Original Apple “wins” over Nvidia

This is illogical. Unless our understanding of “How does AI make money” is wrong from the beginning.

It’s not that Apple is getting stronger, it’s that AI is too expensive

Let’s talk it through first. This time the throne changes hands, Apple has not become stronger, but AI has become more expensive.

The narrative of the market is switching from “who has the largest investment number” to “who can turn investment into profit.”

Just look at a set of comparisons. HSBC estimates that capital expenditures on AI data centers will account for approximately 2.5% of Apple’s forecast sales in 2026. For Meta, Google, Amazon, and Microsoft, the figure is 39%.

Microsoft, Google, Amazon, and Meta each have capital expenditures in the hundreds of billions of dollars. Apple’s AI-related expenditures in 2026 will be approximately US$11 billion to US$13 billion, which is less than a fraction of others.

It’s easy to spend money, but it’s hard to pay back the debt. Alphabet and Amazon’s free cash flow has turned negative, forcing them to increasingly turn to the bond market for financing. Interest rates are still rising, and borrowing costs are rising. S&P directly lowered Meta’s outlook to negative on the grounds that “capital expenditures and free operating cash flow face significant uncertainty.”

Of course, the most intuitive thing is the “collective discount on U.S. technology stocks” on the last trading day of October. Intel fell 28.58% in a single day, the largest single-day decline in more than 40 years; the Philadelphia Semiconductor Index fell 7.2%, the largest decline since June; CoreWeave and Nebius, which provide computing power rental for AI, both fell by more than 30% in October.

The Nasdaq 100 fell 6.7% in October, its worst month since March 2025.

Apple expects free cash flow to grow 38% to $142 billion in fiscal 2026. It has approximately US$146 billion in cash and securities; its stock price rose 27% during the year, ranking first among the seven giants.

On the one hand, it is burning money until it issuance of bonds, and on the other hand, it is piled up in cash. Funds are flowing from the computing power narrative to the cash flow narrative, which is a shift from preference stories to a return to the basic logic of business.

If AI is not heavily stocked, why should it be repriced?

The premium the market gives Apple mainly comes from three levels of logic.

The first level is hedging properties. Among the Big Seven, Apple is the only “quasi-utility” company. Its revenue does not depend on the success or failure of a certain technology cycle, but only on the fact that 2.5 billion devices are still in the hands of users. When everyone is betting on the return cycle of AI, predictability itself has become a scarce commodity. Tony Meadows, head of investment at BRI in the UK, put it bluntly: Apple is less exposed to capital expenditure-intensive models, and this round of repricing reflects confidence in the sustainability of profitability rather than speculation on the prospects of AI.

The second level is indirect realization. Apple is a rare “low-investment beneficiary.” Regardless of whether OpenAI, Google or Alibaba wins in the end, if AI wants to reach users, it must go through Apple’s entrance. It doesn’t race, but it is the track itself.

The third level is the financial structure. The service business gross profit margin is 75.6%, and the product gross profit margin is approximately 40%. The hardware is responsible for bringing people in, and the service is responsible for collecting money. This structure makes Apple’s AI investment naturally “light”. The end-side model runs on the chip that the user has already purchased, and the computing power cost has already been paid by the hardware.

But don’t rush to sing too much, Apple also has many problems of its own.

The first is that the valuation is not cheap. TTM trades at a price-to-earnings ratio of 30 to 36 times, and nearly 36 times expected earnings over the next 12 months, which is at its highest level since 2008 and in the 99th percentile of the historical valuation distribution. It has become the only stock among the Big Seven with a negative implied return in the next 12 months. At least six analysts have downgraded their ratings since the financial report at the end of July. The buy recommendation ratio is less than 60%, while Microsoft, Amazon, and Meta all have more than 90%.

Siri no longer bounces, but it’s not easy to roll out in the global market. The new Siri AI was launched with iOS 27 in beta form on September 14, 2026, supporting 16 languages, including Simplified Chinese. However, the beta version of some AI functions will not be opened to users in mainland China until November 4; however, the complete experience with the ability to autonomously call across applications will still have to wait until next month.

Apple’s much-hyped “visual intelligence” is driven by Google Gemini. Apple pays about US$1 billion per year to license a 1.2 trillion parameter model. Complex calculations are hosted on Google Cloud, and the Chinese version is switched to Alibaba Qianwen.

Some AI capabilities are essentially “outsourced.” In January 2026, Google and Apple officially announced a multi-year strategic cooperation. The next generation of Apple’s basic model is built directly on Gemini’s model architecture and cloud infrastructure, while Apple’s self-developed trillion-parameter model is still on the way.

So the core judgment is just one sentence: the market buys Apple not because it does the best AI, but because it doesn’t have to spend money on AI. These two things have completely different valuation implications.

Can Apple find a way to make money independent of AI?

Yes, and it has been done three times.

This style of play has a name, called “toll station mode”: no roads are built, only tolls are set up.

During the cloud computing boom, Apple did not build its own. Instead, it directly bought Amazon AWS and resold it to users to make a profit on the price difference. In the search era, it did not develop itself, but signed a pre-installation agreement with Google, earning tens of billions of dollars every year. Now that AI is here, it outsources models to OpenAI, Google, and Alibaba, and guards the entrance itself.

