Why the Street Is Wrong About Apple's Services Multiple

The Services consensus formed between 2018 and 2022 when the segment grew from $40 billion of revenue to $78 billion at gross margins north of 70%. That was a genuinely impressive run. The market correctly identified Services as the high-multiple component of the Apple complex and assigned it a software-style valuation in sum-of-parts models. The bull case argued that the embedded base of 1.4 billion active iOS devices was the moat, and that the Services attach rate would compound for another decade.

The consensus also draws on the App Store economics. The 30% take-rate on covered developer transactions, the 15% rate on small developers and on subscriptions after year one, and the search advertising revenue from the Google default arrangement combined to produce a high-margin earnings stream. At the peak, Services contributed roughly 20% of total revenue and over 30% of total gross profit. The market correctly scaled the multiple to that contribution mix.

The regulatory backdrop sat at the periphery of the consensus through 2022. The Epic Games litigation was treated as a one-off. The European Commission cases on contactless payments and music streaming were treated as manageable. The Department of Justice search engine litigation was viewed as a Google issue rather than an Apple issue. None of these characterisations are wrong individually. Aggregated, they understate the cumulative exposure that has now built up against the Services revenue base.

Line one is the Services growth deceleration. Strip the price increases out of Services and the underlying volume growth rate is mid-single digits at most. The Apple Music price hike from $9.99 to $10.99, the iCloud tier reshuffle, and the Apple TV+ price step-up have each contributed several hundred basis points of optical revenue growth without any underlying subscriber expansion. Real subscriber growth in the trailing twelve months has been roughly 4-6% on the disclosed paid subscription metric of just over a billion. That is not the trajectory that justifies a software-style multiple premium.

Line two is the App Store regulatory exposure. The Digital Markets Act in Europe took effect in early 2024 and has been generating active enforcement decisions. The most recent ruling required Apple to allow alternative app stores and direct-to-consumer purchase rails, both of which directly compress the take-rate economics. In the US, the Epic Games appeal cycle is still active and the recent district court findings on anti-steering provisions have already begun reshaping the merchant disclosure flow. The cumulative revenue exposure to take-rate compression is in the $4-7 billion annualised range over the next 36 months.

Line three is the search engine default arrangement. The DOJ's case against Google's search distribution practices puts the multi-billion-dollar annual payment from Google to Apple at risk. The estimates of that payment range from $18 billion to $22 billion annually. If the remedy phase of the litigation prohibits the default arrangement or requires renegotiation at a materially lower fee, the loss to Apple's Services revenue could exceed $10 billion. That is a structural hit that does not yet appear in any consensus model.

Across three large regulatory exposures, the cumulative downside is meaningful. The Valuation Desk models a 6-9% Services revenue compression scenario over 24 months as the base case, with a 12-15% compression as the bear case if the search default goes against Apple.

Apple generated $416 billion of revenue in fiscal 2025, $133 billion of operating income, and $98.8 billion of free cash flow. Those are extraordinary absolute numbers. The issue is the rate-of-change rather than the absolute level. Revenue growth was 6.4% year-on-year, operating income growth was 8.0%, and free cash flow growth was 9.5%. Those are decent numbers for a mature business at a 22-24x multiple. They are inadequate for a 32x multiple.

Gross margin trajectory tells a similar story. Total gross margin expanded from 43.3% in 2022 to 46.9% in 2025, a 360 basis point expansion that has been driven by mix shift toward Services and component cost deflation in the iPhone supply chain. Both drivers are at or near saturation. Component deflation has compressed in the trailing four quarters as the memory cycle has turned. Services margin has stabilised in the 70-72% range and is unlikely to expand further given the regulatory pressure on take-rates.

The iPhone unit volume trajectory is the unspoken concern. iPhone revenue in 2025 was approximately $202 billion, against $200 billion in 2022. Three years of essentially flat iPhone revenue at constant currency. The China market in particular has compressed, with iPhone unit volume in mainland China down roughly 18% over 24 months on consensus estimates. The Vision Pro launch did not move the needle on hardware diversification. The AI feature rollout has been delayed and the upgrade cycle uplift has not materialised in the data.

The historical pattern for multiple compression in mega-cap technology names is well-defined. The catalyst is rarely a single negative print. It is typically a combination of two or three smaller negative inflections that compound. For Apple, the candidates are clear. A China iPhone print that confirms a fourth consecutive quarter of unit volume decline, a Services growth print below 9% that breaks the consensus floor, or a remedy ruling in the search default litigation that prices in a $5+ billion annual hit. Any two of those three trigger a serious revisit of the multiple.

