Jun 11, 2026
On June 11, The Capitol Forum held a conference call with J. Wyatt Fore, partner at Shinder Cantor Lerner, to discuss his recent article, “Don’t Look Down: Antitrust at the Edge.” The full transcript, which has been modified slightly for accuracy, can be found below.
TEDDY DOWNEY: Hello, everyone. Welcome. I’m Teddy Downey, Executive Editor here at The Capitol Forum. Today, I am super excited to be joined by J. Wyatt Fore, Partner at Shinder Cantor Lerner. Wyatt recently authored the article, “Don’t Look Down, Antitrust at the Edge,” which examines competition issues emerging in the edge computing market and how established antitrust theories may apply to the next generation of AI infrastructure. Wyatt, thank you so much for joining us today.
J. WYATT FORE: Thank you for having me, Teddy. It’s a pleasure to be here. I’m a longtime listener and reader of The Capitol Forum. So, I’m pleased to actually make an appearance.
TEDDY DOWNEY: I appreciate that. And before we get started, if anyone has questions, please put them in the questions pane. Or you can put them in the chat. We’ll get to them later in the conversation.
And so, Wyatt, before we get into your blog posts and discussing the edge, I am consistently so impressed by the quality of lawyers and thinkers that you have at Shinder Cantor Lerner. I would love if you could share with the audience, where does that culture come from? How do you consistently have such, I think, sort of really thoughtful and rigorous work being done at the firm?
J. WYATT FORE: Wonderful. Well, thank you for the compliment. I mean, for those of you who are listening who don’t know, I’m a Partner at Shinder Cantor Lerner, which is an antitrust boutique, which has offices both in New York and D.C.
We launched in October 2024. So, we are coming up on our two-year anniversary, which is hard to believe. Because it feels like, in many respects, a lifetime ago that we launched. And really, our mission is just to provide excellent antitrust advice.
I do think that we are a boutique. We’re fairly small. We are small but mighty. I do think that we punch way above our weight. And part of the reason why I think that we are able to tackle really difficult questions that our clients bring to us is because I think that we select not only for people who are antitrust nerds—and I don’t need to tell you, Teddy, that there’s lots of antitrust nerds out there. But I do think that people who end up at my firm tend to be people who are very curious and interested about the way the world works. I think that a lot of folks who work at our firm had previous lives doing something that was different from the law, or at least different from specifically antitrust, where they have dug deep into the facts of an industry to explore how it works.
And so, I do think that we do a great job hiring not only for intelligence but also for people who just like the work and are really curious about it and want to move projects forward. And selfishly, it means that it’s a wonderful place to work. I really enjoy all of my colleagues, and I can trust them with complicated, difficult projects and problems, both on a legal side but also a factual side.
Whenever you’re involved in litigation or an investigation or counseling, really the question is what’s best for the client and how do we do it as efficiently as possible. And I think that because we’re a smaller boutique, and because we’ve done, I think, a great job assembling a team, it allows us to accomplish big projects on difficult questions at a scale that I think is market leader.
So, thank you for the compliment. I’m glad that we’re making waves outside of our four walls.
TEDDY DOWNEY: No, absolutely. I’m super impressed. It sounds like you have different backgrounds from the people that work there. They’re not just all antitrust law lifers. They’re coming from different backgrounds, journalism, history, that type of thing. Can you tell us, just really quickly, your background, how you got into antitrust?
J. WYATT FORE: Sure. Well, I’ve always been really interested in economic history. I mean, my friends make fun of me because on my bedside table I always have my fiction books and then my nonfiction books that are out there.
I have always really interested in especially the history of the Industrial Revolution. I mean, if you think about the world in 1640 and then you think about the world in 1840, there was a dramatic change, right? I mean, it’s sort of a Sleepy Hollow story. Where a guy goes to sleep and he wakes up and the world is totally different. And this wasn’t just a difference in terms of technology. This is difference in terms of how companies are organized, how firms are organized, how economic production happens, and then everything downstream that happens from that too. Because if there’s changes in economics, there’s obviously changes in politics and society, et cetera.
And so, if you’re looking at the late 1700s, early 1800s, there’s this sea change, not only in the rise of factories, but also the rise of global corporations, legal persons, the rise of global insurance, the rise of democracy movements. I mean, if you look at the history of the voting franchise in England, you have revolutionary moments in 1790 and 1830, and all of that flows downstream.
And in many respects, the world that we live in today looks much more like that world of the late 1800s than the world did in 2005. We have breakthrough technologies and artificial intelligence and machine learning and autonomous vehicles and network connectivity. And as a result, if you talk to, I think anyone today who was alive in 2005, they’ll say that society has deeply changed too. And we don’t need to go into all the obviously political changes that are happening. But I do think that the political and social changes that have happened in the last 20 years are a direct result of the economic and technological changes that have happened. And really, there’s no better example than the 1800s.
So, I’m sorry, you asked me a different question. And now you got me ranting. Yes, I think that I’m the kind of person I’ve always been interested in history. It’s kind of natural that I became an antitrust lawyer and wrapped this back.
