The strongest argument against the QNX thesis lives in Beijing. Here’s how real it is — and why it’s slower than the headlines.
By The 🍌🐀 (The Banana Rat)
Scope & disclosure. A two-sided editorial assessment of the China risk to BlackBerry’s QNX, not investment advice. As-of date: 2026-07-01. Factual claims are sourced in the endnotes; inferences are labeled as thesis. Conflict-of-interest disclosure: the author holds a position in BlackBerry (BB), so treat this bear case as one written by a bull deliberately arguing against himself. Do your own research.
Part of the 🍌🐀 Physical AI & Edge Compute field guide → the hub for the whole thesis.
Most of what the 🍌🐀 has written about QNX is bullish, so let’s do the honest thing and build the strongest case against. It isn’t NVIDIA muscling in, and it isn’t AMD. The single biggest risk to the QNX thesis is a ~34-million-vehicle market that would rather not depend on foreign software at all — and that has the engineering muscle to do something about it. If you’re long this story, this is the argument that should keep you up at night. So let’s steelman it properly, then test how much of it survives contact with the record.
The bear case, made as strong as it deserves
China builds more cars than anyone — 34.5 million in 2025, nearly half of them electric [1], and the software-defined vehicle is a national priority. Three real vectors are pushing at QNX’s position:
- Automakers writing their own OS. NIO unveiled SkyOS in 2023, and its structure is pointed straight at QNX: a SkyOS-M microkernel that industry reports describe as replacing QNX kernel services, and a SkyOS-L real-time layer replacing the AUTOSAR middleware. SkyOS-L reached mass production on NIO’s NT2.0 platform; full SkyOS is slated for NT3.0 [2].
- A domestic microkernel with a certified rival’s pedigree. Huawei’s HarmonyOS/HongMeng microkernel has been ISO 26262 ASIL-D certified since 2020 and Common Criteria EAL 6+ since 2023, and Huawei claims inter-process throughput several times QNX’s [3]. That is not a toy — it’s a credentialed competitor.
- An open-source national effort. The China Association of Automobile Manufacturers backed an open-source automotive microkernel (iSOFT’s EasyAda, later EasyXMen) under a domestic license, with FAW, Changan, Geely and Li Auto among its backers; by June 2026 iSOFT became the first Chinese firm to contribute AUTOSAR baseline code [4][5]. The stated ambition, in the industry-report framing, was to gradually replace QNX.
Stack that against a self-sufficiency climate (China’s “Xinchuang” localization drive aims to swap out foreign software across state systems) — and the bear thesis writes itself: QNX’s largest opportunity pool is also where it is least welcome, and the locals are coming for the kernel.
That’s the case. It’s real. Now here’s why it’s slower and narrower than the headline version.
Rebuttal 1: no rule actually forces QNX out of the safety layer
The scary version of this story assumes a government mandate. There isn’t one — not for this. China’s flagship localization policy (the “Xinchuang” drive, SASAC’s Document 79) targets office, ERP, operating-system and IT infrastructure in state organizations; its documented scope does not name automotive or embedded safety-critical software [6]. A December 2024 procurement rule grants a domestic price preference, but it’s general, not an auto-safety ban [6]. Anyone telling you Beijing has ordered carmakers off QNX is overstating the record. The pressure is climate and incentive, not a decree.
Rebuttal 2: most “domestic OS” wins are in a room QNX was already leaving
Look closely at what the domestic stacks actually replace. Xiaomi’s HyperOS is Android-based and runs the cockpit; BYD, ECARX (Geely) and Zeekr’s visible efforts are cockpit/infotainment platforms on Qualcomm silicon [5]. That’s the dashboard — the infotainment domain QNX has long been happy to cede to Android and Linux while it keeps the part that matters: the certified safety and control slice. The battle you read about in headlines is mostly being fought in the room QNX was already walking out of. Displacing QNX in the ASIL-D safety domain is a different, far harder job — and outside Huawei’s HongMeng, no domestic OS has yet shown an independent ASIL-D certificate paired with a production safety-domain track record [3]. Even NIO’s full QNX-kernel replacement is still forward-looking, not shipped [2]. (Why that final slice is so hard to take is its own story: The Moat Is Made of Paperwork.)
