←News detail

Phoenix TechnologyUpdated Nov 3 14:17

Apple's fiscal 2026 revenue reaches $478.8 billion as AI inference costs push up hardware pricing

AI summary

Apple released results for the fourth quarter and full fiscal year 2026, with full-year revenue reaching $478.8 billion and quarterly revenue of $114.477 billion, up 12% year over year. CEO Ternus said on the call that memory prices are rising at an unprecedented pace, putting direct pressure on hardware pricing, and gross margin guidance for next quarter falls back to 45%–46%. Siri handles more than 2.5 billion requests per day, and cloud inference spending pushed R&D expenses up to $45.765 billion, up 32.5% year over year; Apple is exploring bringing advanced AI capabilities into iCloud+ subscriptions.

Why read this

AI score 60

Apple's earnings call offers first-hand clues on how AI inference costs are passed through to hardware pricing and subscription models, revealing changes in the cost structure.

Topics

Tags

Body · AI translation

Everything at Apple is getting more expensive, and it's all because of AI

Ifanr Download the app for exclusive early access November 03, 2026 12:07:46 From Beijing

Today, Apple released its results for the fourth fiscal quarter and the full 2026 fiscal year. This is also the first report card since John Ternus succeeded Cook as CEO.

Judging by the numbers on paper, the performance can be described as quite ferocious, directly recording Apple's best fiscal year in history.

Quarterly revenue was $114.477 billion, up 12% year over year; net profit was $30.711 billion, up 12% year over year. Full-year total revenue even reached a record $478.8 billion.

Even Greater China, which had been under pressure for consecutive periods, bucked the trend with quarterly revenue growing 21%, thanks to the autumn new-model upgrade wave.

But in the subsequent earnings call, the tone suddenly changed.

Ternus, newly in office, and management were repeatedly pressed at the meeting not only about the impressive sales records, but about "out-of-control costs," "memory price increases," and "capacity constraints."

Right at the critical moment of Double 11 and the year-end shopping season, Apple's profit outlook was somewhat conservative: gross margin guidance for next quarter fell back to 45%–46%, far below this quarter's 47.9%.

Faced with analysts' questions about soaring costs and hardware pricing, Ternus did not evade:

These memory prices are unlike anything we have ever seen, and they are indeed putting direct pressure on our hardware pricing. But when we set prices, we always focus on building the best products and delivering value, and installment and trade-in plans are becoming key factors supporting purchasing power.

Image | Bloomberg

2.5 billion Siri requests a day, and Apple wants users to pay for it themselves

For a long time, Apple walked on two legs and made money very comfortably: on one hand, it earned rich premium margins from hardware; on the other hand, it relied on Apple Music, Apple TV+, iCloud+, and the bundled Apple One to continuously accumulate high-stickiness subscription revenue.

English text is being prepared.

But Siri AI completely changed all of that.

Ternus revealed on the call that Siri now needs to handle more than 2.5 billion user requests per day.

Cross-app scheduling, long-text rewriting, and complex reasoning that on-device small models cannot handle all need to rely on Apple's Private Cloud Compute (private cloud computing cluster).

Every time a user asks Siri to organize meeting minutes, generate images, or retrieve schedules, the cloud servers burn real money in tokens and electricity.

And this "operation and maintenance tax" that did not exist in the past is squeezing Apple's profit margins at an alarming rate.

The financial report shows that Apple's R&D spending for fiscal 2026 soared to $45.765 billion, up 32.5% year over year, more than twice the growth rate of total revenue.

CFO Parekh said bluntly when answering analysts: the surge in R&D expenses was driven squarely up by large-model cloud infrastructure and inference computing spending.

Faced with sky-high bills from billions of calls per day, even Apple, with hundreds of billions in cash, feels the heavy pain.

Therefore, Apple has no choice but to pass this expense on to consumers, and Ternus did not hide this on the earnings call:

Today's iCloud+ is no longer limited to cloud storage. We are exploring integrating more advanced AI capabilities into service subscriptions, including setting reasonable server invocation thresholds and providing expansion plans, which will open up entirely new growth space for the services business.

Later, when we invoke advanced AI capabilities on the iPhone, a prompt will likely pop up on the screen: today's free quota has been used up, please upgrade your iCloud+ plan.

In addition to the endless token fees generated in the cloud, the most direct impact of the AI wave on ordinary consumers is that hardware across the board is getting more expensive.

It is still that narrative we are all too familiar with: global AI data centers' frenzied buying of HBM and high-spec DRAM has severely damaged the consumer chip supply chain.

Ternus admitted at the meeting that this round of memory cost increases was "unprecedented." To lock in supply, Apple's capital and manufacturing procurement spending in a single quarter reached as high as $33 billion.