TEDDY DOWNEY: Yeah, myself, I love to learn about the history of business and the history of law. And I read a great book, “Maxwell Perkins: Editor of Genius.” And it’s about Maxwell Perkins, the editor of famous authors, F. Scott Fitzgerald, Ernest Hemingway. And people are like, they read it, and they’re thinking, oh, I’m learning all about F. Scott Fitzgerald and Ernest Hemingway. And I’m reading it, I’m like, this is a fascinating insight into the history of the publishing industry, and Scribner’s. I couldn’t be—I think we’re like minded and being nerds in that way. But good to know that that’s how you get such a great culture over there. I’m going to try to emulate that. They said that I can over here.
Now, let’s talk about the blog posts, why everyone’s here. If you can start us off with just what is edge computing, it just seems super, super important to get that background, lay that groundwork first.
J. WYATT FORE: Sure. So, edge computing refers to, I think, an architectural shift. Back in the day, It’s so funny, because this is not only within recent memory, but it feels like just yesterday. But during the cloud computing architecture, the idea is that I’m over here. I create all this data. I send it all the way to a data center in Loudoun County. There the inference is happening. The intelligence is happening. And then the results are sent back over all the cables and all the networks or whatever back to me over here.
Edge computing inverts that by changing where compute to—edge computing changes that architecture, because the computation happens where the actual data is generated. And so, that’s on devices and regional facilities at base stations, hospitals and universities, inside your car, inside your phone, right? So, the idea is that the computation that’s happening isn’t happening over at a data center in Loudoun County. It’s actually happening where the data is actually generated and where the inferences are needed. Like the actual intelligence that’s needed from the AI is happening.
So, it’s a shift of the architecture from something that’s happening way, way far away to the sort of edge of the network, the nodes of the network. And that’s where the actual data is generated and needed.
TEDDY DOWNEY: Your article focuses a lot on NVIDIA and Google. Maybe we could talk about NVIDIA first. How is it positioning itself in the edge computing market? And why does that raise antitrust questions?
J. WYATT FORE: Sure. So, all of this comes with a big caveat that the market is very quickly evolving. I was joking with Teddy, with you, earlier that we’re going to listen back to this podcast episode in a year and sort of laugh at how out of date it is.
So, all of this comes with the caveat that everything is just changing dramatically. It feels like every month there’s another major development. NVIDIA is an interesting company because people, I think, in the ordinary world think of it as a chip company—which it is. It makes these sort of frontier edge chips that allow the actual sort of intensive computation to happen that powers a lot of AI. And I think that that’s how people often think of it. But NVIDIA isn’t only a chip company. It also provides other ancillary services that are associated with chips.
And so, I think the most important thing for the listeners to know about is the CUDA platform. It’s a programming platform that allows you to essentially write code that works on the chips. And why that’s important is because in many respects, CUDA is a large competitive advantage that NVIDIA has because CUDA, which was started, I think, in 2006, 2010, somewhere around there, somewhere in the late 2000s, it’s been around for about 15 years.
And so, anyone who is designing software on these sort of frontier chips that NVIDIA sort of famously pioneered, drafted it on CUDA. And so, if you are creating software that operates on these chips, chances are you’ve written it on the CUDA platform. And so, that means that it’s really difficult to sort of leave the NVIDIA chip platform if you have sort of entered into this world.
I believe in 2023, 2024, whenever there was the big sort of chip supply chain crisis in the wake of COVID, and a lot of people were shifting to other manufacturers or other designers of GPUs that NVIDIA was providing. And they actually had a lot of challenges. Because it turns out like just because you buy the chips doesn’t mean that you can make them work. Because you don’t have the software to operate on those chips which makes it more challenging.
NVIDIA also has a couple other ancillary devices or ancillary products that I think are important for this discussion. One of the most important I think is Mellanox ConnectX, which is networking. Mellanox previously was a separate company. It was acquired by NVIDIA, I think in 2020, somewhere around there. And basically, that’s like the wires and cables that go within the chips. Because if you’re talking about speed and latency, it’s not only important how quickly the information is computed on the chip. What’s also important is how quickly the information is traveling between the chips. And between the sort of system and the user, right? So, Mellanox is part of that story. It’s like you can think of it as just like very high performance ethernet cords, for lack of a better word. So, that’s important.
And then the last thing I think is Jetson Edge Hardware. And those are hardware that is primarily active in the edge situation. And that is the same thing. It’s not only the chips. It’s the entire system, right? It’s the chips. It’s the CUDAbased stack. It’s everything together.
And so, when you’re starting to see NVIDIA’s business model. The chips are important. But, for example, I believe in January, 2026, NVIDIA came out and they said, we’re going to have this new systems offering to customers called the Vera Rubin system. And Vera Rubin, if you think about it in your mind—and I’ve never seen one of these in person. So, this is what I have in my mind. It’s not just the chips. It’s the entire system, the GPUs. It’s the CPUs. It’s like the wiring that goes in there. It’s the system hardware. It’s like the sort of plug and play.
So, it’s not only the frontier chips, but everything. And so, I say that because that’s really important from an antitrust perspective. Because antitrust historically, especially in emerging technologies, is really interested about bundling and tying. If you think about the sort of original IBM computer that was built in the 60s, it was everything. It was your mainframe computer. It was your software. It was your operating system. There was only one product you bought, and that was IBM. And that originally created antitrust problems. Because you can see how, well, what if someone wants an IBM mainframe, but an operating system that’s different on it? Or someone wants a different software on.
