This is "Wall Street Week." I'm David Westin bringing you stories of capitalism. Local resistance to data centers has popped up across the country and even around the world. We take a look at one place that believes it's turned the quest for more compute to its advantage. Plus, where is Europe in the AI race? We talk with the co-founder of Mistral, the French company carrying the torch for the rest of Europe. And a tale of two Argentinas under President Milei: one thriving and the other falling behind.
But we start with the business of elections in the U.S. With just over a week until the nation votes on control of Congress and the governorships of 33 states, enormous amounts of money are pouring into campaigns, parties and PACs. Where is it all coming from? And what difference is it likely to make? Blair Effron of Centerview Partners has been involved in democratic fundraising for years, and he says at least one thing is certain: elections aren't getting any cheaper.
The cost of these races has gone up exponentially. A Senate race is going to be $200 million plus, a congressional race, which used to be $5 million, $10 to $15 million, easy to spend. And what I would say is that, believe it or not, the most money still comes from small dollar donors. That goes direct to the candidates. What you see a lot from the business community is going to PACs. I think most people who have an interest in things policy and things D.C. are figuring out how to be helpful to both sides.
Does the money matter? I mean, in this sense, we've seen some races, for example, the primary in Michigan, my home state of Michigan, El-Sayed against Stevens. And there was a big difference in the money, but it didn't translate into the votes.
It still comes down to the candidate. If you don't have the right candidate, I don't care how much money you spend. And I would tell you this, David, the amount of money being spent, you and I can agree is so large that at some point there's a saturation point like there is in anything. My bet is that for most of these races the funding piece has gone past the saturation point and it's going to come down to the talent of the candidate, and that's who voters are thinking about.
The U.S. business community and the economy has changed a lot in recent years. Are we seeing money coming from different sources than we did? I mean, 10 years ago you would not have had hyperscalers and crypto.
Ten years ago, the tech community wasn't involved with politics at all on either side. Clearly, anyone, any industry that has a stake in the regulatory regimes that come out of Washington has an interest in giving. And the AI companies have been some of the biggest contributors to both sides. Doesn't surprise me. They're certainly in the sights of a lot of voters, a lot of people in D.C., in terms of what the regulatory regime ought to look like.
Those of us, at least some of us in the media, talk a lot about backlash against big money. And you've seen some rallies, actually, with AOC and with Bernie Sanders about ballot versus billionaires and things like that. Does that actually translate into the election? I mean, can it be a negative effect to raise a lot of money from very wealthy people?
Well, first of all, I think it absolutely can be. If you are deemed as buying something. If you are deemed as putting in your own personal means in a big way. You want to see a broad base of support. So the idea that I started off as saying small donors still make up the vast majority of giving, it's a big deal. And I think that there is nobody on either side of the aisle that would argue that there's too much money going into the system. And obviously the Supreme Court said it was OK. And I think we would do well to figure out how to scale it back. But it's hard to figure out how that ever happens.
Are you concerned at all about the big dollars crowding out the small dollars?
I am always concerned when instead of using the word "dollar," "voice." I'm always concerned when someone's voice is more important than someone else's voice because of what they've done financially. There shouldn't be any room for that. Yet we know there is. And I think that we ought to be shining a light on what it means to give a lot of money. And then what happens in the year following in terms of how your business is treated. Now, I would tell you most businesses I know are not spending that kind of money. They're balanced and, frankly, a lot of them stay out of it. OK? It's not—they want to stay in their lane. And the highest ethics and they are willing to make their bet with D.C. based on the power of whatever argument they're going to with the regulatory regime. But I do think the amount of money generally coming from small pockets of Americans is troubling for anybody.
President Trump, whether you think it's great or it's not great, has changed the relationship between contributors and the White House, has opened up the doors. We see it all the time with visits to the White House. Do you think he's changed it for the long term? Has he really taken away some of the resistance of corporate America to really becoming much more closely aligned financially with politics?
