Everything About AI
NextFin News - John Ternus's first earnings call as Apple's chief executive delivered a clean beat on the numbers that matter most - the iPhone - while quietly revealing the company's first real plan to charge for Siri AI, the biggest upgrade to the assistant in its history. Apple reported fiscal fourth-quarter earnings per share of $2.10, up 13.5% year over year and ahead of the $1.99 consensus, on revenue of $114.48 billion, an 11.7% increase that set a September-quarter record and topped the $113.38 billion estimate. iPhone revenue jumped 17.5% to $57.58 billion, also a September-quarter record, and Services reached an all-time high of $31.57 billion. Yet the stock wavered, because management's guidance pointed the other way: gross margin for the December quarter is expected to fall to 45%-46% from 47.9%, and Services growth is guided to 8%-9%, below the roughly 10.7% the market had penciled in.
The tension is the story. Ternus, who took over from Tim Cook on September 1 after nearly 15 years, used his debut call to frame Apple's artificial-intelligence push not as a cost center but as a future revenue line. Server-side Siri AI queries carry real inference costs, he and CFO Kevan Parekh said, and Apple will let users buy additional capacity once usage limits kick in - the clearest signal yet that Apple Intelligence is moving from a free iPhone feature toward a paid tier of the services bundle. The question for investors is whether Siri AI can become the next leg of Apple's $31.6 billion Services engine before margin pressure and slower services growth cap the stock's rerating.
The Beat, the Guidance, and Why the Market Shrugged
On the surface, this was the kind of quarter that usually sends Apple shares higher. Earnings beat by more than 5%, revenue cleared estimates by 1.3%, and the installed base of active devices exceeded 2.5 billion - an all-time high across major product categories and geographies. iPhone strength was broad-based: management highlighted records in emerging markets including India and Southeast Asia, and the September quarter held up despite supply constraints and foreign-exchange headwinds. The Mac business surged 23.7% to $10.79 billion on the back of the MacBook Neo and MacBook Pro, setting a September-quarter record for customers new to the platform. Products as a whole rose 12.5% to $82.90 billion, accounting for 72.4% of total revenue.
But the market trades the next quarter, not the last one, and the December outlook landed soft on the two metrics that carry the most weight with this shareholder base. Gross margin is guided to 45%-46%, a drop of roughly 200-300 basis points from the 47.9% printed for the quarter ended September 26. Services growth is guided to 8%-9%, a deceleration from the 9.8% posted in Q4 and well below the double-digit pace investors had come to expect from the segment that justifies Apple's premium valuation. For a stock trading at nearly 38 times trailing earnings, a margin squeeze plus a services slowdown is exactly the combination that turns a beat into a "sell the news" session. Shares finished the reporting day down about 1.8% at $338.32, slipped roughly 1% in after-hours trading, then drifted near flat as the call unfolded.
The margin story has a concrete cause. Rising memory chip costs have pushed component prices higher across the industry, and Apple's own pricing power - the same force that lifted iPhone revenue 17.5% - is now working in reverse on the cost side. Management said it has received most of the refunds tied to earlier tariff overpayments, which added about one percentage point to gross margin and roughly $0.05 to earnings per share for the quarter. Strip that one-time benefit out and the underlying margin trajectory looks softer still. That is the cyclical leg of the setup: component inflation is mean-reverting by nature, and once memory prices roll over, the margin bridge should rebuild.
Siri AI's Biggest Upgrade Becomes a Monetization Question
The more consequential development, and the one that will define Ternus's tenure, sits in the artificial-intelligence layer. The redesigned Siri - the biggest overhaul of the assistant since its 2011 debut - now works like a ChatGPT-style chatbot: it holds conversations, drafts emails, searches through texts and photos, and understands what is on a user's screen. It is the centerpiece of Apple's new home-device push, running in beta on the upgraded Apple TV 4K and HomeView, where it is designed to act as the control layer for the living room.
