Ben Thompson
What this covers
Ben Thompson sits down with Tyler Cowen to explain how tech companies actually work—not through the lens of moral judgment or accusations of incompetence, but by examining their business models, incentives, and the structural forces that shape their decisions. The conversation moves across the digital economy, from advertising and social media to e-commerce and China's startup scene, building out a framework for understanding why companies do what they do. Rather than assuming strategic blunders stem from stupidity, Thompson traces them to company culture, network effects, and the shift from controlling supply to controlling demand—a distinction that reorders how you should analyze tech competition entirely.
The core insight threading through the discussion is Controlling Demand vs Controlling Supply: in traditional industries, power came from limiting what consumers could access; in tech, it comes from owning users and attention, which makes acquisitions of user bases exponentially more valuable than they appear on a spreadsheet. This framework illuminates everything from why Amazon tolerates years of thin returns on logistics infrastructure to why Regulation as Competitive Moat actually advantages large incumbents over smaller competitors. Thompson also explores how apparent failures—Microsoft's consumer stumbles, the initial shape of online advertising, blockchain's unresolved killer app—can be explained by the mismatch between old-economy thinking and what new formats allow. The conversation touches on basketball analytics, the Cultural Revolution's role in China's competitive advantage, the scale of internet markets, and why Convenience Always Wins even when a product works imperfectly. Throughout, the emphasis lands on examining the actual structure of incentives and value chain economics rather than celebrating or condemning individual companies.
Thompson argues that tech companies' strategic behavior—including their PR failures, M&A success, and platform dominance—is best understood through their business models, network effects, and the shift of power from controlling supply to controlling consumer demand, rather than through assumptions of stupidity or pure morality.
- Power in tech comes from controlling demand (owning users/attention), not controlling supply/distribution
- Apparent strategic mistakes usually have rational explanations rooted in company culture and incentives
- Network effects and zero-marginal-cost economics fundamentally change how strategy should be analyzed
Incumbent tech platforms remain dominant even against marginally better competitors because their scale and integration create compounding advantages.
- The scale of the internet is astronomical—massive numbers of new blogs are created daily even when 'no one' supposedly blogs anymore—which means a paid newsletter targeting the entire world can find plenty of subscribers in any niche (Apple, China, basketball), and even Silicon Valley underestimates what it means to be able to access billions of people.
“the internet — it’s bigger than anyone thinks. Even people in Silicon Valley, I think, underestimate at times just what it really means to be able to access billions of people.”
- Each new generation wants to be somewhere different from the prior one—as kids not wanting to be on Facebook drove Snapchat—so new social services will keep emerging; the real barrier to a meaningful third entrant is not user adoption but monetization, because Facebook owns so much of the social advertising market.
“Every new generation comes along, and they want to be somewhere different than the generation that came before. That was very much what drove Snapchat”
- Although Facebook and Instagram are very different products to consumers, they are essentially the same product to advertisers—bought the same way through the same platform with the same salespeople and ad formats—which is precisely why Facebook owning Instagram is a competition problem in digital advertising: a marginally-better competitor like Snapchat isn't worth the effort when Instagram is so cheap to add to an existing Facebook buy.
“if you look at it from an advertiser perspective, they’re very much the same product. You buy the ads in the same way through the same platform.”
- Large companies like Google and Amazon can launch many products that fail without consequence because their core products with huge network effects (or money-making businesses) throw off profits that subsidize endless experimentation—Amazon functions almost like an internal VC reinvesting profits into many initiatives that don't work out.
“it doesn’t matter because they have that core product with the huge network effects that lets them do that end fail.”
- The filter bubble narrative is almost too neat; the most polarized age groups tend to be older generations exposed to things like cable news, and since platforms gain power by giving people what they want, polarization may largely reflect that people want polarizing content—so attributing causation to Facebook is questionable, though Facebook's own no-effect study was deeply problematic and its effect is likely greater than Facebook admits.
“it’s a little easy and cliché to blame Facebook when Facebook may actually just be a reflection of society at large.”
Owning users and attention, not controlling supply, is the fundamental source of power in internet-era tech.
