Apple Lost Its DMA Appeal. The App Store Tax Survived Anyway.

On July 8, 2026, the General Court of the European Union dismissed Apple's attempt to escape the Digital Markets Act. The judgments, in joined cases T-1079/23 and T-1080/23 plus case T-214/24, upheld the European Commission's September 2023 decision designating iOS and the App Store as “core platform services” run by a “gatekeeper.”^[1]Apple's separate challenge over iMessage was tossed as inadmissible, because iMessage was never actually put under DMA obligations, so there was nothing to annul.^[2]

Apple's cleverest argument, and the one worth understanding, was that it does not run one app store. It runs five. iPhone, iPad, Mac, Apple Watch, Apple TV. Different code, different review pipelines, different economics. Break them apart and most of them fall under the DMA's user and revenue thresholds, and the regulation loses its grip. The court was not remotely interested. In its words, the stores have “the same purpose, namely to connect app developers with end users in order to facilitate the distribution of software applications.”^[2] One purpose, one service, one gatekeeper.

Apple can still appeal to the Court of Justice, but only on points of law, not on the facts.^[3] Practically, this is the end of the road for the argument that Apple should never have been regulated in the first place. Which makes it a good moment to ask the question nobody in the press release war is asking honestly: after two and a half years of this, did the DMA actually change anything?

What the DMA actually requires, in plain English

Strip away the article numbers and the DMA asks Apple to do four things on iOS in Europe.

• Sideloading and alternative marketplaces. Users must be able to install apps from somewhere other than the App Store, and third parties must be allowed to run their own app stores on iOS.
• Steering. Developers must be free to tell users, inside the app, that a cheaper price exists on the web, and link them there. This is the one Apple fought hardest and lost most expensively.
• Browser engine choice.Browsers on iOS no longer have to be skins over WebKit (Apple's rendering engine, the thing that turns HTML into pixels). Chrome could ship Blink. Firefox could ship Gecko.
• Interoperability.Third-party hardware and software must get access to iOS features Apple's own products already use. In March 2025 the Commission issued specification decisions covering nine iOS connectivity features, mostly the ones that make an Apple Watch feel magic and a Garmin feel like a peripheral.^[4]

Apple's argument was structural, and it was never going to work

The five-stores theory is the kind of argument that wins in a world where regulators reason from implementation details. It loses in a world where they reason from function. The DMA was drafted by people who watched a decade of platform companies win antitrust cases on definitional technicalities, and they wrote a law that defines the service by what it does for users, not by how the code is organized.

Which is why the reasoning matters more than the outcome. If the court had accepted that a watch app store and a phone app store are different services because they use different SDKs, every gatekeeper in Europe would have spent the next decade shipping architectural changes designed purely to fall below thresholds. The judgment closes that door. That is the real precedent here, and it extends well past Apple.

Apple's public response was the same one it has given since 2023. The company said it “firmly believe[s] the DMA's mandate goes beyond what is lawful and proportionate, threatening to erode decades of privacy and security protections we've built.”^[3] That is a policy argument dressed as a legal one. It is also, in part, true. More on that below.

Now look at what Apple actually built to comply

Here is the thing that should bother anyone who cheered the ruling. Go read the current EU terms. As of January 1, 2026, a developer selling digital goods in the EU can pay an initial acquisition fee of 2% on purchases from new users, a store services fee of 5% or 13% depending on which tier of App Store features they want, and a Core Technology Commission of 5% on digital sales, which follows the developer even if they distribute entirely outside the App Store.^[7]

Stack those and a developer using external payments can still hand Apple something in the neighborhood of the old commission. The old €0.50-per-install Core Technology Fee is gone, and good riddance, because a per-install charge on free apps was an existential threat to anyone who ever went viral. What replaced it is smarter: a percentage, applied to revenue, on every distribution path.^[7]

That is the case for the cynical read, and it is a strong one. Apple did not open the platform. It re-priced access to it. The DMA said you cannot charge developers for using their own payment processor, so Apple invented a fee for using the platform's intellectual property instead, which is legally distinct and economically nearly identical.

