Fewer Than 3% of Deals Get Investigated. Selling to the Monopolist Is What Triggers the FTC.

When Lina Khan was appointed chair of the Federal Trade Commission at 32, she was the youngest person to hold that job in the agency's history and one of the youngest leaders of any federal agency. Hosts Sarah Guo and Elad Gil introduce her as the official who has challenged Nvidia, Meta, and Microsoft, pushed into healthcare and pharma, and made Silicon Valley argue about a "sheriff." Khan's own origin story is quieter. She started as a business reporter whose assignment was to document how U.S. markets had evolved.

The beat took her into airlines, book publishing, chicken farming, and aluminum, the nooks where four decades of mergers had concentrated whole industries. She watched consumers drive much farther to reach the nearest hospital, chicken farmers who could sell to only one company, and small businesses facing a handful of gatekeepers who controlled access to markets and could dictate worse terms. Those reporting trips convinced her that how we structure markets is not a fact of nature. It is a policy question: monopolies in every market, or more competition.

Historically, she says, America chose competition, and she believes that choice is what produced both phenomenal growth and so many of the companies the tech industry celebrates. The through-line of the episode is that the same choice is now on the table for AI. Guo and Gil want a usable map: when will a deal get investigated, how should founders think about selling, and what does Khan want the model layer to look like. She spends the next 25 minutes making the map more boring, and more specific, than the sheriff headline.

Before Yale Law School, Khan spent about six months talking to businesses that sold through Amazon and to investors and analysts who were trying to guess the company's endgame. Amazon's returns were still fairly low. The people inside the market, she says, were already living with a form of structural dominance that had not yet shown up as higher consumer prices. That fieldwork became the student essay *Amazon's Antitrust Paradox*. She puts a caveat on it immediately: writing as a law student is very different from being a law enforcer.

The paper's claim was historical as much as it was about one company. U.S. antitrust, in her telling, had moved from a broad view of the many ways a firm can exercise monopoly power to a narrow metric of short-term price and output effects. If consumers were not paying more right away, squeezing rivals did not count. She wanted the older prism back: dominance can kill potential competitors, lock channels, and squeeze suppliers without a visible price spike. That is still the lens she brings to young digital markets, including ones that do not yet have a price war to measure.

Gil presses on a case that still bothers founders: the FTC's challenge to Meta's purchase of a 30-person VR fitness app in a market that barely existed. Khan's answer starts with the statute. Congress told the agencies to review mergers and ask whether they *may substantially lessen competition*. That language makes the work predictive by design. Enforcers have to name the real dimension of rivalry, price, quality, or innovation, and then ask whether the deal is being used to snuff out a competitor in an existing market or a future one.

Nascent markets are the hard case. It is not obvious whether a category will fizzle in a few years or become the next great thing, and she does not pretend otherwise. The context for the last few years of FTC and DOJ work, she says, is the prior two decades of default hands-off. Enforcers told themselves that digital markets moved so fast, with such low startup costs, that a few people in a garage would discipline any monopoly that appeared. What actually happened, in her account, is that digital markets tip. Network effects and self-reinforcing data advantages make it harder to break in once a firm is dominant.

That is why both agencies now have pending cases alleging that acquisitions by companies such as Facebook and Google over the last couple of decades were designed to snuff out competition. Inflection points matter most, she argues, because they are when a paradigm can shift and when incumbents are most afraid. Microsoft went after Netscape not because Netscape would replace Windows, but because a browser threatened to make the operating system irrelevant. The enforcement action that followed, in her telling, is what let Google, Amazon, and Facebook grow rather than be "strangled in the crib." AI, she implies, is the new inflection.

Gil's second-order worry is the one every AI investor has: if enforcement gets stricter, some acquisitions will not happen, and the people and capital will not recycle into the next wave. Khan splits the issue. Talent does not need a merger to land in the right place. And even if the goal is an acquisition exit, the FTC's job is only to stop the deal from deepening a monopolistic moat. Thousands of acquisitions still close every year. Less than 3% of deals reported to the FTC and DOJ even get investigated, and an even smaller share is challenged in court.

Her concrete example is pharma, not a chatbot. Sanofi was the monopolist in a drug for Pompe disease, a treatment that requires patients to get hooked up to an IV every couple of weeks. Maze was developing an oral alternative that could be a large improvement. The FTC sued to block Sanofi from buying Maze, on the theory that a monopolist has little incentive to rush a cannibalizing product to market. Sanofi abandoned the deal. Maze then partnered with a firm that did not already hold a monopoly in that market. The FTC had no legal objection. The terms, Khan says, were as good for Maze or better, and the drug is on track to reach patients sooner.

