How to Break Up Big Tech — and Why Antitrust Is Suddenly Back

For most of my adult life, the idea that you could break up big tech sounded like a fantasy — the kind of thing a few academics muttered about while the rest of us clicked “accept.” A handful of companies sat astride commerce, information and attention, and questioning that felt as pointless as questioning the weather. Yet here we are: regulators in the United States and Europe are once again treating dominance as a problem to be solved rather than a fact to be admired. Antitrust, dormant for a generation, is suddenly back. I want to walk through why it went quiet, why it returned, what tools are actually on the table, and — the part most people skip — where breaking up market power stops short of fixing the deeper thing.

Why antitrust went to sleep

To understand the revival you have to understand the long slumber. For roughly forty years, competition enforcement in America operated under a single, narrow question: are consumers paying higher prices? This is the “consumer welfare standard,” and it quietly rewrote what monopoly meant. If a company grew enormous but its products stayed cheap — or free — then by this logic there was no harm and nothing to answer for.

Big tech was practically engineered to sail through that test. Search cost nothing. Social networks cost nothing. Two-day shipping felt like a gift. How do you argue “consumer harm” about a service the consumer pays no money for? Enforcers looked at the price tag, saw a zero, and moved on. Meanwhile the actual currency — our data, our attention, our dependence — never showed up on the ledger they were trained to read.

So the giants grew, and they grew in a particular way. They didn’t just win markets; they became the ground the markets stood on. An app has to pass through an app store. A merchant has to appear in a search result or a marketplace. A publisher lives or dies by an algorithm it cannot see. This is the shift from competitor to gatekeeper, and it is the heart of why the old framework failed. A gatekeeper doesn’t need to raise prices to exercise power. It sets the terms of the game, then quietly plays the game itself. I’ve written before about how technology gets captured — how a tool that starts as a convenience hardens into an intermediary you can no longer route around — and gatekeeping is that pattern operating at the scale of the entire economy.

Why it came roaring back

The revival wasn’t one event; it was accumulated evidence finally becoming impossible to ignore. A generation of researchers and, eventually, regulators started arguing that the price-only test had missed almost everything that mattered. Quality decays. Privacy erodes. Small competitors get bought before they can threaten anyone, or get cloned and starved. Innovation narrows to whatever the platform permits. None of that shows up as a higher price, and all of it is harm.

A gatekeeper doesn’t need to raise prices to exercise power. It sets the terms of the game, then quietly plays the game itself.

Two things sharpened the mood. First, the sheer concentration of the moment — a few firms controlling the pipes of digital life — made the abstract concrete. When the same company runs the marketplace and sells the best-selling products on it, or owns the phone platform and the store that taxes everything sold through it, the conflict of interest is not subtle. Second, artificial intelligence raised the stakes overnight. The same incumbents who own the data, the cloud and the distribution are now positioned to own the next platform shift too. I’ve laid out that danger in detail in the AI monopoly: the worry isn’t just that today’s giants are big, it’s that they may be locking in the next era before it has properly begun. Ask who owns AI and you keep arriving at the same short list of names.

In Europe this shift became law. The Digital Markets Act names certain very large platforms as “gatekeepers” and imposes obligations on them before any wrongdoing has to be proven in a years-long case — do-nots and must-dos baked in up front: don’t self-preference your own services, do let users uninstall pre-loaded apps, do open up messaging and app distribution. In the United States the pressure has come more through litigation, with agencies bringing and reviving cases against the largest platforms. I won’t pretend to narrate outcomes that are still being fought over; several are ongoing, and regulators have argued positions that courts have not fully ruled on. But the direction is unmistakable. The presumption has flipped from “bigness is fine unless prices rise” toward “gatekeeping itself deserves scrutiny.”

The tools actually on the table

“Break up big tech” is a slogan. Underneath it sits a real toolkit, and the tools differ enormously in how blunt they are.

Structural break-ups

This is the boldest option, and it has precedent. In 1911 the US Supreme Court broke Standard Oil into dozens of separate companies. In 1984 AT&T was split apart, its long-distance business severed from the local carriers — a divestiture that genuinely reshaped the telecoms landscape and, many argue, cleared room for decades of innovation that followed. Structural remedies mean forcing a company to sell off or spin out a piece of itself: separate the marketplace from the products sold on it, the ad exchange from the ad buyer, the operating system from the app store. The logic is simple and durable — remove the conflict of interest by removing the ability to sit on both sides of the table.

Blocking acquisitions

Much of big tech’s dominance was purchased, not built. A rising rival is easiest to neutralise before it grows — buy the photo-sharing app, the messaging upstart, the smart-home company, the AI lab. Merger review is the cheapest, least disruptive intervention available, because it prevents concentration instead of trying to unwind it later. Enforcers have grown far more willing to challenge deals, and to scrutinise the quieter arrangements — investments, licensing, “partnerships” — through which incumbents attach themselves to promising startups without technically buying them.