And so, when technological development happens, what happens is you usually get like one giant system. And then the market sort of chips away at it. And then it sort of breaks it apart into individual ones. And so, you can see something similar happening here that’s happened in previous technological revolutions, where NVIDIA is offering this product. If you think about the Vera Rubin system, this product that has GPUs and CPUs and software and connected tissue and all that kind of stuff, you can start to see how, from an antitrust perspective, that starts to look like bundling or tying, even if it’s just at the very forefront. I’m not accusing anyone of an antitrust violation, but you can sort of see how antitrust—which talks about relevant markets, discrete markets—that this sort of starts to raise sort of antitrust questions.
TEDDY DOWNEY: Yeah, this is not like an atypical thing in tech even, recent tech. I mean, Amazon web services, it’s hard to pull out because they have all these applications on it that otherwise are more expensive to do on your own. There’s all these lock-in effects.
You’re saying NVIDIA does this through (1) through an acquisition. Sounds like you think probably that should have been blocked, potentially, or at least should have gotten more scrutiny than it did, potentially. And just vertically integrating to lock the customer in. And you’re seeing this already, but are you seeing this already not on the edge? Or are you particularly worried about it on the edge?
J. WYATT FORE: I think I’m particularly worried about the edge right now. I shouldn’t say I’m particularly worried about the edge, but I am particularly interested in the edge. Because in recent history, in relatively recent history, we have seen market entry into sort of what I’ll call frontier chips market, even though that’s like query whether that’s a relevant market. Like Google has their TPUs. We have Tranium. We have NVIDIA’s GPUs. Like query to what extent they are interchangeable. Because they are different products that are kind of gesturing at the same thing. But we are seeing interesting innovations there because everyone always accused NVIDIA of being a monopolist in sort of these frontier chips.
And then the next step is you see sort of other large tech companies moving into this. But that’s more at the cloud level and more at the model training level. What is interesting about the edge is that NVIDIA is still very dominant at the edge level.
I mean, a lot of these sort of market entries that you’re seeing from Google and Tranium, et cetera, are happening sort of at the cloud level, at the model training level. If you think about what a big giant data center—and group of data centers—is really good at, it’s training your models. So, that like models can be trained. And so, therefore, they can go and make inferences. If you think about AI, there’s like training on the one hand for compute, and then there’s inferences on the other. Edge is really an inference story, especially right now. And NVIDIA is still, I think, in the inference at the edge level, still the dominant firm, even if you are seeing some market entry at the sort of upstream cloud training level.
So, this is all to say that I’m interested in NVIDIA at the edge, but I think that that’s just because that’s where the market is moving. In 2023, I believe that two thirds of compute was used for training, and then one third was used for inference. And then there was a recent McKinsey study really in the last six months that said 2026 is going to be the first year that inference is actually more than 50 percent. And so, the world is going towards edge. NVIDIA currently has an edge in edge, for lack of a better word, there.
But the market is developing, right? Who knows what it’s going to look like in three years? Even if we’ve seen market entry upstream, there might be market entry downstream. Who knows what practices are happening there?
TEDDY DOWNEY: What do you think about one way that NVIDIA seems to be able to stay ahead of their peers, same with Apple, is access to Taiwan semiconductor volume, those kinds of contracts as another way of locking out, you know, sort of keeping competitors at bay?
J. WYATT FORE: Yeah, I mean, it’s no secret that the manufacturing capability, which is largely based in overseas countries, including Taiwan, as well as South Korea. But Taiwan is most famous. There is a shortage. And as a result, they can command high prices from their suppliers to take care of that shortage, right? That’s supply and demand 101. If you are one of the largest companies in the history of human civilization, you can pay that toll and other people can’t.
Now, is that an antitrust issue? I don’t know. I would need to think more carefully about it. But certainly, it sort of incentivizes investment in more fabrication, theoretically, right? Because if you’re able to obtain really high prices, you want to build as many factories as possible to continue to get your high prices. And so, maybe this is the market working itself out. But obviously, it’s a problem from a market resilience standpoint, as well that we’re reliant on only a handful of places to obtain these highly valuable things.
So, is it an antitrust issue? I don’t know. It could be, especially exclusive contracts that you’re talking about. It could be. But I’m not so sure there’s really an antitrust remedy there. It could just be the market working itself out and dealing with the shortage. I mean, right now, there’s a lot of talk about us living in an AI bubble. Maybe that’s true, maybe it isn’t.
But really, what it reflects is that there’s an unquenchable demand for this right at this moment. And that’s creating some wild market effects. Is that just the competitive market working? Are there antitrust sort of schemes hiding in plain sight? That’s not something that I can have a definitive opinion on or do have a definitive opinion on right now.
TEDDY DOWNEY: It’s just hard to know without seeing the contrast, because it just seems like they have, it’s not really the price that is prohibitive. It’s the volume allocation. It’s like TSMC is picking the winners and losers by who gets access to the most advanced types of chips. They’re only making so many. And who are they contracting with? They’re contracting with NVIDIA and Broadcom and whoever else. And like, oh, you want some of those chips? You’re Google, you’re whoever. You don’t have enough. Oh, you’re going to contract with—I mean, you’re just locked into this monopoly ecosystem, it seems like.
