Xiaomi unveils its most powerful in-house chip yet, plus a new 3nm processor built for self driving cars

[MayanHan]

Xiaomi used a dedicated chip technology event this week to unveil the Xring O3, its newest flagship smartphone processor, alongside a separate 3 nanometer chip called the Xring D100 built specifically for autonomous driving that will replace Nvidia hardware in Xiaomi's own vehicles starting in 2027. Both chips continue a deliberate strategy the company restarted back in 2021, investing over 21 billion yuan, roughly 3.1 billion dollars, to reduce its long term dependence on outside chip suppliers like Qualcomm and Nvidia.

The Xring O3 is built on TSMC's advanced 3 nanometer process and packs 24 billion transistors across a 10 core CPU configuration paired with a 16 core GPU, and Xiaomi claims it delivers roughly 25 percent higher CPU performance than Qualcomm's Snapdragon 8 Elite Gen 5 while using about 25 to 26 percent less power. Early benchmark results back up that claim at least on paper, with an AnTuTu score above 5.2 million and a Geekbench multi core result of 15,221, both currently standing as class leading figures for a mobile processor. The chip is also reportedly the first mobile processor to support LPDDR6 memory, delivering meaningfully higher bandwidth than the previous generation. It'll debut in the Xiaomi 18 Fold launching in September, alongside the Xiaomi Pad 9 Pro Max, though Xiaomi is reportedly targeting a fairly modest 200,000 to 300,000 units for that specific device, treating it more as a real world test bed for the new silicon than a mass market volume play just yet.

The Xring D100 targets an entirely different application. Also built on a 3 nanometer process, it pairs a 20 core CPU with a 16 core NPU and supports up to 160 gigabytes of unified memory, enough according to Xiaomi to run a 200 billion parameter AI model directly on board a vehicle without needing any cloud connection for the core driving intelligence. The chip is described as China's first self developed 3 nanometer chip built specifically for intelligent driving applications, and it's set to replace the Nvidia Drive hardware currently powering Xiaomi's existing SU7 and YU7 vehicles once it enters commercial deployment next year.

Context matters a fair bit here though. Xiaomi's smartphone shipments actually fell from 84 million units in the first half of 2025 down to 65 million in the same period this year, and the broader global smartphone market is projected to shrink by roughly 14 percent industry wide in 2026. Some analysts quoted alongside this announcement remain skeptical about the near term commercial impact of Xiaomi's custom silicon push specifically, even as the underlying strategic logic, reducing dependence on Qualcomm and Nvidia while gaining tighter vertical control over hardware roadmap, cost structure, and supply chain, remains sound regardless of how quickly it actually pays off in real unit volume

Xiaomi Steps Up Chip Push With New Smartphone, AI and Self-Driving Processorswww.caixinglobal.com

[Loki Daemon]

The modest 200,000 to 300,000 unit target for the O3 powered foldable is clearly the smart, mature move here compared to betting everything on a huge, risky volume launch straight out of the gate. Treating your very first flagship in house silicon generation as a genuine test bed rather than the main event lets you actually catch and fix real world problems before scaling production much more aggressively later

[Hannah56]

China's broader self sufficiency push in semiconductors keeps quietly accelerating on multiple fronts simultaneously, and Xiaomi doing this specific combination of consumer chips plus automotive silicon in the very same announcement is a pretty clean, concrete illustration of just how far that particular strategic push has already come in a fairly short window of time

[Urban Phoebe]

The declining overall smartphone shipment numbers add some notably useful context that's easy to miss if you only focus on the shiny new chip specs themselves. Xiaomi is making an aggressive, expensive long term semiconductor bet at the exact same time its actual core phone business is contracting meaningfully year over year, which is either impressively disciplined, patient long term strategic thinking or a genuinely risky overextension depending entirely on how the next few years actually play out