Selective exit: when vendors leave the EU market

When commentators speak of regulation creating leverage, the implicit model is: regulator passes rule, vendor adjusts product, market continues with adjusted product. The European reality of 2025–2026 is increasingly different. Vendors have a third option: withdraw the product from the EU market entirely.

This chapter documents the cases.

Apple Intelligence — and what came after

In June 2024 at WWDC, Apple announced Apple Intelligence — a suite of generative AI features integrated into iOS, macOS, and iPadOS, including enhanced Siri, generative writing tools, image generation, and ChatGPT integration. Apple initially declined to ship Apple Intelligence in the EU at launch, citing "regulatory uncertainties brought about by the Digital Markets Act." After roughly nine months of delay, the features arrived in the EU with iOS 18.4 in April 2025.

The Apple Intelligence delay turned out to be the opening move, not the whole story. In a September 2025 statement on the DMA's impact on EU users, Apple confirmed that two further features remain unavailable in the European Union: iPhone Mirroring (which lets a Mac control an iPhone) and Live Translation with AirPods. Apple's stated reason in both cases is the same — interoperability obligations under the DMA would, in Apple's view, force these features to be exposed to third parties in ways the company is unwilling to support.

For European Apple customers, this is concrete loss: identical hardware, identical operating system, but pieces of advertised functionality are not delivered. And the pattern matters more than any single feature. The selective-exit toolkit — delay, partial release, indefinite hold — is now a routine part of how a major US vendor responds to EU regulation.

Meta multimodal Llama

Meta announced in July 2024 that its future multimodal AI models will not be offered in the EU — citing "the unpredictability of the European regulatory environment". Multimodal Llama variants — capable of handling images, audio, and other inputs alongside text — are available in the U.S. but not in the EU.

This is not just a consumer-facing product withdrawal. Meta's models are widely used by other developers (academics, researchers, startups building on Llama). The EU exclusion cascades into the entire ecosystem of European developers building on Meta's open-source models.

What is novel and what is not

Selective product withdrawal is not new. Companies have always made country-by-country availability decisions based on regulatory complexity, market size, and operational cost. What is new is the scale — the EU is now large enough and regulated enough that withdrawal is a realistic option for a major U.S. firm — and the specificity — the products being withheld are exactly the AI products that the regulatory framework is trying to govern.

The DMA was designed to create competition through interoperability requirements and gatekeeper accountability. The actual effect, in the AI case, is that the products subject to the regulation never reach the EU market in the first place. There is nothing to interoperate with, no gatekeeper to discipline, because the gatekeeper is not present.

What it means for European users

The EU is becoming, in some technology categories, a lagging market. European businesses adopting AI cannot always use the same tools their U.S. or Asian competitors are using on the same timeline. European universities and researchers have less access to certain frontier-model capabilities. European consumers see advertising for features that are not available to them.

This is the cost of regulation without alternatives. The EU has the power to deter foreign products, but not to substitute them. The result is a gap.

What it means for the strategic argument

Selective exit is the empirical evidence that "fines-and-litigate" is not the only response to EU regulation. The third option — withdrawal — is real, increasingly used, and rationally chosen by vendors when regulatory cost exceeds market benefit.