- Modern big tech companies enjoy good public standing because their business model is making people happy and they succeed by being better than competitors, unlike old monopolies that controlled distribution and had no need to serve customers well because customers had nowhere else to go.
“They’re very different than the dominant companies of old that were like monopolies based on controlling distribution. They had no need to service their customers well because their customers couldn’t go nowhere else. In this case, they succeed by virtue of being better than everyone else”
- In traditional value chains you gained power by controlling supply so consumers had to come to you on your terms, but in tech you gain power by controlling demand—owning all the users so suppliers come to you on your terms; therefore acquiring a company's user base and adding it to yours increases your power exponentially in the value chains you compete in, which is why winner's curse doesn't apply to consumer tech acquisitions.
“it used to be that you gained power on a value chain by controlling supply, and consumers had to come to you on your terms. Whereas, in tech, you gain power by controlling demand, by having all the users and the suppliers come to you on your terms.”
- Facebook's enormous power comes from being the digitization of your friends and family, which is why it has penetrated everywhere—in places like blue-collar Wisconsin, friends and family are central and Facebook is the social network while Twitter is merely known of but unused, a reality less visible to Silicon Valley where many people are running away from their friends and family.
“Because Facebook is the digitization of your friends and family. Where I’m from, friends and family is very central in a way that in Silicon Valley — where a lot of people are running away from their friends and family”
- On the internet power comes from owning attention and owning the consumer, and ByteDance's algorithmic approach has gained the attention of a huge consumer audience in a way almost no one in China outside Tencent has, giving it significant potential for advertising and driving users to various services.
“In the internet, generally, power comes from owning attention, from owning the consumer, and they have the attention of a lot of consumers in a way that, pretty much, no one in China, outside of Tencent, does.”
Smart founders and managers make seemingly obvious mistakes because their company culture, business model, and prior success create systematic blind spots.
- Founder-led companies are missionary in their sense of mission and true believers in their own good intentions, which creates blind spots: they cannot even imagine anyone thinking poorly of them because they know their own good intentions—a reverse attribution error that leaves them unprepared to answer legitimate criticism.
“It’s not that they’re bad at public relations. It’s that they can’t even imagine the possibility of anyone thinking about them poorly because they know their own intentions, and their intentions, they feel, are good. It’s the old attribution error”
- When companies make seemingly obvious strategic mistakes, the explanation is almost never that they are stupid—these are brilliant people; rather, the mistakes are driven by company culture, business model, and prior success that creates blind spots, as when Microsoft assumed it was good at consumer products after winning via enterprise-derived network effects and then built consumer products that failed.
“a lot of people — they don’t have anything beyond “Because they’re stupid.” The reality is, if you know these companies, you know the people in them — they are anything but stupid.”
- The fundamental and broadly generalizable premise of Christensen's Innovator's Dilemma is that managers can do the 'right thing'—serving their best customers and adding features people want, exactly what they're incentivized to do—and yet fail and become susceptible to disruption precisely because of it.
“the fundamental premise is managers can do the “right thing” and fail.”
- The most useful business reading for Thompson is founding stories of companies and their very early decisions, because companies' later seemingly-dumb decisions are often embedded in their culture, and understanding that early culture explains why they do what they do.
“The books I find particularly useful for what I do is the founding stories of companies and going back to decisions made very early because... it’s often embedded in their culture about why they do that, and understanding that is useful.”
Amazon's modular structure and capital-intensive real-world assets let it succeed in conglomerate expansion where other tech companies would fail.
- Amazon is the anti-Apple: rather than tightly integrated, it is a modular entity where teams have well-defined ways to work with other teams (exemplified by how AWS was built to serve internal needs as it would any third party), and this modular internal structure—combined with capital-intensive, hard-to-duplicate infrastructure—makes Amazon far better suited to expand in a conglomerate structure than other tech companies.
“Amazon is the anti-Apple in many respects, where it all fits together absolutely perfectly. Whereas Amazon’s a very modular entity, where all the different teams have well-defined ways to work with other teams that lets it build other things that sit on top of that.”