Three times without missing a beat.

The forecast given by the agency is not ugly. FY2028 revenue will be US$534.1 billion, and the service business will have a CAGR of 12.4%. It is the fastest growing part of the revenue structure and has the best profit quality.

But the ceiling is equally clear.

First, this model is strongly bound to the iPhone replacement cycle. If mobile phones cannot be sold, there will be no cars passing through the toll booths. Second, antitrust pressure on the App Store’s commission is ongoing, and European and American supervision has no intention of letting up.

So this path can work, but it is “steady-state rent collection”, not a “high-growth story.”

what does that mean? This means that if you look at Apple through the valuation framework of technology stocks, you will always feel that it is “not sexy enough”; if you look at it through the framework of consumer stocks, it is one of the best cash flow assets in the world. This round of market repricing is essentially a switching of valuation coordinates.

The global market is strong, but the Chinese market is weak

Apple has reached the top position in the world, but it has seen the most contradictory readings in China. Its revenue growth rate is the first in the world, but its market share in China is relatively weak.

In the third quarter of this year, Apple shipped 11.6 million units in China, a year-on-year increase of 15%, and a share of 20%. Huawei’s sales reached 12.4 million units in the third quarter, an increase of 18%, and its share reached 21%. The two companies took a combined share of 41%, about 10 percentage points higher than the same period last year. The market is concentrating towards the duopoly, and Apple has changed from the only king to one of the top two.

According to the authoritative organization IDC, iPhone shipments in the quarter were approximately 10.33 million units, a year-on-year decrease of 4.3%, but the share increased from 15.8% to 17.8%. The volume is decreasing but the share is increasing, indicating that the entire market is shrinking faster.

Moreover, Apple’s price range has begun to lose ground. In the mainstream high-end segment of 5,000 to 6,000 yuan, Huawei has overtaken Apple to take the first place. Huawei’s share of sales above 8,000 yuan has exceeded 30% for the first time, reaching a record high. Apple still holds an absolute advantage of 73.5% in the market above 6,000 yuan and above 8,000 yuan, but it is losing ground.

Revenue is rising and share is losing. Apple can only use price to exchange volume.

During the 618 period, the iPhone 17 Pro series saw a direct discount of 1,000 yuan, with superimposed subsidies saving up to 2,000 yuan, and the standard version fell below 4,500 yuan for the first time. Huawei responded to the challenge on the same day, and the Mate X6 dropped by 3,000 yuan. By September this year, iPhone 18 Pro/Pro Max each increased by US$100, and the basic model was postponed to next spring. Only two high-end models are on sale in the autumn. The volume has not increased, but the average price has increased.

It’s not that Apple can’t be sold in China, it’s that it’s expensive to sell and hard to keep. In the past, we relied on brand premiums to sell products, but now we have to rely on price cuts and subsidies to increase volume. Volume has returned, and the gold content is lower than two years ago.

Ternus, who took over as CEO, said at the financial report meeting that he was satisfied with the performance of the Chinese market. Growth mainly came from iPhone, followed by Mac. This is true, but the market reaction was that the stock price fell by more than 3% after hours. Investors care about the expected shortfall of US$18.75 billion in the next quarter, not the year-on-year growth of US$17.5 billion.

In China, the difference between 20% and 21% shares is not one percentage point, but who sets the rules. Apple still holds high-end pricing power, but it is no longer the Apple that doesn’t need to explain, cut prices, or rush for time.

寫在最後

According to the actual measurement of Apple Smart, which has already received the beta version, the National Bank Apple Smart is more like a safe test for Apple in the domestic AI mobile phone track. Its end-side native functions focus on practical capabilities such as translation, transcription, and photo processing, while complex creation and in-depth Q&A are filled by Qianwen. However, the AI ​​mobile phone battlefield has been raging in China for several years. Real-time translation, photo elimination, AI summarization, and intelligent creation are no longer new selling points.

Complex creations and in-depth questions and answers are left to Qianwen, and short-term experience can be completed quickly. But there is a long-term worry. When users get used to “AI capabilities = Qianwen”, the value perception of Apple’s own model will be diluted, and this part of the experience is subject to cooperation terms and data compliance, and is not a moat that Apple can completely control. The intelligent agents of domestic manufacturers use their own base + their own scheduling, while Apple’s “own framework + other people’s brains”, the value of the route is different.

Apple still has a chance to catch up with the time window for domestic manufacturers to roll out smart phones, but the window is narrowing.

References:

Apple’s National Bank AI first-hand actual test Source: Zhidongxi

Everything at Apple is becoming more expensive, and it’s all because of AI. Source: Ai Faner

• END –

Related Reading

• World, line up to sell goods to China!2026-11-12
• Ni Runfeng, the “King of Color TV”, passed away. From a workshop technician to the head of a state-owned enterprise, his ups and downs in half a life defined an era of home appliances.2026-11-12
• Original “Fearless” Wang Jianlin regains control of Wanda Business Management, and his net worth has evaporated by 200 billion in ten years2026-11-12
• Original: The middle-class people are crazy about grabbing Archaeopteryx in Sam2026-11-12
• There is no more double 11 in the world, leaving businesses miserable2026-11-11