Why Xiaomi built the Xring O3 during a profit slump

Xiaomi put the Xring O3 into mass production six days after reporting its third straight quarter of falling profit.

The chip’s own numbers point to supply security and differentiation rather than a cheaper bill of materials.

Xiaomi unveiled the Xring O3 on Monday and said the 3nm processor has already entered mass production. It is a bigger and more expensive chip than the one it replaces, and Xiaomi plans to build very few of them. One source told Reutersthat shipments are targeted at 200,000 to 300,000 units. The company sold 65 million handsets in the first half of this year, based on Visible Alpha data from S&P Global.

The spending sits against three consecutive quarters of falling profit. Adjusted net profit, which strips out investment gains and share-based compensation, dropped 42.6% year-on-year in the second quarter to 6.2 billion yuan, Reutersreported. First-half R&D went up 25.6% to 18.2 billion yuan, with AI-related work accounting for close to 30% of it.

The Xring O3 is a bigger chip, not a cheaper one

IT Home reported that the O3 carries 24 billion transistors on a 133mm² die. Xiaomi’s own figures for the Xring O1, recorded on Baidu Baike, put that chip at 19 billion transistors on 109mm². The transistor count is up 26%. The die is up roughly 22%. A report by South China Morning Post put the density gain at 5%, and according to the two people familiar with the matter who spoke to Reuters, both generations are built on TSMC’s 3nm process.

That combination tells you where the performance came from. Xiaomi did not move to a denser node between the O1 and the O3. It bought the gains largely by making the chip physically bigger, which on a leading-edge process means fewer dies per wafer, more exposure to defect rates, and a higher unit cost.

Then there is the volume. One source told Reuters that shipments of the O3 are targeted at 200,000 to 300,000 units. Xiaomi told analysts last week that devices running the O1, counting smartphones, tablets and watches, had passed one million units cumulatively since launch. Sources told Reuters that roughly 150,000 of those were smartphones. Xiaomi sold 65 million handsets in the first half of 2026, on Visible Alpha data from S&P Global cited by Reuters.

So why is Xiaomi still building the Xring O3?

The answer is not cost-focused, at least not yet. Buying a flagship chip from Qualcomm eats more than 20% of what a Chinese premium phone sells for, according to a Shanghai Observer analysis cited on Xiaomi’s Baidu Baike entry. Making your own removes that cost, so the savings are real. They are just very small next to what Xiaomi has spent.

A source told Reuters the Xring programme has cost more than 20 billion yuan (US$3 billion) so far. Three hundred thousand phones will not come close to covering that. Xiaomi only gets its money back if the chip ends up in a large part of its premium range, and if the same research pays for more than phones. That is what the other two chips are for.

The Xring O100 is a 6nm accelerator built to run Xiaomi’s MiMo model on the device. The 3nm Xring D100 is for autonomous driving, and Xiaomi says development is finished. Reuters sources said TSMC has been contracted to build both. Phones, tablets, watches and cars can share one engineering bill.

There is a second reason, and it may matter more. Every vendor buying a Qualcomm chip gets the same AI hardware as its rivals, because it is the same chip. Owning the design lets Xiaomi build the AI block around its own models instead of waiting for Qualcomm’s roadmap to get there. Ivan Lam, senior analyst at Counterpoint Research, told SCMP that Xiaomi’s progress puts it in the top tier of China’s self-developed mobile SoCs, and that the investment has returned nothing financially so far while the long-term commitment looks sound.

One dependency traded for another

Xiaomi’s stated goal of reducing reliance on external chip suppliers holds for Qualcomm and MediaTek. It does not hold for TSMC. Xiaomi designs the Xring line; it does not fabricate it, and Reuters sources place all three new chips at the same foundry on 3nm and 6nm. Designing your own silicon on a Taiwanese leading-edge process substitutes a commercial dependency that can be renegotiated for a geopolitical one that cannot.

The contrast with Huawei makes the point. Huawei is preparing its Kirin 2026 processor with a new LogicFolding architecture for its Mate handsets, SCMP reported, and it builds that way because leading-edge foundry capacity is close to it. Xiaomi still has that access. Everything it has spent on Xring depends on keeping it, and that is a decision made in Washington rather than in Beijing or Taipei.

Jiemian News, cited by TrendForce, reported that the O3 debuts on the Xiaomi 18 Fold in September and will also appear in the Xiaomi Pad 9 Max. Xiaomi has not indicated the chip will go beyond those, which leaves the mid-range volume that carries the company across Southeast Asia running on the same Qualcomm and MediaTek silicon it ran on last year.

The immediate test is a hard one. Huawei shipped 1.6 million foldable phones in China in the second quarter for a 68% share, followed by Honor at 13.7%, on Smart Analytics Global figures cited by Reuters. Xiaomi is entering that segment with a chip it has never shipped at volume, in a year when IDC expects global smartphone shipments to fall 14%.

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.

Tech Wire Asia is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.