- Amazon is more politically savvy than people realize: it fought sales tax for years, then when the wind shifted agreed to pay everywhere while simultaneously building distribution centers near major cities—turning the tax into an advantage; the Wayfair decision now burdens small e-commerce businesses facing thousands of tax jurisdictions while Amazon handles it easily, and Amazon similarly got ahead of minimum-wage rises for cheap PR while clawing back bonuses.
“When they saw the wind shifting, they said, “We’re going to pay sales tax everywhere, and oh, by the way, we’re building distribution centers next to all the major cities.” So they actually turn it into an advantage”
- Amazon's strategy of building expensive distribution centers and logistics chains—dismissed by others as not delivering good returns on capital—pays off not in a three-to-five-year range but in creating an impenetrable moat from physical real-world assets that no competitor can afford to replicate in the long run.
“They do all this really hard, expensive stuff that everyone says doesn’t make sense... without realizing that the payoff is not in the three-to-five-year range. It’s in the impenetrable moat that comes from actually having physical assets in the real world that no one can afford to compete with in the long run.”
- Apple is fundamentally a personal computer company and what counts as a personal computer has migrated ever closer to the body—desk, bag, pocket, and now wrist; the closer something is to you, the more the small details of the physical experience matter, which is why Apple's strengths in feel and interaction make it best positioned for wrist devices and eventually face-worn augmented reality.
“the closer something is to you, the more the small stuff really matters.”
Product discovery requires experimentation and willingness to let most attempts fail, not top-down strategic planning.
- You need to build new products to discover what they're for, because trying to anticipate and plan everything generally produces a less dynamic outcome than experimenting and seeing what works—history shows the Silicon Valley willingness to let people do crazy things, most of which fail, drives innovation better than a top-down 'this is what it's for' process.
“We have a lot of history that trying to anticipate and plan everything generally results in a less dynamic outcome than experimenting and seeing what works.”
- We do see tons of bad strategy in the form of failed startups—that is the model of Silicon Valley; the VC model is not about finding great strategy and the right team, but funding many teams, keeping the ones that gain traction, and building from there, so strategy is a follow-on to initial execution and product-market fit.
“The venture capital model, in many respects, is not about having a great strategy and in finding the right team to implement it. No, it’s funding a whole bunch of teams, get the ones that get traction, and then build up from there. So, strategy is very much a follow-on to initial execution.”
- Netflix expanded to ~160 countries quickly mostly by flipping a switch—often without even localizing the app into the country's language at first—which was less a viable localized product than a demonstration that, unlike traditional networks focused on a single market and on monetizing upfront, Netflix competes asymmetrically by targeting the entire world and monetizing over many years; deeper localization came only in the last couple of years.
“a lot of it was literally just flipping a switch, and “You can now download our app and sign up.” There was no real localization effort”
Internet economics—zero marginal cost and destroyed scarcity—create fundamentally different business dynamics than traditional models academic concepts assume.
- Every time a new paradigm arrives, everyone first tries to copy the prior format (e.g., slapping ads next to articles online as in newspapers), which fails; it's the second wave that figures out what is native to the new format (e.g., ads injected into the feed)—and blockchain's killer use will be whatever is uniquely possible with it, not the recreation of existing products.
“every time there’s a new paradigm that comes along, everyone starts out trying to copy what came before, but in the new format... It’s the second wave that figures out what is native to the format.”
- The internet drove distribution cost to zero but also ended scarcity since anything can be copied endlessly; what makes blockchain-derived products interesting is they can retain that near-zero distribution cost while also being scarce—uniquely allowing something to span the whole world while having only one of each particular item.
“what makes blockchain–derived products interesting is, they can have that zero-distribution-cost aspect... while also being scarce. That’s a really interesting concept of being able to basically go over the whole entire world, yet there be only one of every particular item.”
- Academic business concepts claimed to be universal break down for internet companies because once you put zero marginal costs into the equation it becomes a totally different business that is hard to draw concepts from, which is the void Stratechery aimed to fill with practical, immediately useful strategic analysis of current decisions rather than abstract case studies.
“once you put zero marginal costs into the equation, you can think it through if you think really hard, but it’s a totally different business, and it’s hard to draw concepts from.”