I think you're absolutely right. I think that in politics, generally, when something happens and something moves forward, very rarely does it move backward with the change of office. So I do think some of the rules, some of the giving we're seeing, I don't think it slows down. And I'm not sure that a candidate, be it Republican or Democrat, is going to try as opposed to do what it takes to stay competitive. Donald Trump has changed the game on what giving is. And, to be clear, I think the game will be changed in terms of transparency. The game will be changed in terms of doing everything according to the letter of the law. But in terms of the amounts being raised, I just don't think that goes backwards.
This is an exciting election from a campaign finance perspective. Maybe too exciting.
Brandice Canes-Wrone is professor of political science at Stanford and a senior fellow at the Hoover Institution.
We seem to be on track to increase. Like other recent elections, there's a larger role for super PACs. That's part of the increase and they've been growing. And then there's the recent June 2026 Supreme Court decision that's allowing the parties to coordinate more with candidates explicitly.
We have AI, goodness knows, as a major factor today. And for that matter, crypto as well, by the way, also sports betting is more active than it has been in the past. Do we have any sense of whether those dollars tend to go to the Republicans versus the Democrats?
We do have some sense. It's interesting on AI, for instance, while, on the whole, there's a lot of money going to Republicans, certainly more than you'd think of the tech industry going 10 to 15 years ago, it is more bipartisan than may meet the eye. So if you were to look at, for instance, Anthropic's contributions of its employees, those do tend to lean Democrat. And there are super PACs on both sides. So Leading The Future is a very active super PAC that's kind of less hard on the guardrails, for instance, on AI, and Guardrails Alliance is a super PAC that, as its name suggests, is in favor of stricter guardrails on AI. Even Leading The Future, however, is making bipartisan contributions. They've been very active in Democratic primaries. They've also made contributions to Democrat and Republican candidates. It leans more in the Republican direction, but this isn't a hard 100 percent, you know, to zero by any means. The Democrats, of course, are likely to win the House. They may win the Senate. So you could say that organizations like Leading The Future are behaving as we'd think of, you know, traditional finance companies behaving. They're going to be looking at who's likely to win. They're going to be looking at having candidates from both parties that are going to represent, as they see it, their industry's interests.
Looking forward to after the midterms, is it too early to speculate about what, if any, policy differences may come about because of these elections?
I don't think it's too early. So, first, one thing we know about money is where it has the largest effect is in the primaries and actually also in referenda. It does have a marginal effect in general elections. No political scientist worth their salt would advise a candidate, "Don't worry about money. Don't, you know, don't raise, let your opponent go crazy and don't do any—don't advertise." But, of course, many people vote for the party they always vote for. So the effects are going to be biggest in these contexts that don't have partisan cue. A kind of amazingly in California in a democratic—what's likely to be a democratic wave election, the new wealth tax may not pass, which is a referenda issue on which there's been a lot of money. It's also a tax that the Democratic governor Newsom has opposed. So this is not—while it's largely a democratic proposal, it's not that all Democrats have come on board. So there are a lot of factors at play. But certainly advertising about the effects that the tax might have on, you know, the mobility of those who would be paying this wealth tax is having an effect on, at least, the polling. I guess we'll see other results.
As I mentioned earlier, a lot of these newer industries have been very active in primaries, and you see some cases, for example, in New York, in the 12th District, Bores loses, who was very much in favor of heavy AI regulation, and those types of examples—and this was a case where AI, kind of anti-heavy AI regulation super PACs were very involved, and those types of primary victories are going to have an impact down the line on kind of who's serving from both parties on issues like AI regulation. Crypto was very involved in Alabama's Senate race. Felix Barry Moore, who is a big proponent of federal regulation that's advantageous to the crypto industry. So I would actually, instead of thinking about the general election, I'd say the answer is yes.
And in the interest of full disclosure, Michael Bloomberg, founder and majority owner of Bloomberg LP, has contributed some $80 million to Democratic candidates in these midterm elections.
Coming up, data centers have gotten a bad name in communities across the country, but one county in northern Virginia has embraced them with some strings attached.
