Xiaomi (1810.HK): A $78B Profitable Phones-Cars-AI Conglomerate Down 57%, While a Pre-Revenue AI Lab Trades at $128B

$1810.HK, Xiaomi is a roughly $78 billion Hong Kong-listed conglomerate that makes smartphones, electric cars, in-house chips, and an open-source AI model called MiMo. The stock is down 57 percent from its 2025 peak. To frame the oddity of that drawdown: Zhipu (Z.ai), a pure-play Chinese AI lab with a frontier-class model, recently crossed a $128 billion market capitalisation on less than $120 million of revenue.

Sit with those two numbers. A profitable, diversified hardware-plus-services-plus-AI conglomerate trades at $78 billion. A pre-revenue model lab trades at $128 billion. That gap is the whole story of why Xiaomi is interesting right now, and it is also the reason the factor framework reads it as a Hold rather than a Buy. StockRank 36 of 100. Decent business quality, fair value, and momentum on the floor. This post is an attempt to lay out why all three are true at once.

Why It Is Down: The Memory-Chip Squeeze

The proximate cause of the drawdown is not demand. It is cost. DRAM and NAND memory prices have roughly doubled on AI-datacentre demand, and memory now exceeds half of a smartphone's bill of materials. When the single largest input to your flagship product doubles in price, and you cannot pass all of it through to a price-sensitive consumer, the margin absorbs the difference.

Xiaomi's smartphone gross margin fell to 10 percent. Founder Lei Jun has warned publicly that the squeeze could last two years. There is an irony worth naming here for readers of this catalog: the same AI memory boom that is crushing Xiaomi's phone margin is the boom that is minting fortunes for the upstream substrate and memory-infrastructure names. We wrote about one of them this week, AT&S (ATS.VI), whose customers are funding new factories precisely because the demand for leading-edge memory and substrate is insatiable. Xiaomi sits on the other side of that same trade: a buyer of the component whose price is being bid up by the AI build-out.

The Business Model: Razor and Blades

The reason a 10 percent smartphone gross margin does not break the investment case is that the phone was never where Xiaomi intended to make its money.

Xiaomi sells hardware close to cost. Management has historically committed to capping the net margin on hardware at around 5 percent. The phone is the razor. The blades are the services that run on the device for years after the sale. Internet services, mainly advertising inside the Xiaomi software ecosystem, are roughly 8 percent of revenue at a 76 percent gross margin, and that high-margin stream grows with every new device sold into the installed base.

In Hamilton Helmer's framework, the durable Power here is not the hardware manufacturing, which is a scale-and-cost game any large Chinese OEM can contest. It is the installed-base monetisation: every device shipped near cost enlarges the pool of users from whom the 76-percent-margin services revenue can be harvested for years. The honest assessment is that this is a real but contested moat. Xiaomi's ecosystem lock-in is softer than Apple's, the services attach rate is lower, and the Chinese Android landscape is more competitive than the Western duopoly. We covered how to think about that kind of durability question in our explainer on what an economic moat actually is. The point for Xiaomi is that the services engine is the part of the business that justifies selling the hardware at cost, and it is the part the memory squeeze does not touch.

The Car Nobody Expected the Ford CEO to Daily

The most striking single data point in the bull case is not a financial metric. It is a behaviour. Ford's CEO, Jim Farley, flew a Xiaomi SU7 from Shanghai to Chicago and has continued to use it as his daily driver. His verdict, on the record: “I don't want to give it up.”

The man running one of the largest legacy automakers in the world picked a phone-maker's first serious car as his personal benchmark. That is the kind of signal that does not show up in a spreadsheet but tells you a great deal about the product's competitive position. Xiaomi did not build a credible electric car. It built a car that the competition is studying.

Investing Through the Storm

Rather than retrench into the margin squeeze, Xiaomi is spending into it. Research and development ran at 9 percent of Q1 revenue, up 33 percent year on year, and is guided to a record $5.6 billion for 2026. The forward catalysts management has put on the table:

• A 550,000-car delivery target for 2026. Up 34 percent on the prior year, on the strength of the SU7 ramp.
• The new SU7 production ramp. The model changeover that hurt Q1 volume is also the source of the second-half recovery management is guiding to.
• A planned Europe entry in H2 2027. The first move of the EV business beyond China, and the optionality that a domestic-only narrative does not price.

Spending a record R&D budget into a margin trough is the behaviour of a management team that believes the trough is temporary and the competitive window is now. It is also the behaviour that depresses near-term earnings, which is part of why the Value factor reads only fair rather than cheap.

The Dream Premium Goes to Others

Lei Jun is regularly called China's Elon Musk: phones to cars to in-house silicon, all under one founder-operator with a cult following. The comparison is apt on the breadth of ambition. It breaks down on the multiple.

The market has not granted Lei Jun the Musk-Jobs reality distortion field. Tesla and the frontier AI labs are priced on the dream: a multiple that assumes the founder bends the future to the roadmap. Xiaomi is priced on the numbers: a profitable conglomerate trading at a fair multiple of the earnings it actually reports. The dream premium, the thing that takes a pre-revenue Zhipu to $128 billion, goes to others.

Lei Jun gets called China's Musk: phones to cars to in-house chips. The difference is the market has not granted him the reality distortion field. The dream premium goes to others.

For a value-minded investor, that is not necessarily bad news. A business priced on its numbers rather than its dream is a business where the downside is anchored to something real. The question is whether the absence of a dream premium is a permanent feature of how the market sees a Chinese hardware conglomerate, or a temporary consequence of the momentum drawdown that reverses when the margin trough passes.