The security trade-off is real, and both sides are lying about it

I want to be careful here because this is where the discourse gets stupid. Apple's claim is that sideloading exposes users to scams, fake banking apps, and malware, and that multiple marketplaces with different review standards fragment trust.^[9] The developer-advocacy claim is that this is pure pretext and sideloading is safe because macOS has always allowed it.

Both are wrong in the same way: they treat “security” as one thing. It is two.

Technical securityis the sandbox, code signing, entitlements, and notarization (Apple's automated malware scan and signing of any app distributed through an alternative marketplace). Sideloaded iOS apps still get all of it. The kernel does not care where the binary came from. On the technical layer, Apple's DMA implementation gives up almost nothing, and any engineer who has shipped an iOS app knows it.

Social securityis the part where a 70-year-old installing a marketplace to get a cheaper Spotify subscription is now making a trust decision that iOS used to make for them. That risk is real. It is not a technical exploit, it is a social-engineering surface, and social engineering is how almost everyone actually gets owned. Apple is right about this and mostly refuses to say it clearly, because “our users are less sophisticated than we let on” is not a keynote line.

The tell that Apple is not arguing in complete good faith is elsewhere. Since iOS 17.4, third-party browser engines have been legal in the EU. More than two years on, not one has shipped.^[10] That is not a security outcome, it is a friction outcome. Apple requires a separate EU-only app rather than an EU variant of an existing one, gives web developers no way to test against alternative engines on iOS, and has not committed to keeping browsers updated for users who travel outside the EU for more than 30 days.^[10] None of that protects anyone. All of it makes shipping Blink on iOS economically irrational. Compliance that is technically complete and practically impossible is a design choice, and Apple made it deliberately.

What this does to the Services line, which is the whole point

Follow the money. In the quarter ended March 28, 2026, Apple posted record Services revenue of $30.9 billion, up 16.2% year over year, on total revenue of $111.2 billion.^[12] Services gross margin was 76.7%. Products gross margin was 38.7%.^[13] Every dollar of Services is worth roughly two dollars of hardware at the gross line. That is why the App Store is the single most strategically protected asset in the company, and why Apple has spent three years and hundreds of millions in fines and legal fees defending a fee structure most people outside Cupertino think is indefensible.

Now the honest part. The DMA has not dented that number and it probably will not. Services grew 16% through the same period that Apple was being fined €500 million and forced to open iOS. The App Store is a habit, not a lock-in, for the vast majority of European consumers. Nobody is installing a third-party marketplace to save 30 cents on a puzzle game. The developers who benefit are the ones with enough scale and brand to pull users out to the web, which is exactly who: Spotify, Netflix, Epic, Amazon. That is a transfer of margin from Apple to large developers, not to consumers.

So who actually won

My honest read, and I say this as someone who ships software and would love the developer-friendly answer to be the correct one: the DMA is good, and it has been badly executed.

It is good because the principle is right. Owning the operating system should not entitle you to a permanent tax on every transaction that happens on top of it, and the court just confirmed that you cannot reorganize your codebase to escape that principle. That matters. It is badly executed because the law regulated the mechanism (the payment sheet, the link, the store) instead of the outcome (the take rate). Apple responded exactly the way any rational profit-maximizing firm would: it complied with the mechanism and preserved the outcome. You can be angry about that. You cannot be surprised by it.

The Commission's next move tells you whether this regime has teeth. If it goes after the Core Technology Commission as a circumvention of Article 5(4), the DMA becomes a real constraint on Apple's economics. If it does not, the DMA ends up being the most expensive API-compatibility project in corporate history: a lot of new surfaces, a lot of new paperwork, and a Services line that keeps compounding at 16%.

Apple lost the case. Check the margin line next quarter and tell me who lost the war.