The historical analogue is the 2012-2013 Apple compression cycle, when the multiple compressed from 16x to 9x over 18 months as the iPhone growth rate decelerated and the margin profile compressed. The current setup is not identical but the rhyme is unmistakable. The market over-paid for an iPhone growth assumption then; the market is over-paying for a Services growth assumption now. The mechanism of multiple compression is the same.

The other historical analogue is Microsoft 2000-2002, where the multiple compressed from 50x to 18x as the antitrust overhang, the dot-com revaluation, and the modest growth deceleration combined to produce a 50% drawdown. We are not predicting a 50% Apple drawdown. We are pointing out that the regulatory overhang is structurally similar and the multiple risk is meaningful at the current premium.

Mid-quarter data points have not been kind to the iPhone hardware story. Foxconn's reported revenue in the September-October 2025 window came in 7-9% below the comparable 2024 print, suggesting a build pace that does not support the consensus iPhone unit assumption for fiscal Q1 2026. Taiwan Semiconductor's mobile-related wafer commentary has been similarly soft. Counterpoint Research's mainland China weekly tracker shows iPhone share in the premium segment has slipped roughly four percentage points over six months as Huawei's recovery and Xiaomi's flagship push have eaten into the high-end. None of these are catastrophic on their own. Stack them and the picture is of an iPhone refresh cycle that is finding the upper bound of its addressable buyer set faster than the consensus model assumes.

The carrier promotional pulse in the US tells a related story. The major US carriers stepped up handset subsidies into the holiday quarter at promotional levels last seen in 2017. Aggressive carrier promotion typically signals demand that needs subsidising rather than demand that is pulling itself through the channel. Strip the carrier subsidy out of the demand build and the underlying organic upgrade rate looks closer to the 2016-2017 trough than to the 2020-2021 super-cycle that anchors the bull case. The data is consistent across three independent sources. The setup does not support the embedded growth assumption.

Vision Pro shipped roughly 350,000 units in fiscal 2025 against an early bull case modelling several million units annually. Real revenue contribution in 2025 was approximately $1.2 billion, or 0.3% of total revenue. The product is a strategic option but the optionality has not converted to a material revenue line. The price point of $3,499, the limited content library, and the form factor constraints have together kept the addressable market narrower than the consensus model assumed. The 2026 trajectory does not change that arithmetic.

Apple Intelligence, the AI feature rollout layered onto the iOS 18 family, has been positioned as the upgrade catalyst that drives an iPhone refresh super-cycle. The data so far does not support that thesis. The mainland China rollout has been delayed indefinitely on regulatory grounds, removing the largest single growth opportunity. The European rollout has been reduced in scope to comply with DMA requirements. The US rollout has produced engagement metrics that are below the comparable Samsung Galaxy AI launch metrics in the same product window. None of this means Apple Intelligence is a failure. It does mean the expected iPhone refresh super-cycle has not arrived on the consensus timeline. Without the AI-driven hardware refresh, the multiple cannot hold at 32x earnings on either the hardware growth thesis or the Services growth thesis.

The buyback machinery is the other support beam under the Apple multiple. Apple bought back approximately $94 billion of stock in fiscal 2025 at average prices of roughly $230. The buyback authorisation outstanding is over $90 billion. The argument runs that aggressive buyback compounds EPS growth, which justifies the multiple. The problem with that logic at 32x earnings is that buyback math turns on the relationship between the FCF yield and the cost of equity. At a 2.5% FCF yield against a 9-10% cost of equity, the buyback is mildly value-destructive on a strict capital allocation framework. Buying stock at a 2.5% yield is no longer accretive when the same capital could be deployed at a 5%+ Treasury yield. The capital allocation calculus has shifted with the rate environment in a way the consensus has been slow to recognise.

The better historical analogue is IBM in 2014-2018, where the company executed roughly $50 billion of buybacks while the underlying business was decelerating. The buyback supported the multiple for a window but did not prevent the eventual de-rating once the operational story caught up. Apple's situation is fundamentally healthier than IBM's was, but the mechanism is the same. Buyback at high multiples of declining-growth earnings buys time; it does not buy a permanent multiple.