This is a story about making sure a deal is good for both sides. Between March and August of this year, data center projects tripled in the U.S., and the hyperscalers putting up these data centers are certain they're good for their businesses. But many communities are questioning whether what is good for the companies is also good for them. Many communities, but not necessarily all.
What it means for a community is jobs and lots of money and lower taxes.
For his part, President Trump has little doubt that data centers are a good deal for America.
If you want to create tremendous wealth, if you want your house to be more valuable, and everything else, you're going to want a data center. If you want to go through poverty, crime, and squalor, I would say, don't approve data centers. You have your choice.
And some people living closest to the centers agree with Mr. Trump. In Ashburn, Virginia, Paul Serrano walks out his front door every morning to see a data center, and he isn't complaining.
The value of my property has gone up by almost two thirds since I had moved in here, since 2013. But the property taxes have been almost flat. And the reason for that is because the tax rate has been going down steadily. In fact, if you factor in inflation, the real dollars that I pay for property taxes are 12 percent less than they were 13 years ago.
Serrano lives in what's known as Data Center Alley in Loudoun County, just next to Dulles Airport in Washington D.C. It's an affluent county, historically known for its horse country and rolling hills bordered by white fences. But in recent years it's become famous for its abundance of data centers. So many that they're expected to generate about $1.3 billion next year. About 40 percent of the county's total tax revenue.
Just launched a new high school last year. $144 million high school with all the bells and whistles. Plus, we've been able to add a lot of recreation facilities. We think that this is a very livable community, and a lot of that is thanks to the tax revenue that we've been able to get from the commercial sector and the data centers.
I was a rock and roll disc jockey who came into a job trying to figure out what I was selling, right?
Buddy Rizer is the executive director of economic development in Loudoun County. A former radio DJ, he moved to the county in 2007.
Well, you know, everybody grows up wanting to be something. And I grew up wanting to be a disc jockey and was lucky enough to get in the industry and really wanted to get back into helping the communities that I was part of. And that was how I kind of stumbled across economic development and been here in Loudoun since about 2007. We had the airport, but we were a bedroom community largely dependent on residential real estate for our tax base. In fact, 81 percent of all the revenue we received was from residential real estate. When the housing bubble burst, we lost a lot of our revenue, and that was the real entry point in which I joined Loudoun County as an economic developer. We needed to find new revenue sources, and that needed to come from the commercial sector.
Loudoun County had been a participant in the dot-com era. We had really, because of AOL and Unanet and a lot of different things, we were really building a lot during that late 90s, early 2000s. So when the dot-com bubble burst, we were left with a lot of assets. Namely, we were left with a ton of fiber, we were left with some great talent and some buildings that were built but never occupied. So, after a little research, I decided, you know, let's pitch these empty buildings to Amazon and Google and Microsoft and Digital Reality and all the people out there that were really starting to build on the Internet infrastructure. We kind of said, you know what? I think we have something. And, certainly, it's turned out pretty good for Loudoun County in a lot of ways. When you drive through here, you see the roads are just amazing. You know, beautiful landscaping.
Rizer says the county has about 230 data centers with another 20 or 30 coming. But even a good deal has its limits. In September, following increasing pushback from locals, the county moved to pause approvals for some new data centers.
We're at the end of our life cycle. You can no longer build data centers by right here in Loudoun County. There are some that are already permitted and will be going forward. But, you know, until the energy situation is sorted out and we get a better understanding of where we are, we're not really looking to bring new data centers into Loudoun County. And I think that's OK, right? I mean, I think there's opportunity for everywhere. And I think we've been able to take advantage of the opportunity we've had. But that doesn't mean that we want them everywhere. We want to be very thoughtful in our future growth.
So what is the difference between a good deal for all concerned and one that favors only the big tech companies? And what can other communities learn from Loudoun County?
Well, places have more leverage than they thought. It's turning out. And I think that is how they will get better outcomes by asking for it.
