Every time a customer taps a card at a coffee counter, roughly 2.5 percent of the purchase flows silently to the card networks as interchange — and for years, merchants who pointed buyers toward cheaper payment rails were penalized under anti-steering clauses. The mobile economy is now entering its own interchange moment, as Brussels, London and dozens of U.S. state capitals dismantle the anti-steering architecture that has governed app distribution for nearly two decades.
In the six weeks through mid-August 2026, the European Commission fined Google €890 million for, among other breaches, blocking developers from steering users to cheaper purchase channels, while the U.K. Competition and Markets Authority advanced draft conduct requirements that would compel Apple and Google to permit off-platform steering at regulated, evidence-based fees. In parallel, both platform owners are pushing a new compliance stack — age assurance, accountability APIs, target-API mandates — down onto the very developers they are being ordered to liberate.
MARGIN MIGRATION, NOT MARGIN MAGIC
The 30 percent headline commission is already a fiction: effective take rates run 0–30 percent, with most studios at 15 percent or less under small-business programs and subscription step-downs. What steering mandates reprice is not the rate card but the balance-sheet location of cost — payment processing, tax remittance and fraud liability migrate from the platform’s P&L to the developer’s. For a subscription app at $5 million in revenue, swapping a 15 percent platform fee for 3 percent card processing plus billing headcount is a windfall; for a two-person studio, it is a fixed-cost treadmill.
“While it is only fair for Apple and Google to be compensated for the services they provide, any fees they charge must be justified through a robust, evidence-led framework involving due reference to both cost and value.”
— Will Hayter, Executive Director for Digital Markets, CMA, June 30, 2026
Note the concession embedded in that sentence: steering is not free. The fee migrates from a commission line to an access line, and every mobile finance team should model it accordingly.
WHAT THE MICROSOFT DECREE ACTUALLY TEACHES
The closest precedent is United States v. Microsoft. The 2002 consent decree barred the exclusive contracts and bundling at issue but left the platform intact; what followed was not the fragmentation of Windows but its slow irrelevance, as gravity shifted to the mobile platforms Microsoft had been restrained from defending with its full arsenal. The lesson for 2026 is precise: remedies rarely redistribute power to the plaintiff class they name; they redistribute it to the next platform shift. Studios that spent 2003–2007 litigating license terms missed the iPhone; today’s winners will treat regulated distribution as a commodity input and reinvest the difference wherever discovery happens next.
THE COMPLIANCE LEDGER IS A REGRESSIVE TAX
Even as distribution opens, participation costs are climbing. Google Play enforces an Aug. 31, 2026 target-API 36 deadline; Apple’s social-media declaration turns mandatory for new apps in September; Google ships age assurance globally by year-end; and U.S. state App Store Accountability Acts have already pushed platform-level API adoption downstream as verification work for independent developers. Each item is defensible in isolation — child safety is not a negotiable good — but in aggregate they behave as a per-jurisdiction fixed cost, and fixed costs do not scale down. A studio shipping in 40 jurisdictions now maintains 40 regulatory diffs, an overhead only publishers with compliance departments can absorb. Pro-competition regulation thus disciplines incumbents at the platform layer while entrenching incumbents at the application layer.
IN DEFENSE OF THE TOLL: CURATION AS A PRODUCT
The bear case for walled gardens is not fiction. Curation is a security product consumers cannot price individually but buy in aggregate: Apple’s own audit reports over $2.2 billion in fraudulent transactions blocked in 2025, part of an $11 billion six-year total, and more than 90 percent of the $1.4 trillion in App Store commerce last year was commission-free — a direct rebuttal to the blanket-30-percent narrative. Tim Sweeney’s maxim — “when lots of stores compete, the result is a combination of better prices for you, better deals for developers, and more investment in new content” — describes an equilibrium, not a transition plan; the transition externalizes fraud, refund and support liability onto thinner balance sheets. Regulate steering to marginal cost with no contribution to security R&D, and the subsidy runs from users who value a managed store toward the arbitrageurs who leave it.
THE NEXT GATEKEEPER WEARS A CHAT INTERFACE
The deeper implication is that the steering fight may be a rear-guard action. The EU’s refusal of an 18-month DMA exemption for Apple’s Siri overhaul — and the resulting delay of on-device AI features in Europe — signals that interoperability demands will now attach to assistants, not just stores. Once an agent, not a human, browses the ranked grid to choose software, steering becomes a question of who controls the recommendation weights. Android’s 2026 tooling push — a stable Android CLI and AI Studio build pipelines — pushes the marginal cost of producing an app toward zero, moving scarcity from production to distribution. Store-competition theory assumes users choose stores; agentic discovery deletes that layer. The 2027 antitrust docket will be about assistant neutrality, and 2026’s steering rules risk looking like 2009’s feature-phone interoperability rules: technically valid, economically beside the point.
PAPER WALLS AND THE COMPLIANCE-THEATER RISK
Skepticism also runs the other direction, and it is equally well-founded. Every previous opening regime — the EU’s alternative-marketplace terms, U.S. link-out disclosures — shipped with friction: click-through warnings, fee schedules drafted by defendants’ economists, disclosure UX designed to convert openness into alarm. The 2026 rules may likewise produce checkbox compliance rather than functional competition, in which case the “opening” appears in press releases, not in take-rate data. Add the possibility that compliance load itself entrenches the duopoly — a rulebook only Apple and Google can afford to implement — and the honest base case sits between liberation and theater: modest, real fee compression captured disproportionately by developers already large enough to run billing, legal and fraud operations.
A WORKING PLAYBOOK FOR THE NEXT TWO QUARTERS
- Model both rails now. Build external checkout behind a feature flag where steering is legal; run the 15-percent-versus-3-percent-plus-headcount math per SKU cohort, not in aggregate.
- Treat compliance as product. Calendar the API 36 deadline (Aug. 31, 2026), Apple’s September declaration and age-assurance rollouts; budget them as engineering headcount, not legal trivia.
- Internalize the fraud line. Off-platform billing inherits the chargebacks and stolen-card exposure the store used to absorb; price fraud tooling and reserves accordingly.
- Own the relationship. First-party accounts, email and CRM are the only assets that survive both store redesigns and agent-mediated discovery.
- For citizens: keep payments, health and kids’ usage inside managed stores, and treat unsolicited “cheaper outside the app” prompts like a card skimmer — that is where the fraud statistic migrates.
BASELINE SCENARIO: FEBRUARY 2027
Six months out, the CMA’s steering requirement finalizes close to draft form and Google’s global terms normalize off-platform checkout on Android; iOS steering stays narrower, contested through an EU-style fee-dispute process. Effective take rates on steered subscription revenue settle in the low single digits while store-mediated revenue remains at 15–30 percent — a two-tier price system for the same inventory. The first DMA-style complaint against assistant recommendation behavior lands in Brussels, and at least one mid-size studio publicly attributes layoffs to per-jurisdiction compliance overhead, supplying the regressive-tax thesis its first causal data point. The walls do not come down; they get repriced. The developers who read the fee schedule as a term sheet, rather than a verdict, are the ones who compound the difference.
Primary sources: European Commission DMA non-compliance decisions, July 23, 2026; U.K. CMA steering conduct requirement consultation, June 30, 2026; Apple Newsroom ecosystem and fraud reports, June 2026.