Capital Allocation: Real Buybacks, Net Dilution

The capital-return story has a genuine contradiction in it that an honest writeup has to surface rather than bury.

On one hand, the buybacks are real. There is no dividend (and there has not been one since the 2018 IPO), but Xiaomi announced a fresh HK$20 billion (roughly $2.6 billion) repurchase plan, and 2026 repurchases already exceed the entire 2025 total. That is a meaningful commitment of capital to retiring stock at a depressed price.

On the other hand, a 2025 equity placement raised HK$42 billion. The net effect is that the share count rose rather than fell. The buyback, real as it is, was more than offset by the raise. An investor reading only the buyback headline would conclude the float is shrinking. It is not. The placement funded the EV and AI investment programme, which may well be the right strategic choice, but it means the per-share compounding effect of the buyback has not yet materialised. This is the kind of detail that reading buyback disclosures carefully is designed to catch.

The MoatMap Scorecard: Q56 V55 M16, StockRank 36

Here is the Xiaomi MoatMap StockRank:

• Quality: 56/100. Decent. The services engine, the installed base, and the diversification across phones, EVs and chips give a middling-to-good quality reading. Not a fortress, not a cigar butt.
• Value: 55/100. Fair. After the 57 percent drawdown the stock is no longer expensive, but the depressed near-term earnings (memory squeeze plus record R&D plus EV losses) keep it from screening as outright cheap. Fairly priced, not a bargain.
• Momentum: 16/100. On the floor. This is the factor doing the most work in the composite. Down 57 percent from the peak, the market is voting hard against the name.
• Composite StockRank: 36/100. Hold. Decent quality and fair value cannot overcome momentum this weak. The factor framework is explicitly not calling this a buy; it is calling it a name to watch.

This is a genuinely different shape from most of the names in our catalog. It is not the cash-rich derated deep-value setup of the Greater-China names, and it is not the momentum-led AI-infrastructure rerating of AT&S. It is a quality-and-value-balanced business that the market has knocked down on a real but plausibly temporary margin shock. We covered how to read a Hold-rank profile like this in our guide to factor investing.

The Question Worth Sitting With

The EV arm is where the bull and bear cases collide most directly. Xiaomi's electric-vehicle business slipped into a RMB 3.1 billion operating loss after turning profitable in 2025, as the new-SU7 model changeover gutted Q1 volume. Despite that, management reaffirmed a 34 percent increase in 2026 EV deliveries.

Realistic?

The bull read says yes: the Q1 loss is a changeover artefact, the SU7 ramp is real, the product is good enough that the CEO of Ford dailies one, the 550,000-car target is achievable on second-half volume, and the Europe entry in 2027 adds a leg the market is not pricing. In this read, the memory squeeze passes inside two years, the services engine keeps compounding on a growing installed base, and the dream premium eventually arrives once the numbers stop falling. A Hold today becomes a Buy when momentum turns.

The bear read says the 34 percent delivery guidance, set against a quarter that just produced an operating loss, is the kind of target that gets quietly walked back; that the memory squeeze could run longer than two years; that the EV business is burning cash in the most competitive automotive market on earth; that the 2025 placement shows management will dilute shareholders to fund the dream; and that the absence of a dream premium is the market's correct assessment of a hardware conglomerate with thin hardware margins, not a temporary mispricing.

Both reads are defensible, and unlike the Strong-Buy names elsewhere in this catalog, the factor framework is not taking a side. StockRank 36 is the framework saying, in effect, the quality and the value are real but the momentum is too weak to act on yet. Watch it. Let the margin trough and the EV guidance resolve before the framework will upgrade it.

Companion Reading

Xiaomi connects to several threads we have already written:

• AT&S (ATS.VI) for the other side of the memory trade: the AI-driven DRAM and NAND boom that is crushing Xiaomi's phone margin is the same boom funding AT&S's new factories. One name is the buyer of the squeezed component; the other is the supplier benefiting from the squeeze.
• Fast Retailing (9983.T) for the founder-architect question. Lei Jun is to Xiaomi what Tadashi Yanai is to Uniqlo: the visionary whose ambition the market is being asked to underwrite. Both raise the question of how much of the multiple is the system and how much is the person.
• Penguin Solutions (PENG) for the AI-infrastructure layer that explains the memory demand in the first place: the inference memory wall is the demand-side reason DRAM and NAND prices doubled.

The Bottom Line

Xiaomi is the clearest example in our catalog of a business the market is pricing on its numbers while it spends to build a dream the market refuses to pay for. The razor-and-blades services engine is real. The car is good enough that the CEO of Ford will not give his up. The R&D commitment through the storm is real. The buybacks are real. So is the memory squeeze, the EV operating loss, the net dilution from the 2025 placement, and the momentum sitting at 16 out of 100.

The factor framework does not resolve the tension for you. StockRank 36 is a Hold: decent quality, fair value, broken momentum. It is a name to watch through the margin trough and the EV-guidance test rather than a name to act on today. For investors thinking about how a Hold-rank position fits a broader book, our guide to reviewing your portfolio for weak spots is the right framework for sizing a watch-list name where the central catalyst (does the margin trough pass and the EV guidance hold) is still unresolved.

For the full breakdown including the segment economics, the services attach math, the EV delivery cadence, the buyback-versus-placement share-count history, and the memory-cost sensitivity, the Xiaomi Deep Dive is the place to go.

This article is for informational purposes only and is not investment advice. The author may be long names covered on MoatMap.