Mark Muro is a senior fellow at the Brookings Institution and authored a paper titled "Turning the Data Center Boom into Long Term Local Prosperity."
I think the important thing is that the data center in itself needs to be part of some plan for the future, some strategy for regional identity. Another problem that I think communities face is many of them don't have thoughtful strategy for the future, a sense of what they want to be. So that needs to be in place. And with that, will come requests. Communities should be able to and ready to negotiate in a tough but fair way and try to extract enough to make these deals valuable to them. So it might be negotiating a compute partnership with the local university that has, you know, computer science needs. Maybe it is about housing. Maybe it's about workforce training. Communities need to lift their aspirations. They should not be afraid to ask for much more here. And I think that's been part of the problem. They've been cowed. There's been a race fear of missing out, fear that the data center plum will go to the next county over or the next state over. And therefore there's been a fear of asking for more.
I would say that I don't go a day without hearing from somebody and somewhere in the world who wants to pick our brain and see how we can give them advice on what we've learned. I think that there are things that we have learned along the way as far as how do you cite the buildings and what kind of engineering are you looking at and how do you make sure that you're using resources that are available and not taxing the resources that aren't. But I think that it's also about, you know, where do you put them, how do you make sure that they're not near residential. How do you make sure that they're in the locations where you want them. So that's why I say define the opportunity. Decide where you want them and say, you know, you want them in this industrial park. And then you want to make sure that the infrastructure is already there and you want to make sure that everything you need is in the right places because you can build this in a way that is not overly impactful for citizens.
But if that's what the local communities can do, what can the data center developers do? Equinix is an Internet and data center company based in California. Adaire Fox-Martin is its CEO.
We operate over 281 data centers and we do that all the way around the world. And one of our differentiating prospects is because we're focused on connecting businesses to businesses, we tend to build in metro locations so close to where people are. And our facilities are not enormous facilities at all. The average size of our data center is about 10 megawatts, with the largest being in the 35 or so megawatt category in the U.S.
So data centers were necessary and facilitated the Internet. Now we hear about them in connection with AI. How much of the data centers that we're talking about are connected to AI as opposed to the rest of the Internet?
Well, in our case, the vast majority of the work that we underpin is work that kind of makes possible very, very ordinary daily things. You know, so for example, if you've bought some stocks today, you've probably traversed our platform. If you've had a video call today, you've probably traversed our platform. If you've accessed your health care app or your banking app, then you've probably traversed our platform.
Do data centers have a public relations problem?
I think there is an education problem. You know, if I think about Equinix and the kind of things that we enable on a day-to-day basis, I'm not sure that that is widely understood by everyone in the public domain. The things that data centers make possible, and just like not every single building on a landscape needs to be a skyscraper, not all data centers are built for the same format or the same function. And so I do feel that the industry has a duty to continue to educate about the different formats, the type of capabilities that they provide, and how that impacts the community and the environment. I think it's very important to emphasize that we don't see ourselves as a visitor, you know, to these communities. We absolutely see ourselves as a neighbor. So our strategy has always been to engage very, very early, to explain, to listen, to listen to what the community needs, to understand, and then to work in a transparent way, you know, to meet the requirements of that community as we become one of their important neighbors.
With the amount of capital pouring into AI, a demand for data centers is unlikely to slow any time soon. The challenge is meeting that demand without alienating local communities.
It's mind-boggling the trillions of dollars that are going into that. I think people forget, you know, data centers aren't the end product. You know, that's not what the product is. The data centers are what enables you and I to have this conversation. It's every text you send and every post you like and every photo you take and everything you buy online. I read a poll, I think it was New York Times, said that 71 percent of American adults don't want data centers in their community. Yet 96 percent of American adults use the Internet every day. So, I mean, there's obviously a disconnect there, right? We've got to figure that part out.
Next, most of those data centers going up in the U.S. will support a handful of big U.S. tech firms. We take a look at the lone European horse in the AI race: Mistral.
