of tax dollars, but it's that form of jobs and how we're supporting

infrastructure needs. So we get that, and we spend a bunch of time working inside the community to make sure we're as good a partner as we can be. Honestly, I'd have expected a more robust answer, because that's the, the criticism of data centers everywhere. Sure. They're sucking all the electricity, and they're sucking all the water. Yeah.

And of course you want to be good partners, but it's a valid critique, no? Well, I think it's something that probably requires a more in-depth conversation. I mean, if you look at it from a statistical, one of you, from what we've seen, is the consumption, as consumption actually grows within data center communities, the cost of power actually reduces. So we have seen actually a positive benefit, net benefit, of data center. stepping into the community. For sure, we have to be a good partner in that equation

and that we have active conversations with communities to do so. But on the whole, we very much feel like that we're in that ad. There are some cases where data centers can, in theory, bring down utility costs. The catch is that those theories require places with extra energy capacity. But big companies are betting that with all this extra demand for power, energy supply is going to find a way to keep up with new technology and innovation that will eventually bring down costs. That, of course, is not where we are right now. The biggest AI name in town is just a mile away from the digital realty building.

So Megan and I drove over and parked at a construction site near Nvidia HQ. They're probably building another building, right? I mean, almost certainly they're building another building. Also in our line of site, along with Nvidia and the construction site, were two other data centers run by two different companies. Everything is data center. Unbelievable. Seriously.

Yeah. Between us, Megan and I have been to a total of three data centers now. Talk to me for a second, though, about the people who get left behind because, you know, you're in Silicon Valley, incredible concentration of wealth. Same with L.A., and New York and, you know, Washington, D.C. But there are huge chunks of this country where people don't even have broadband Internet yet. Forget AI for crying out loud. Yeah, you know, and I spoke to a lot of people in rural areas where these big hyperscale data centers are going in.

Many of them, you know, we had to talk on the phone because they're like, I don't do Zoom. I don't do video conferencing. And they are asking the question, what is in this for me? That's why I came up here to see what the AI economy looks like on the ground and what it's going to mean for you. Even if you don't use Claude or ChatGPT, AI is still part of your life. There could be a data center you never knew was there, or you could be invested in companies benefiting from the AI boom through your retirement account without even knowing it.

AI and its tradeoffs are here to stay. But how those tradeoffs balance out is the question we don't have an answer to quite yet. This final note on the way out today, if you want to see some of the places we've been like that big, empty data center. You can do that over on our Instagram. It's Marketplace APM. Also, as we mentioned, Megan's been doing a whole lot of reporting on AI infrastructure over at Marketplace Tech. We're going to take kind of a trip down memory lane. Like going on a vintage manhole cover hunt for a look at technology booms of the past.

So we're just going out on the street. We're just going out on the street real briefly. Also visiting a place that has been called the Winchester Mystery House of the Internet. It's kind of spooky down here. And trying to figure out if electrical vehicle trash can become AI data center treasure. I drove an EV for quite a long time. I don't think I ever pictured what the actual battery pack looked like inside my vehicle. Yeah, they're not that beautiful objects. They're just kind of a little black skateboard.

It's not a high design object now. With a bunch of red and black wires coming out of it. But this is really the core of an electric vehicle. You can hear those stories. on Marketplace Tech wherever you get your podcast, just follow us there. Jordan Manjee's Anil Maharaj, Janet Winn, Olga Oxman, and Virginia K. Smith are the digital team around here. I'm Kai Rizdao. We will see you tomorrow, everybody. This is APM.

Ready to make the most of your money? Sign up to receive weekly tips from Marketplace designed to help you make smarter financial decisions. Plus, you'll also be the first to know about exclusive Marketplace merchandise and local events. Text Marketplace to 80568 to sign up.