Conduct remedies

Short of a break-up, you can regulate behaviour: bar self-preferencing, ban the most coercive default deals, require a company to deal fairly with rivals who depend on its platform. Conduct remedies are less drastic and politically easier. They’re also the easiest to game. A firm with a thousand engineers and a strong incentive can usually find the next workaround faster than a regulator can write the next rule. Behavioural remedies need constant supervision, which is precisely what under-resourced agencies struggle to provide.

Interoperability and the app-store fights

The most interesting tool may be the least dramatic: forcing systems to open up. The app-store battles are the clearest front — fights over whether a platform can force every developer through its own payment system, take a large cut, and forbid apps from even mentioning a cheaper option elsewhere. Prise those doors open and you change the economics for millions of developers without dismantling a single company. Interoperability generalises the idea: make platforms expose the seams so others can build on, connect to, and compete with them.

How a break-up could actually work — and where it stops

Picture the cleanest version. You take a company that runs both a marketplace and a line of products competing on that marketplace, and you separate them into two firms with different owners. Overnight the incentive to rig search results in your own favour disappears, because “your favour” no longer exists. Do the same with an operating system and its app store, or an ad exchange and the brokers bidding into it. On paper, elegant.

Now the honest part, the part a cheerleader would skip. Breaking up market power does not, by itself, fix who owns the data and the infrastructure. Split a giant into three and you may get three companies each still sitting on a mountain of behavioural data no newcomer can match, still running the cloud regions half the internet is hosted on, still enjoying the network effects that made them dominant in the first place. Standard Oil’s descendants remained enormous and profitable for a century. A structural remedy redistributes power between corporate entities; it does not necessarily return any of it to you.

Breaking up market power does not, by itself, fix who owns the data and the infrastructure.

There are real trade-offs, too, and I don’t think pretending otherwise helps anyone. Some of these platforms are genuinely convenient because they’re integrated; clumsy separation can degrade the very services people rely on. Break-ups take years and armies of lawyers, during which the technology moves on and the remedy risks fighting the last war. And a fragmented market is not automatically a fairer one — it can simply be a more chaotic version of the same extraction. None of this is an argument for doing nothing. It’s an argument for being clear-eyed about what a break-up buys and what it doesn’t.

The complementary fixes

This is why I’ve come to see antitrust as necessary but not sufficient. If the deeper problem is ownership of data and infrastructure, then the remedies have to reach there too.

• Interoperability and data portability. If you could leave a social network and take your connections with you — if your messages, your followers, your history were yours to carry elsewhere — the lock-in that makes these platforms unassailable would loosen. Portability turns a walled garden back into a place you can choose to stay, rather than a place you can’t afford to leave.
• Data rights. Treat the behavioural data these systems harvest as something you have a stake in, not merely a resource they own because they collected it. This reframes the whole contest: not “which company controls the data” but “on what terms is it collected and used at all.”
• Public and open alternatives. Some infrastructure may be too foundational to leave entirely to private gatekeepers — protocols, standards, even publicly funded models and datasets that anyone can build on. Not every layer of digital life has to be someone’s toll booth.

These aren’t alternatives to antitrust; they’re what makes antitrust stick. A break-up removes a conflict of interest today. Interoperability, data rights and open alternatives change the underlying conditions so the same concentration doesn’t simply reassemble itself tomorrow under new corporate names. This is also where the conversation connects to something larger than competition policy — the argument that we’re drifting toward what some call technofeudalism, an economy where a few platform lords own the ground everyone else works on and collect rent for access. Antitrust chips at the lords; the complementary fixes are about who owns the ground.

The capture lens: rewriting the last step

I keep returning to a single pattern, because it explains so much of this. Powerful technologies tend to move through the same steps: something new and genuinely liberating appears, it spreads, it becomes essential, and then — this is the fourth step — it gets captured by whoever controls the chokepoint, who turns a shared capability into a private toll. Search, social, mobile, the cloud, and now AI have each walked some version of that path. The people who capture the technology take the value; the rest of us pay, in money or data or attention or lost alternatives; and for a long time it looks as though no one can fight back.

But there’s a fifth step, and it is not fixed. It can be rewritten. Antitrust — break-ups, blocked deals, conduct rules, forced interoperability — is one of the main ways a society goes back and edits that final step, refusing to accept capture as the permanent ending. Standard Oil and AT&T are proof that the ending has been rewritten before, against companies that seemed every bit as untouchable as today’s. The revival happening now, on both sides of the Atlantic, is that same act of editing, aimed at platforms that spent a decade being told they were beyond reach.

I’m not naïve about it. Antitrust alone won’t hand the future back to us — it’s slow, it’s partial, and on its own it leaves the data and the infrastructure largely where they sit. But it is a lever, and after a generation of being told there were no levers at all, that matters. The question was never really whether these companies are too big. It’s who takes, who pays, and whether the rest of us get any say in how the story ends. For the first time in a long time, that last step is up for negotiation again — and the worst thing we could do is decide, once more, that it’s just the weather.