Convenience always wins in consumer adoption despite imperfect functionality or other concerns.
- Particularly for consumer products, convenience always wins in the long run, which is why home devices like Alexa will become more popular despite imperfect current functionality—people love convenience, so even partial conveniences like hands-free answers or automated lights will drive adoption.
“The reality is — particularly when it comes to consumer products — is that in the long run, convenience always wins.”
- Smartphones haven't made humans smarter or stupider—humans are humans, and adults tut-tutting kids on phones forget they themselves spent hours on the home phone with friends; the upheaval is here and the best response is to figure out how to manage it rather than pine for a world that's gone, while remaining aware of downsides like bullying that have always accompanied new technology.
“People that are tut-tutting the kids on the phone — they forget that when they were kids they literally sat at home with a phone against their ear, talking to their friends for hours at a time, and if they had a smart phone, they’d be doing the exact same thing.”
Internet writers must maintain awareness that their most vocal feedback does not represent their actual audience distribution.
- The internet enables many small-scale, bootstrappable niche businesses serving previously unservable geographically-dispersed audiences via social media and digital marketing, but venture capitalists are tempted to fund these even though there are no venture-level returns to be had—the idea or model isn't bad, but the returns aren't there.
“all those businesses are not venture-style businesses... It’s not that the idea is bad or the business model is even bad. It’s that there’s not venture-level returns there to be had.”
- A key lesson for internet writers is that your core group of readers who message and tweet you all the time are not representative of your audience as a whole, so you must maintain a mental filter to remain aware of that bias while still valuing the immediate, adjustable feedback the internet provides.
“you have a core group of readers that are going to send you messages and tweet you all the time, and they’re not representative of your audience as a whole. Having a mental filter around that, to be aware of it.”
- Amazon escapes the PR trouble Facebook and Google face partly because it is removed from media tensions and is customer-focused, but also because its explicit strategy is to not engage—whereas Facebook and Google are eager to explain themselves, and it is that very explaining that gets them into trouble.
“by and large, they aren’t generally in the thick of it, whereas Facebook and Google, particularly Facebook — they’re very eager to explain themselves. It’s arguably that explaining themselves that gets themselves in trouble more than anything else.”
- The Milwaukee Bucks will have a great year and Giannis will win MVP because they replaced the worst coach in the league and now have the best shot profile in the entire league in the preseason—shooting threes or shots at the rim at an ~89 percent rate.
“They actually have the best shot profile in the entire league in the preseason, which is shooting threes or shooting shots at the rim. They are 89 percent or something like that. They are going to have a great year. Giannis is going to win MVP.”
- Giving fouled teams the option to either take free throws or take the ball out of bounds would fix two problems: intentional hacking of poor free-throw shooters that grinds games to a halt (since a foul should punish the defense, not the offense), and the end-of-game parade to the free-throw line, together making games faster and more exciting.
“I would also give teams the option, when fouled, that they can either take penalty free throws, or they can take the ball out of bounds... The point of a foul is to punish the defense, not to punish the offense.”
- Apple has been the most successful company in China because its device-selling model with differentiated software avoids piracy and hardware copying, but that same exposure makes it uniquely vulnerable: if China wants to apply pressure in the trade war it can squeeze Apple on both the supply side (manufacturing) and the demand side (sales)—a risk other tech companies lack because China isn't a meaningful market for them.
“Apple is uniquely exposed right now to the trade war that’s going on... if China really wants to put the squeeze on, they can squeeze Apple on both sides, on the supply side and on the demand side.”
- The hardest part of living in Taiwan as a foreigner is not the language but the constant sense of always being visibly different—never knowing whether a slight is due to being a white guy—and those who can adopt an 'it is what it is' attitude survive while those who let it drive them crazy leave within a month or two.
“every time you’re in public, at all moments, you’re always aware that you are different... if you can chill out and be like, “It is what it is,” then you’ll have a better chance of surviving.”
- From an intellectual standpoint an MBA is probably overrated—you won't become a strategic maven from reading case studies, and MBAs get a deserved bad rap for thinking they know everything—but MBAs are well documented to do wonders for your salary; the value Thompson got was that it gave him the language to articulate business intuitions he already had.