This heavy upstream hardware cost was quickly pasted in the form of price tags in front of every consumer. In the latest product line, Apple has cut prices—or rather raised them—on almost all volume-driving base models:

The starting price of Mac mini soared from 4,499 yuan all the way to 6,999 yuan; iPad Air rose to 5,999 yuan; the starting thresholds for MacBook Air and iPhone 18 Pro were both pushed up to the 10,000-yuan mark (9,999 yuan).

Raising prices alone was not enough; Apple even thought carefully about the pace of new product launches—delaying the cheap, easy-to-sell base models and putting only the most expensive new products on sale.

At this year's September fall event, Apple broke a nearly two-decade-old convention: the volume-driving standard iPhone 18 was delayed, leaving only the high-priced Pro series and the first foldable iPhone Duo for the entire second half of the year.

Goldman Sachs analysts specifically pressed on the reason for the standard model's delay during the earnings call. Ternus was vague about it, but the business logic had long been self-evident—in a year of soaring component prices, Apple must go all out to sell higher-priced, higher-margin products.

We are pleased to have launched the iPhone Duo and 18 Pro in September, while the iPhone 17 and iPhone Air continue to round out the lineup. Of course, we never comment on future product roadmaps.

However, the iPhone Duo, priced as high as 21,999 yuan, ran into a double grilling over quality control and production capacity as soon as it hit the market.

Not long after the first batch of new phones went on sale, consumer complaints about cracked inner screens and black or green lines began appearing online one after another.

Image | 9to5Mac

In an exclusive interview with Reuters, Ternus had to step in personally to put out the fire.

Only a very small percentage of iPhone Duos experienced display failures, and Apple is doing everything it can to replace the entire device free of charge for affected users.

Just as thorny as the cracked screens is the production ramp-up dilemma for the foldable inner display and the complex titanium alloy hinge.

Ternus said bluntly on the earnings call that iPhone Duo supply is currently "in a constrained state, and at this stage it is completely impossible to predict when supply and demand will come into balance."

The same ultra-high-end strategy is also playing out on the Mac: Apple launched its first MacBook Pro with touch functionality (PROLED XDR display), with a starting price set directly at 22,999 yuan.

Faced with the ultra-high pricing brought on by hardware price increases, ordinary consumers' wallets are under unprecedented pressure.

When analysts asked on the call whether consumers could still bear the steep prices, the remedy Ternus offered seemed rather resigned:

Apple will go all in on the Apple Upgrade Program and 24-month interest-free installments, trying to use the "a cup of coffee a day" pitch to offset the sharp pain caused by hardware price increases.

Apple doesn't produce tokens; it's just a porter for large models

Under the cloud of across-the-board pressure on consumer hardware and steadily climbing costs, one business in this earnings report achieved a surprisingly explosive breakout—Mac.

In the fourth fiscal quarter, the Mac business posted quarterly revenue of $10.8 billion, up 24% year over year, far outpacing all other hardware categories.

The volume-driving MacBook Air, the newly launched MacBook Neo, and the regular back-to-school upgrade wave were still, on the surface, the mainstays supporting that $10.8 billion pie.

But beyond the conventional consumer base, the Mac is also quietly seeing another unexpected source of new demand.

According to Reuters and The Information, as the cost of deploying large models spirals out of control, the hefty token bills for renting cloud computing power from Microsoft, Amazon, or Nvidia are forcing many small and medium-sized businesses and developers to turn their gaze local—

and high-spec Macs equipped with ultra-high-bandwidth unified memory are becoming their refuge from cloud computing inflation.

Apple Silicon chips' inherent high-bandwidth unified memory architecture means a Mac Studio or Mac mini configured with 128GB or even 192GB of unified memory, priced at just tens of thousands of yuan, can run open-source large models at the 70B scale directly locally and offline.

This not only eliminates pay-as-you-go cloud token fees but also removes enterprises' concerns about data leaks.

Ternus also confirmed this momentum on the earnings call, noting that the Mac is enjoying its best enterprise purchasing cycle in history, and that it even won a bulk order for 170,000 MacBook Neos in the Philippines for education and government-enterprise use.

Rather than celebrating the "best fiscal year ever," the first report card Ternus delivered at the helm of Apple is more like a pragmatic business showdown in the face of runaway computing bills.

When tokens become so expensive that even Apple feels the pain, the narrative of a free dream-building AI era has officially turned the page.

Author | Zheng Tingxu

Editor | Xiao Qinpeng

"Special statement: The content above (including videos, pictures, and audio if any) is uploaded and posted by users of "Dafeng Hao," a self-media platform under Phoenix Net. This platform merely provides information storage space services.

Notice: The content above (including the videos, pictures and audios if any) is uploaded and posted by the user of Dafeng Hao, which is a social media platform and merely provides information storage space services."

Source

Phoenix Technology