So, I agree. If anyone out there has one of these contracts lying around, please find me. My information is on the website. You can signal me.
J. WYATT FORE: I was about to say, Teddy, a lot of investigative journalists would really love that, right?
TEDDY DOWNEY: Yeah, exactly. I would love to get my hands on that contract. Let’s talk move onto Google. We’re already starting to get questions. We’ll move to the list of questions in a little bit. Let’s talk about Google. You also see some issues potentially with Google on the edge. What did you find there?
J. WYATT FORE: So, I think that Google, one of the really interesting things here is that on the one hand, you can tell a story about all this, that’s a competitive market. And on the other hand, it seems like every firm is sort of making the same play and that’s vertical integration, right? And it’s vertical integration in order to make it very difficult for customers to switch out.
And I think that Google has a very similar story, right? Like if you are playing with Gemini, Gemini is trained on TPU chips. Which is Google’s proprietary chips, et cetera. And I think it’s a similar story. It’s that in many respects, they feel that what the market is disciplining them to do is to vertically integrate and sort of compete in every sector of the staff.
So, specifically to edge, as I mentioned, Nvidia really does appear to continue to be the dominant player there. But just recently Google announced a recent edge initiative to offer products to the market. But I think most importantly is the news with Apple. That Siri is integrating with Gemini. And obviously, for anyone who followed closely the search litigation, which I think is a lot of your listeners, this should sound like a very similar story.
Google and Apple are coming to an agreement, which is technically non-exclusive. But it allows this access to this immense volume of data that allows the model to continuously improve itself. And obviously, search is different from AI because Gemini is not the dominant provider of inference models. In fact, in many respects, they’re not even the largest.
But access to these sources of data and this sort of stickiness of the user. Because if Siri—which a lot of people have an iPhone in their pocket—if Siri is how you’re primarily interacting with AI models, well, it’s really difficult to switch away from that. And also, Google is soaking up all this data, which allows it to continuously improve its models. And so, you can see a similar story there as search. Just the difference—obviously a very important one—is that Google does not appear at this moment to have monopoly power in any relevant market related to that.
And so, I only throw that out there because it does seem like the next sort of round. And I think that this is true for all of edge computing that a critical battle line is going to be access to this data, to use it to make inferences and to make the models better.
And whether that’s in your phone on Siri or whether that’s in a factory floor, think about how much data is created in the factory floor. Like the temperature of the machines, the vibrations, the robots going around, the pH of the air, whatever. Think about how much data is in the hospital. Sicknesses microbes, all that kind of stuff. And the ability to deal with that data at the edge, I think, that’s going to be where the market is moving. So, I just think it’s really interesting.
TEDDY DOWNEY: Yeah. And that agreement between Apple and Google in search was deemed illegal, right? The judge said that was an illegal agreement.
J. WYATT FORE: Right.
TEDDY DOWNEY: Now, the judge shockingly did absolutely nothing about it. Basically said, you can keep doing it. I mean, with very minor changes, which intellectually just blew my mind. To this day, I don’t understand how he did the mental gymnastics to end up where he did, but I guess some people are like that. They didn’t work at a place like yours ever, clearly. So, it’s kind of interesting to me that you would do the exact same playbook, even though a judge just said it was illegal. If there’s no consequences, I might as well, right?
J. WYATT FORE: Right. So, it’s the problem of antitrust remedies, right? It’s like by the time a monopoly is entrenched, by the time a monopoly is so entrenched that there are any competitive effects that are so easily discernible that you can win a court case. And then to get to the remedy phase, it takes seven years, ten years.
And so, this is one of the problems that I think that enforcers right now are actively dealing with this. I think that a lot of enforcers feel a bit burned by the Google search litigation where they won on the law and have lost on the remedies. I think that there is a lot of nervousness there because on the one hand, they don’t want to kill the goose that laid the golden egg. We are very lucky to live in a country like the United States, where people have the freedom to make innovative products. And we want to continue that innovation.
They also don’t want to wait too late and let one particular firm get so dominant that even if you were to find a monopolization violation that effectively they would be on, they would be beyond certain remedies. And so, I do think that there is a remedy story there that in AI that people are learning from Search that, okay, well, there is a concern about moving too early. But there’s also a concern about moving too late.
TEDDY DOWNEY: Yeah. Well, we’ll see what happens with Google ad tech. It could have a different conclusion there. Stay on Google for a second in autonomous vehicles. I mean, you could argue that’s one area where Google does have a bit of a monopoly position with Waymo. They’re currently getting a really crazy amount of data from owning Waymo. Do you think they could weaponize that the way they did with search, weaponize the data they get from Waymo?
J. WYATT FORE: Yeah. I mean, I do think that autonomous vehicles—I mean, look. I’m old enough to remember when we were talking about autonomous vehicles even in the eighties. I mean, if you look at “Back to the Future II,” the joke is that in 2005, really far in the future, everyone’s going to have autonomous vehicles. We still don’t have them. But I do think that is one area where there is just an immense amount of data Like whether there’s a wreck, how fast is the car going? Just a million micro decisions that any driver of a car makes. And so, I do think that autonomous vehicles is a prime candidate for one of the applications of edge AI.