“From an intellectual pursuit, probably a little overrated... MBAs do do wonders for your salary, and I think that’s very well documented.”
- The bigger moral problem with performance-enhancing drugs is not professional athletes—who are paid tremendously and bear their own risks—but the spillover effect on high school and college players who must live normal lives afterward; leagues are only incentivized to do the minimum to avoid bad press, and removing marijuana from drug-testing programs could make those programs stricter and more equitable for everyone.
“when you have a high school player or a college player that is just in it, and they’re going to have to go live a normal life afterwards, then it’s much more problematic from a moral perspective.”
- TV advertising has stayed stronger than expected because its advertisers—the mass-market CPG, car, beer, and phone companies seeking least-common-denominator reach—are integrated with big-box retailers, car-based shopping, and sports into one interconnected old economy; all parts of that value chain are under simultaneous internet threat, so by standing together the weak stand longer, but when it collapses it will be a total collapse, giving way to internet attention markets and individualized niche marketing.
“the reason why TV advertising has remained stronger than people anticipated is because the TV advertisers are unique, are tied in. It’s all integrated.”
- NBA teams remained in an equilibrium of taking too few three-pointers—despite the obvious math that a third of threes equals half of twos—because coaches and front offices are largely former players with decades of conditioning about 'how you play basketball,' so the shift required outsiders bringing a math-based mindset (like the Rockets' Daryl Morey), which generated resistance and took time to break through.
“coaches and front-office executives — probably the majority or a fair number — are former players. They came up through the game, and so they have 10, 20, 30, 40 years of conditioning about, “This is how you play basketball.””
China's scale, density, and labor abundance create higher valuations and service possibilities than the more dispersed United States.
- Chinese startup valuations like ByteDance's $75 billion make sense because China's 1.2-1.3 billion people form a common market—same language, same shopping, far more uniform than the EU but three times the US—and because Chinese cities are so large and dense and labor is so abundant, the level of services and labor-intensive offerings possible, and thus the upside, is far higher than in the more spread-out US.
“All things with China start and end with the size of the market and the potential there.”
- The Cultural Revolution is one of the most underrated events of the last hundred years not just for its deaths but because it was a scorched-earth clearing that let China start economically from scratch (enabling long growth) and produced a bare-knuckled, fast-moving, brutally competitive business culture; India, lacking such a clear-the-decks event, retains thousands of years of culture and colonial legacy that make getting things done harder, so the two countries will develop fundamentally differently.
“the most underrated event... in the last hundred years — is the Cultural Revolution in China... it really was like a scorched earth for China as a whole. Everything started from scratch.”
- Taiwan has produced internationally successful executives because of its island-nation openness and outward orientation—looking favorably to America and Japan and pulling culture from abroad—whereas South Korea is more internally focused, which benefits Korean companies' domestic base (Koreans buy Samsung because it's Korean) but contrasts with Taiwanese who buy whatever phone is best rather than HTC because it's Taiwanese.
“Taiwan has always been very open to the outside world in a very island-nation sort of way, very cognizant of their place in the world and their dependence on other countries.”
Tech expansion into regulation-heavy, real-world sectors like healthcare requires mindset shifts most founders and investors avoid.
- Tech is built on heavy upfront investment and near-zero marginal cost products, but as companies expand into the real world they face inescapable marginal costs—regulatory compliance, ongoing lobbying, and other sticky real-world expenses—which means the biggest growth opportunities lie in regulation-heavy areas like healthcare that most avoid, requiring a mind-shift among VCs and founders.
“Tech, in general, is all about the heavy investment up front, and then basically zero marginal cost products on the back side. The reality is, as you get more into the real world, those marginal costs become more and more difficult to escape.”
- The internet won't fragment into a distinct EU version the way China is distinct; instead, big US tech firms will keep doing business in the rich, populous EU no matter how annoying the regulations, and will actually become stronger there because incumbents are better at handling heavy regulation—so the EU internet will be the same city with a fence, not a different country.
“they’re going to be even stronger in the European Union, in part because dealing with all the regulations like that, it’s going to be a traditional favorite; the incumbents are going to do better at it.”