Until this call, the market's working assumption was that Apple Intelligence would follow the iPhone playbook: a free feature that drives hardware upgrades and keeps users inside the ecosystem. Ternus explicitly walked that back. "There's going to be some usage limits on features that rely on server-side models, which include Siri AI and intelligent photo editing, some other things like that," he said on the call. Parekh added that server-side Siri AI queries carry inference costs, while on-device processing has "no real cost associated with that," and described Apple as being in the "very, very early stages" of sizing the cost mix and any additional compute requirements.
"There's going to be some usage limits on features that rely on server-side models, which include Siri AI and intelligent photo editing, some other things like that."
The architecture of the answer matters. Apple is not building a standalone ChatGPT-style subscription. Instead, expanded access to server-side Siri AI and other compute-heavy Apple Intelligence features will be tied to iCloud+ plans, which start at $0.99 a month for 50GB and $2.99 for 200GB. That is a deliberately Apple-shaped monetization path: it bundles AI capacity into the storage tier users already pay for, avoids the awkward question of whether anyone will pay $20 a month for Siri, and turns every heavy AI user into an iCloud+ upgrade candidate. With more than 1.5 billion paid subscriptions already in the base and 2.5 billion active devices, even a low-single-digit attach rate on a few dollars of incremental monthly revenue compounds into a meaningful services line item.
There is a second-order point most coverage is missing. The usage-limit-and-upgrade model does more than monetize AI - it reveals which features are expensive enough to ration, and therefore which ones users actually value. Every time a user hits a server-side limit and chooses to pay rather than walk away, Apple learns the price elasticity of a specific AI capability. That is proprietary demand data no competitor can see at scale, and it gives Apple a feedback loop for pricing that pure-software AI labs, still burning cash on free tiers, do not have. The assistant becomes both a product and a market-research engine.
But the competitive bar has moved while Apple was rebuilding Siri. The redesigned assistant can converse, draft, and search a user's device - capabilities that were impressive when Apple scoped the project, but are now table stakes. Rival agents can book travel, fill out online forms, cancel appointments, and file customer-service complaints: the mundane but necessary tasks that define the "personal secretary" use case most people actually want from AI. Apple's Siri AI understands your screen; the question is whether understanding is enough when competitors are already acting.
Cyclical Margin Pressure Meets a Structural Services Test
The cyclical-versus-structural call here has two separate legs, and conflating them produces the wrong investment conclusion.
On margins, the pressure is cyclical. Memory and logic component pricing is a commodity cycle with a well-documented mean-reversion pattern: past upturns in the memory market have consistently been followed by price rollovers as new fabrication capacity comes online, and Apple's gross margin has historically rebuilt once component costs normalize and mix shifts back toward higher-margin Services. The current 45%-46% guide reflects a cost spike, not a permanent step-down in pricing power, and the one-time tariff-refund benefit masking the true baseline will lapse regardless. Evidence floor met: a short-term supply-driven cost shock, a demonstrated historical recovery pattern, and a mean-reverting driver. The margin leg should revert.
On Services, the question is structural, and it is the harder one. The 8%-9% growth guide against a market expectation near 10.7% is not a component-cost story - it is a question of whether the installed base can keep generating double-digit high-margin revenue as hardware upgrade cycles lengthen and as AI features that were once paid (cloud storage, advanced photo tools, writing assistance) migrate into free or bundled tiers. If Siri AI monetization via iCloud+ merely cannibalizes existing subscription revenue rather than adding incremental spend, Services growth settles into a mid-single-digit utility rate and the nearly 38-times earnings multiple has no fundamental anchor. Evidence floor for a structural claim: a permanent shift in how value is captured (bundled AI capacity instead of standalone subscriptions), a business-model change that does not self-correct, and a valuation multiple that depends on the old growth rate persisting. That case is not yet proven - it is the hypothesis this earnings call put to the test.