I’m not totally sure about the market structure at the moment, just because it is one of those areas where there are certain front runners and certain other ones. But the vehicle market itself is going through such dramatic change right now too, right? With the rise of electric vehicles from China with the sort of rebalancing of automotive supply chains, et cetera.
But I do think that it’s going to be a key, I think test use case, of edge AI as to whether or not vehicles themselves can drive themselves based on that data. And also, if you are the edge provider in all of the cars and you’re the only ones getting the data, that does raise pretty serious antitrust issues. Whether or not one could have sole access over all that data, I think is a logistical and structural question that enforcers I’m sure would look at. But I do think that it is an area where you are going to see increasingly edge AI applications, and also that data is going to be incredibly valuable.
TEDDY DOWNEY: NVIDIA recently introduced an open source autonomous vehicle software called Alpamayo, and they’re hoping that other car companies get locked into that ecosystem. So, they have to use NVIDIA hardware, just like NVIDIA is doing with CUDA. Obviously, they don’t have the edge that Google has. But are you concerned that you could see a situation that you saw with sort of Apple and Google and cars where it’s just like they get this kind of duopoly and you have all sorts of associated competition problems?
J. WYATT FORE: Right, yeah. No, I mean, that does raise, I think, serious competition problems if there were to arise a duopoly. And the fact that cars—it’s not only important for your car to be able to drive you, it’s also really important for it to not hit other cars. Which means that they have to talk to each other.
And so, on the one hand, that means that the market might naturally only produce one or two products, right? Because in that respect, it’s kind of a networked industry, and networked industries tend towards monopoly or duopoly for that reason. Because it’s more efficient for the market to only produce one or two firms rather than a panoply. But also, you can imagine that that could be a huge competition problem, right? Like, if you only have one or two firms dominating a firm, and also they have to talk to each other, well that starts raising Section 1 and Section 2 issues.
Obviously, this is a bit hypothetical at the moment. But you can start to see why this data would be so valuable to the market leader. Like why it would incentivize a market leader or a firm to move into this market and to dominate it, because it’s very valuable. And also, there are strong incentives to, once you have the monopoly, to keep it.
TEDDY DOWNEY: You write a little bit about telecom and 5G infrastructure in the blog. What’s your concern there?
J. WYATT FORE: Oh, man. Well, telecom has a long history with the antitrust laws, and for many reasons, right? I mean, the AT&T breakup is still the example that everyone reaches to in their mind of an antitrust breakup. And I do think that the AT&T story is not told enough.
Like, everyone knows that the Bell system was broken up. But with the benefit of hindsight, we can see why the AT&T breakup was so important. It’s because AT&T, which controlled the copper wire network, stopped ancillary devices like answering machines from plugging into the AT&T system or other phones, unless they were an AT&T product.
Well, one of the most important products that happened from the 60s through the 90s was the modem. So, the breakup of AT&T and the squashing of that monopoly power, all of a sudden, allowed this flourishing of modems and this innovation level that happened on top of the telecom to happen. And the reason why I say that is because you can see how a monopoly position on one hand, in one market, can cause really serious anticompetitive effects, especially a threat to innovation at markets built on top of it.
And so, the reason I talk about telecom here is because when everyone thinks about AI, they think about Nvidia and Google and these really sexy, innovative Silicon Valley companies. But a lot of this infrastructure, not all of it, in fact, a lot of it’s being built outside of it. But a lot of this infrastructure exists on top of the telecom network, AT&T, Verizon, T-Mobile, et cetera. And so, the reason I bring that up is because we had this entire debate during the Obama administration about net neutrality. Should these sort of internet service providers, like the rails of the internet, be considered common carriers so that they had to treat everyone reasonably and fairly equally?
And you can imagine a situation in which an AI provider is able—who is, again, one of the largest companies in the history of human civilization, has the ability to pay a toll on a telecom network that a smaller competitor might not be able to. And so, if the telecom network has the ability to sort of pick winners and losers, based on who will share in the monopoly rents with it, that all of a sudden, starts to look like a pretty serious threat to competition.
So, again, I want to emphasize that this is just theoretical. I’m sort of looking at my looking glass. I don’t have any accusations to make or anything, but it’s something that as an antitrust practitioner—and as someone who cares a lot about innovation and creating an innovative culture and making sure that smaller companies that are created in a garage can compete with the biggest guys on the block—it seems like the telecom network is one area where this might be a problem.
TEDDY DOWNEY: There’s one more industry and then we’ll get to a list of questions. We’ve got a couple. Let’s talk about healthcare for a second. Patient privacy regulations magnify the value of local processing and storage over third-party clouds, making edge infrastructure especially important. In general, how do you think data privacy interacts with edge computing, antitrust concerns? And how might we see that play out in healthcare? In particular, I know we were talking earlier about surgical robots. So, there’s a lot of questions about healthcare that seems super important, particularly around data.
J. WYATT FORE: Oh, absolutely. I mean, healthcare, I think, is another key application for edge AI. Like, think about how much data a hospital produces, about the patients, about the temperature, about diagnoses, et cetera. A surgical robot, if it’s inside you and that a surgeon is using, it’s really important that there’s not a delay of the data and inferences going back and forth. You need something to operate quickly and correctly every single time.