This is a story about rooting for an underdog. The artificial intelligence race until now has been dominated by a handful of U.S. tech giants.
I believe that the future, if we get this right, can be much more like a new renaissance than a new industrial revolution.
Whoever wins AI wins. I think that's very important.
It's productive. It's profitable. It's generating enormous profits. This small trio of executives and the technology they created have become household names in much of the world. And they're committing unheard of amounts of capital. When ChatGPT hit consumers' phones in November 2022, investment in AI was $112 billion per year. By the end of 2025, that number increased more than 200 percent to $345 billion. And this year it's on track to blow past even those vast sums.
When we look back to 2025, where really this borrowing from hyperscalers had really only begun. Through the course of 2026, we're probably looking at around 300 billion in the investment grade market borrowings from hyperscalers themselves.
Next year, we're expecting another 15 percent more in issuance coming from some of the hyperscalers.
The U.S. hyperscalers are largely up against Chinese firms backed by the resources of their entire government as they pursue lower cost alternatives that require less compute. Lost in this titanic struggle thus far has been Europe, which hasn't seen the levels of private or government capital committed to frontier AI models. But France does have a dog in the race, an underdog to be sure, but one that seeks to redefine the competition in a way that it believes can put it at the front of the pack.
The way in which we should look at the world is rather an opposition between the closed model providers that are promoting a centralized version of artificial intelligence and trying to export it to the entire world, and an open version of artificial intelligence where you have multiple actors that are contributing to an open ecosystem and then helping enterprises use that and customize their systems on top of an open infrastructure.
Arthur Mensch is a co-founder of Mistral, the leading AI contender for Europe, and he believes it has a built-in advantage over the closed source, closed weight U.S. hyperscaler approach.
We actually are the leader in that domain. We are now the best open source model providers for sovereignty deployment. That gives us a special status that I think goes way beyond the fact that we're European.
So on the distinction between open source, open weight and closed, what are the advantages and what are the limitations of open source, open weight the way Mistral is?
So the advantages for the enterprises are threefold. The first is that you have business continuity guarantees because you're decoupling the IP layer, which is the model, from the infrastructure layer. So you can actually bring the model into the infrastructure you want. You can use multiple providers and you can make sure that whatever happens to the model providing company, you're still going to be up and running as a business. The second thing is really closely linked to the fact that, when you want to have return on investments on your AI deployment, usually those AI deployments are dealing with very sensitive data. And in many cases, this is not the data that you want to share with a centralized AI provider. And what that means is that a lot of customers of Mistral are actually coming to us because we come and deploy our models close to their data in a way where we don't even see that data.
The third aspect, which has grown in importance in the last few months, is the fact that AI is getting expensive. And so in order to build the business cases of AI deployment, you do need to take into account the cost. And the cost, if you pay the IP layer, if you pay for the models provided by centralized AI providers, you're actually going to pay much more than if you're paying for open source models being deployed on your infrastructure, on the infrastructure of an open source model provider. What that means is for an enterprise, you are asking me about the downside and the upside. There is actually only upside. There is no reason why not to use open source models today, except the fact that it's a little bit harder because it came a little later. But our job at Mistral has been to make it much simpler for our customers to deploy AI systems on top of open source model.
Mistral recently released
We have been under capital and compute constraints for the last two years, but we managed to produce a model extremely strong despite the fact that we were not controlling a lot of compute. This is actually about to change because we've made the right investments for doing so. So we're very excited by the upcoming roadmap because we're just able to scale the training. We're able to scale inference capabilities for our own customers. We do it in a way which is... which I think is a little more frugal. We've been able to produce—the kind of technology we've been able to produce, we produced with much less capital than some of our U.S. counterparts.
Mistral is taking a different approach to AI, one that requires frugality. But how much of that difference in approach is required because of the limitations inherent in being a European tech company?
Mistral is... I wouldn't say it's the exception. It's probably one of the first that we've seen to this scale and size in Europe.
Roxanne Varza is the director of Station F, a European startup incubator.