The structural bull case runs through China. Apple Intelligence is scheduled to reach mainland China in December with iOS 27.3, following a regulatory delay that kept the platform out of Apple's second-largest market for two years. Developer testing begins November 4, and the first wave covers the existing Apple Intelligence toolset - Live Translation, Writing Tools, Visual Intelligence, photo search - with Alibaba's Qwen model powering some Siri, writing, and visual requests. Notably, the next-generation Siri AI announced at WWDC 2026 is excluded from this first China release, a reminder that the deepest AI features remain geographically gated. Still, opening Apple Intelligence to hundreds of millions of Chinese iPhone users - on iPhone 16 and later, iPhone 15 Pro and Pro Max, iPhone Air, and the foldable iPhone Duo - is a latent services catalyst that the December-quarter guide does not yet reflect.
The Counter-Thesis: Apple Is Pricing AI Like a Feature, Not a Platform
The strongest argument against the Siri-monetization thesis is that Apple is solving the wrong problem. AI labs and big-tech rivals are racing toward agentic assistants that act across apps and the open web - booking, transacting, completing workflows. Apple's answer is a better on-device and screen-aware helper, monetized through a storage bundle. That is a feature strategy, not a platform strategy, and it caps the revenue ceiling: users pay for storage capacity, not for intelligence, so the attach rate and the price point are both anchored to a utility product rather than to the value of the work the AI performs.
The counter-thesis has a quantifiable falsifying signal. If, within two quarters of usage limits going live, fewer than 5% of heavy server-side Siri AI users convert to a higher iCloud+ tier, and if Services growth remains below 9% for three consecutive quarters, then the bundled-AI model is not additive and the structural-services concern is confirmed. Conversely, if iCloud+ attach among AI-heavy users exceeds 15% and Services reaccelerates toward double digits as the China rollout lands, the bundling model is working and the margin dip is simply the cost of buying the next growth leg.
What to Watch: Three Horizons, Three Scenarios
Short term (next quarter): the December print is the near-term gate. Consensus stands at $2.91 earnings per share on $155.8 billion revenue for the holiday quarter. The margin guide of 45%-46% implies the beat, if there is one, comes from volume - the iPhone 18 cycle and the new foldable iPhone Duo, which started sales in fiscal first quarter 2027 at a $1,999 starting price - rather than from profitability. A revenue beat with in-line margins would likely stabilize the stock; a margin miss below 45% would reopen the multiple-compression trade.
Medium term (2027): watch the iCloud+ attach data and the Services growth rate. The key metric is not total Services dollars but the slope: a reacceleration toward 10%+ growth alongside rising iCloud+ average revenue per user would confirm that AI capacity is accretive, not cannibalizing. Memory pricing is the second medium-term watch item; a rollover in component contract prices would pull gross margin back toward the high 40s and remove the cyclical overhang.
Long term (structural): the Siri AI question resolves on capability, not pricing. If Apple's assistant closes the agentic gap - moving from screen awareness to cross-app action - the installed base of 2.5 billion devices becomes the largest deployed AI surface in the world, and the bundling model suddenly has a very high ceiling. If Siri remains a smarter helper while rivals ship working agents, Apple's AI spend becomes a defensive cost with a capped return.
Base case: margins trough in the December quarter and rebuild through 2027 as memory costs normalize, while Siri AI monetization adds a low-single-digit growth contribution via iCloud+ - enough to hold Services near 9% but not enough to rerate the multiple. Upside case: China's December Apple Intelligence launch converts at scale and heavy AI users upgrade iCloud+ faster than expected, pushing Services back to double-digit growth and justifying a re-rating toward the high end of Apple's historical range. Downside case: the margin guide proves conservative, memory stays tight into mid-2027, and AI attach disappoints, leaving the stock stuck between a nearly 38-times multiple and mid-single-digit services growth.
The market treated Ternus's first call as a hold-the-line quarter, and on the numbers it was. But the real story is not the 11.7% revenue beat - it is that Apple's new CEO just admitted Siri AI costs real money to run, and outlined the first mechanism to make users pay for it. Whether that mechanism funds the next growth leg or merely re-packages existing subscriptions is the question the next two quarters will answer.
Explore more exclusive insights at nextfin.ai.