I do think that your question about patient privacy is incredibly important, and it’s one that in the antitrust community we have a lot of conversations about. In my mind—and I know that this is a bit controversial, and a lot of intelligent people have different opinions about it, but my view is that data privacy and antitrust are, in some respects, at crossroads or at cross-purposes, because the touchstone of antitrust really is growth. It’s trying to incentivize economic growth by pushing a competitive market. And in many respects, antitrust likes copying and sharing data. Because the more people that have access to data, the more people that can use this input to grow their firm, et cetera.
But data privacy is, in many respects, like a break on that. Because data privacy laws, including important ones that I support, like healthcare—like I don’t want everyone knowing my healthcare data either. But in many respects, it allows the—it’s kind of like a patent or other intellectual property where you’re creating a moat around certain information that only certain people, including certain firms, have access to.
And so, although antitrust law generally likes copying, we like when an innovative product is multiplied across the market, data privacy does not. And so, I do think that it is one of those things where antitrust and data privacy are kind of operating at cross-purposes, and there is a risk that too much data privacy regulation will just entrench the market power of dominant firms. That’s not to—I’m not weighing in and offering a solution to this or taking a stake in this debate. But I do think that it is an opportunity that firms who have unique access to healthcare data might have a leg up on their competitors and might be shielded from that competition based on not sharing that data.
The last thing I’ll say on this is that one of the things I was saying earlier is that there is very much this race in AI to access proprietary data sources, in order to train their models, et cetera. And one of the interesting ones that I found recently is that Microsoft actually signed an agreement with, I believe, Harvard Medical School, someone affiliated with Harvard Medical School, in order to access research, medical research, in order to presumably train its models. And so, Microsoft identifies this as an interesting sort of competitive edge to access to medical knowledge that might be protected from other people, either because of patient privacy or because of intellectual property.
So, I do think that healthcare data is going to be immensely valuable in the edge AI space. And I think it’s one of those that I suspect that at major tech companies on the West Coast that they’re thinking very hard about.
TEDDY DOWNEY: And then, have you looked into the surgical robots at all? Have you looked into how that market is evolving? Or is this kind of an area to watch all these similar issues? Because you’re obviously going to have hardware and edge computing there and sensitive data potentially.
J. WYATT FORE: I think it’s an area to watch. I mean, obviously, there’s the right to repair issues with the surgical robot. That’s not really an AI issue. But you could imagine—I think of this as being the sort of edge AI problem, the vertical integration problem, right to repair, the Kodak problem. Like, they’re all sort of articulations of the same concern. And that’s a firm with significant market power in one market using it to squash competition in another.
And so, surgical robots, I just kind of use as an example because they’re fun to imagine. I’m not so sure if there are issues there right now, but I do think that is an example of something that would be very interesting. And I do think that healthcare edge AI is absolutely a use case for edge and something that we should all be looking at.
TEDDY DOWNEY: Let’s move on to listener questions. First one here, is edge computing a risk to the hyperscalers, i.e. how easy or more cost efficient is it for business slash consumers to rely on edge computing over the data centers for AI usage?
J. WYATT FORE: So, my view, and again, this market is quickly evolving, is that edge computing still needs the hyperscalers. Because the data centers are like the brain of the system and edge computing is like the fingertips or like the nervous system. Like an example I use is that if you’re crossing the street and a car is about to hit you jump back. That decision that your body’s making isn’t at the most advanced parts of your brain. It’s actually just goes to your spinal cord and then you have a reaction to it to draw you back. That’s kind of more similar to edge computing.
Edge computing, obviously, a lot of inference happens there and a lot of computation happens there, but it still needs the brain in order to do a lot of important things. Like a lot of the information is actually just cached back and forth. So, even though a lot of, for example, like the inference computing is happening at the edge, it’s still, in many respects, relying on the hyperscalers. So, I’m not so sure if it’s a threat to their business model as such, so much as the really smart folks who sort of already have done very well in the hyperscaler market or the cloud computing market are already thinking downstream.
So, that’s one of the reasons why I think that NVIDIA is still sort of the dominant player. Like NVIDIA has been very sort of forward-looking, knowing and seeing where this market is going. And so does Google. And so do other large tech companies. So, I’m not so sure it’s a threat to their business model so much as an evolution of it.
TEDDY DOWNEY: Yeah, it’s like a new market that people are getting into, but at some level maybe could come at a cost though. You might see less volume of tokens, or what have you, in the cloud as opposed to in the edge maybe. Maybe that’s another way to think about it.
J. WYATT FORE: Yes.
TEDDY DOWNEY: We talked about NVIDIA. We talked about Google. We talked about Apple a little bit. Any other companies that you think we should talk about in terms of entering this market or having an edge in the edge market, as you said earlier?
J. WYATT FORE: Well, I do think that any of the large cloud computing providers are interesting here. And that’s one of the things that I’m looking at, or thinking about, is I believe the European Commission has an industry study into cloud computing that’s due in 2027. And it’s going to be really interesting to me is how they deal with this question about edge.