Guo's follow-up is predictability: how is an entrepreneur supposed to know. Khan's first rule is blunt. Selling to the company that is already the monopolist in that market is more likely to raise issues than selling to a company with no presence there. The second is homework. At the end of the previous year the agencies published new merger guidelines that lay out the main frameworks they use. Founders who want fewer surprises, she suggests, should read the document rather than the headlines.

She then asks them to run a longer clock. Today's markets with five major players exist in part because those firms were allowed to buy nascent competitors. A world with one or two suitors is worse for a startup's leverage, valuation, and terms than a world with eight, nine, or ten. Strict enforcement may take some exits off the table in the short run; over time it is what keeps the buyer list long. Reduced to a sentence, antitrust is about making sure the best ideas can win, that a founder can get capital and a foothold without a threatened giant using another line of business to cut off the oxygen. Those, she says, are the arteries of commerce.

Guo asks the awkward question hanging over foundation models. If antitrust is about letting ideas win, how much do ideas and research still matter against the cost of GPU data centers. Khan does not answer the compute-versus-ideas ledger directly. She answers with an industrial preference she has already stated in public: she is in favor of open-weight models. Layers of the stack can deprive companies of key inputs. Open weights, she says, let more people tinker without everyone putting up enormous startup costs, and history says openness has been a key driver of that kind of innovation.

The other constituency she is hearing from is not model builders. A few months before the taping, the FTC sat down with authors, graphic designers, even fashion models. Their story was the same: they woke up to find a life's work ingested without anyone asking or paying, then watched models spit out content that appropriated their style and sometimes competed with them. They were quick to say they are not anti-AI and can see benefits for creators and the public. The long-term fear, as Khan relays it, is that if appropriation is allowed, the incentive to invest in information and vigorous journalism collapses. Getting that piece right, she says, will matter.

On proposals such as registering models above a parameter cutoff, she declines to write Congress's bill. She is a law enforcer, not a freelance legislator. What she will say is that not all rules are equal. Founders have told her they fear regulations designed, in effect, to protect the big guys, rules so complicated that you need hundreds of lawyers to know whether you are in violation. The FTC's own rulemaking, she claims, tries to stay on the other side: clear, simple, easy to follow, no army of lawyers or lobbyists required.

The exhibit is the non-compete ban the Commission finalized a few months before the interview, covering the vast majority of employment contracts. California, she notes, has treated non-competes as unenforceable for centuries, and there is research tying that to Silicon Valley's organic diffusion of ideas. The federal rule is easy to follow and creates no special privileges for large employers. Guo calls it entrepreneurial catnip. Khan also wants the process itself open: regular Commission meetings anyone can attend, and her own trips to Silicon Valley, because sitting in Washington over-indexes on incumbents and under-indexes on founders.

Gil notes that the founding generation was mostly in its twenties and thirties, that the first head of the NIH was 26, and that John F. Kennedy entered the House at 29. Why are so few young people in government now. Khan points to the Great Recession. Millennials who graduated into it faced scarce jobs, never fully made up the lost ground, and still feel priced out of buying a home, as if the ladder had been pulled up. That generation, she says, learned viscerally that what government does, or fails to do, changes a life. Wanting laws that create opportunity for millennials is, in her analogy, the same instinct as not wanting market incumbents to lock everyone else out.

Running the agency was a different shock. In prior jobs she could go extremely deep on one or two things and feel she had mastered every last detail. The chair's job demands breadth over depth and abrupt context-switching: which investigation to bring, which theory to include, how to allocate the budget, what workplace flexibility should look like. Her management rule is to name her comparative advantage, protect time for it, and hire a team that covers the rest. Getting the right people in place takes time. Once they were there, she says, the agency could fire on all cylinders.

How should the public measure a body like the FTC. Activity and deterrence are real, and several people have noted how historically active the Commission has been. Khan's own test is narrower: is the agency addressing the major problems in its purview that people actually face. That is why healthcare has been a priority, hospital consolidation, pharma tactics, anything that hikes the cost of medicine until patients ration care, skip doses, get sicker, or die. The other test is whether markets stay open enough for a thousand flowers to bloom, rather than letting incumbents lock out the next wave of founders. That is the same sentence she has been saying since the chicken farms.