And going to your previous question, we are seeing sort of a flurry of smaller companies operating at the edge. There is this flurry of activity where people are being innovative and doing some really interesting things. And I think from an antitrust perspective, that’s one of the things that they should watch out for. Because, again, if firms with a really strong market position in cloud upstream or any sort of input upstream are seeing sort of a lot of money on the table, they might be crowding out those smaller companies.
So, I’m not so sure if I would call any particular out. But I do think that if you are a smaller company—if you’re an enforcer and you’re thinking about watching this market develop, one of the things to watch is whether or not edge computing continues to have a flurry of smaller companies doing interesting, weird, innovative things. Or if you see it go the way of cloud computing, which is essentially four companies. And so, I think that that’s something that enforcers are actively looking at.
We’ve seen a lot of enforcer interest in this entire AI stack, both in the merger investigations and in conduct investigations in the United States, in China, and in Europe, and in the UK. You’re seeing a lot of enforcer interest, but not really any sort of landmark cases. And the reason why is I think that enforcers are very much in watch and see mode, as well as learning mode. They’re using every transaction as an opportunity to glean more information about the market, to see how the firms are operating, to see what the incentives are for different companies and how they make money, so that they have a really smart understanding of how the industry works.
And I think that if I were an enforcer, that might be something that I would be looking at is this edge computing market, both how it exists now, and also to compare it in sort of three to four years and to see, oh, is this just another firm or another market that’s being dominated by a handful of firms? Or are we seeing vigorous competition and the rise of companies that are doing interesting, weird, one-off projects, that are trying to make a go at it?
TEDDY DOWNEY: You mentioned antitrust enforcers not bringing cases. We’ve seen a lot more enforcement and investigations into vertical conduct. Do you think there’s any hesitancy to bring vertical foreclosure cases in edge computing because you’re still running into some court resistance on vertical theories of harm? Tempur-Sealy mattress firm being, I think, a recent example of that.
J. WYATT FORE: I mean, it’s no secret that vertical cases are hard, although not unwinnable. And I will just toot our firm’s own horn. And that’s, we won—or we got a wonderful settlement—in the healthcare space in a vertical case called Sidibe versus Sutter Health. That was about hospital contracting practices.
So, I do think that you’re right, that enforcers are increasingly—they’re much more willing to investigate and prosecute vertical allegations of harm than they were probably 20 years ago. I do think that you’re right that there is some court hesitancy in vertical cases. They are harder to bring.
But my view is that the hard cases are why we were put on this earth, right? If the anti-competitive effects are there, if you see them, then they’re interesting and that shouldn’t deter you. That being said, I think that you’re right, that some courts, it’s harder to explain to them why vertical theories of harm are more challenging, especially in a market where the firm does not have monopoly power. Or if they only simply have market power. Or if they do have monopoly power, but it’s like sort of a closer call.
Also, I think that obviously I’m not an enforcer. Nor have I ever been one. Or a public enforcer. I consider a lot of my work private enforcement. But I do think that they are more in a wait-and-see and looking mode. Nobody wants to bring a case that they feel is weak. Nobody wants to bring a case that they think is weak, that is going to run into resistance. Add the geopolitical dimension on it, right? Add the very quickly moving market on top of that. And I think that that means that folks are more inclined to wait and see rather than take really strong bullish action.
TEDDY DOWNEY: I don’t envy them. I mean, you mentioned before, are we in a bubble? I mean, that seems pretty good evidence to me that there is a bubble here. And it’s just hard because you don’t have any sensible business models, right? Like you don’t know how much revenue is really going to come on the other end of this. It’s like a bit of a hope and a prayer. How do you know what to do in a market where it doesn’t really make sense in a lot of ways? So, I don’t envy them in that respect.
We have one question here. When it comes to defense and intelligence, could edge computing monopolists squeeze defense contractors or even the government itself, as security and military applications become dependent on the technology?
J. WYATT FORE: Well, I do think that warfare is an application of edge computing. I’m remembering back to—do you remember “Star Wars” episode two with “Revenge of the Clones” or whatever, where the autonomous robots are all fighting each other?
TEDDY DOWNEY: Yes, vaguely.
J. WYATT FORE: That’s the image I have—which I know is not fair, but that’s the image I have in my mind.
TEDDY DOWNEY: Yeah. Yeah, they definitely, well, it’s like the humans are fighting the robots or maybe the robots are fighting the robots. I don’t really remember.
J. WYATT FORE: That’s the image I have in my head. But I do think that warfare is probably an application of edge AI, that if public reporting is true, that the Pentagon and policymakers are actively looking at it. In fact, we’re seeing from the Ukraine-Russia conflict that autonomous drones are playing a very large role.
I do think that there has been public reporting about dramatic consolidation in the defense sector. It’s a natural problem because the government is only one buyer. So, there is a monopsony problem in that market on the one hand. But on the other hand, the industry is also consolidated. And so, the government often feels like they are overpaying and not getting super great products. Because they can’t switch away from one prime contractor to another, because there are so few of them.
So, this is all to say that I don’t know. I mean, I think that the market is evolving so quickly and that it’s hard to say. It is certainly a possibility. And also, not only do we not know what the market’s going to look like, we don’t even know what products are going to be. So, it is certainly something I think to be on the lookout for.
TEDDY DOWNEY: Okay. Well, we have one last question. I can’t read it anymore. But the gist of it was, why does Teddy see a monopoly problem in all the markets that we talk about on this call? And I would just inform this regular listener, the podcast is called Second Request, the podcast that explores solutions to monopoly problems. So, the whole premise of the conference calls and the podcast is to look at monopoly issues.
But Wyatt, I’m curious if you think that when you go out in the world, do you see a lot of monopoly problems? Because I know I certainly do. That’s why I got into the investigative journalism business. It just seemed like a huge percentage of the economy seemed to be related to issues around market power, domination, and predation.
And now you could say, well, some of that’s just fraud. Some of that’s just gambling or bad law or what have you, just like generally the effects of laissez-faire policy towards business over the past four years. But there’s obviously also lots of mergers, huge consolidation, huge levels of profit throughout the economy.
I know I see it when I go to the store, pretty much in every aspect of my life, I see competition issues that I’m curious about, curious about the history of the law, curious about the history of business. And I’m wondering if that’s how you go through life as well.
J. WYATT FORE: Regretfully, I think that I do as well. Or at least my friends make fun of me. I used to always say that literally from birth, if you think about hospital consolidation until death, casket price fixing, your entire life is dominated by industries with high levels of concentration. So much empirical work has been shown that nearly every industry is consolidated. It used to be that there were, I mean, it’s even like the examples roll off, right? I remember when there were more than three airlines.
TEDDY DOWNEY: I remember when flying on an airline was a genuinely glorious experience. I mean, you had plenty of room. I was on a flight the other day and I said, I’m going down to the Knicks game. I’m going down to San Antonio with my father. And I look at the woman on Southwest—usually the stewardesses on Southwest are so nice, flight attendant. And I’m like, look, my father is sitting over there. Can I please just get up and sit next to him? No one’s sitting next to him. And she goes, you should have paid for that. And I just, it was astonishing, just such an astonishing commitment to the bit of making you as miserable as possible on a flight, right?
J. WYATT FORE: So, that you pay a little bit more.
TEDDY DOWNEY: I actually couldn’t believe it. I’m like, I bought this ticket day of to go to San Antonio. It was ludicrously expensive. I mean, it’s not like, right. And this woman just had the audacity. You should have paid for that.
And yeah, I mean, it was not like that. I mean, I remember flying as a child. I would fly by myself to New York and they would take care of me. They put a pin on me that a designated person keep an eye on me. Really dramatically different experience to your point.
We actually had a guest on last week, two weeks ago, who said he was in-house counsel at an airline. And he said that if he talked the way that the executives talk on their earnings call—where they say we expect 15 percent price increases later this year—he would have gotten a call from DOJ and worried that his boss would get in real trouble, like go to jail. Like he was legitimately like, Oh, God. I could never let something like that happen. And he heard it on an earnings call now, you know?
So, I think it’s very different. I think it’s a very different sort of political economy that we operate in. Let me ask you this, last thing, any other tech issues that you’re looking at around—we talked about the edge a lot. Anything else that jumps out at you as particularly interesting right now in competition policy, generally?
J. WYATT FORE: In competition policy, generally, I think that healthcare is—I mean, tech is really interesting and sexy and everyone loves talking about tech. If people, I mean, I think that the antitrust issues in healthcare are awful. I mean, it actually might be the worst offender of any industry. Both, I mean, all the hospital consolidations as well as all these sort of anti-competitive contracting practices. The DOJ, the Trump DOJ, has been active on this issue as well as a lot of state attorneys general.
But for some reason, the healthcare stuff just really bothers me personally, because you’re talking about not just a widget, you’re talking about a human being. And it is no secret to any of your listeners that healthcare costs in this country are atrociously expensive. I mean, just heads and shoulders above. I mean, a friend of a friend just moved from Los Angeles to Lisbon, Portugal, and bought private insurance there because he’s American. And the difference between the health insurance costs in California versus Portugal, like both private insurance, both good private insurance, the difference there was enough to send his kid to private school. It’s just out of control.
And there’s a lot of reasons for that. But a lot of those reasons are just we’ve let several companies operate critical choke points in the healthcare supply chain, and they charge monopoly rents, and it increases year over year.
So, I do think that healthcare is something that I’m always passionate about and caring about and paying attention to. This DOJ has been pretty active on it. And they’ve been looking at it as well as a lot of state attorneys general. And the FTC obviously has an outstanding healthcare shop. And they’ve been very focused on this as well.
So, I do think that healthcare is one of those things that does not get enough play from the general antitrust community. Because I do think that it’s the underlying reason why we pay so much in this country for so little.
TEDDY DOWNEY: Yeah, I couldn’t agree more. And Wyatt, this was a really great conversation. I love learning more about your firm, your analysis of the edge. I can’t thank you enough for doing this today.
J. WYATT FORE: Well, thank you for having me on. This was a pleasure.
TEDDY DOWNEY: And to everyone for joining the call. Thank you so much for joining us today and every week for our conference call series. And you can check out this episode. You already heard it. You don’t need to listen again, but this episode and others on Second Request, wherever you get your podcasts. And also, our TCF Investigates podcast is on the same channel. So, I highly encourage you to check that out as well.
Thanks again. This concludes the call. Bye everyone.