Rebuttal 3: QNX is still winning in China
Erosion should show up as lost deals. It isn’t — yet. QNX rides the dominant Chinese cockpit silicon (SemiDrive’s X9 plus QNX powers 50-plus mass-production models across SAIC, Chery, Changan, FAW, GAC and more), it is the certified safety foundation of NVIDIA’s DriveOS that Chinese Thor customers build on, and it kept landing fresh business into 2026 — a mid-eight-figure China design win with a global tier-one for ADAS (reported on the Q2 FY2026 call), the Leapmotor D19 premium SUV (production from April 2026), and a next-generation tie-up with SemiDrive’s X10 SoC [7][8]. BlackBerry’s raised FY2027 QNX guidance ($295–312M) is not what revenue erosion looks like [9]. Tellingly, even the most cautious desk on the stock (RBC, which raised its target to $9 in June 2026 but held a Sector Perform rating) rests its caution on GEM revenue transparency, not China substitution [10]. The people paid to be pessimistic aren’t pointing here.
Rebuttal 4: the plot twist — China’s safety crackdown is a tailwind
Here’s the part the bear case has to swallow. After a fatal crash involving a Xiaomi SU7’s driver-assistance system in April 2025, China’s regulators did an about-face on the move-fast era: they banned “autonomous driving” marketing claims, capped consumer systems at L2 with mandatory driver monitoring, and required regulatory approval before over-the-air ADAS updates [11]. The whole market lurched toward liability-defensible, certified, safety-critical software — which is precisely what QNX sells. BlackBerry’s CEO said as much on the record: China’s demand has shifted toward safety-critical capability, and QNX is positioned for it [11]. The reflex the 🍌🐀 keeps talking about, the certified layer that overrides the AI when it’s wrong, is exactly what Beijing now wants more of, not less.
The honest read
Put it together and the China risk is structurally real but concentrated in the wrong layer for the bear thesis to win quickly. The domestic stacks are advancing fastest in cockpit and gateway, the commoditizing tiers, while the ASIL-D safety kernel, where liability and certification live, is exactly where QNX’s pedigree, install base and the new regulatory mood protect it. This isn’t “China can’t”; Huawei clearly can, and NIO is trying. It’s that the substitution is a multi-year grind at the hardest layer, not the imminent wipeout the headlines imply.
Actionable Takeaway: treat China as the thesis’s biggest swing factor, and watch three specific tells rather than vibes. Confirmation of the bear case: a domestic microkernel earning an independent TÜV ASIL-D certificate and a named production win in the safety/ADAS domain (watch NIO SkyOS NT3.0, Huawei HongMeng moving from cockpit into vehicle-control, and Li Auto’s in-house OS); QNX China royalties visibly shrinking in disclosures; or Xinchuang being formally extended to vehicle safety software. Confirmation of the bull case: continued named QNX China safety-domain wins, QNX guidance holding or rising through FY2027, and the domestic microkernels stalling at the cockpit layer. Right now the scoreboard reads bull — but this is the one row of the thesis worth re-checking every quarter.
The full BlackBerry/QNX thesis (the growth models, the certification moat, and the honest “10× bigger” test) is in the pillar: QNX: The Quiet Operating System Powering the AI Age.
The 🍌🐀 has spoken. 🍌🐀
Sources
[1] China 2025 vehicle output: production 34.531M (+10.4%), sales 34.4M (+9.4%); NEV production 16.63M / sales 16.49M (+28%), ≈47.9% of all vehicle sales; 17th straight year as the world’s largest market. CAAM via gov.cn and CnEVPost, Jan 2026. http://english.scio.gov.cn/in-depth/2026-01/15/content_118280571.html
[2] NIO SkyOS, unveiled Sept 21 2023: “1+4+N” architecture with SkyOS-H hypervisor, SkyOS-M microkernel (industry reports: replaces QNX kernel services), SkyOS-L real-time layer (replaces AUTOSAR middleware); SkyOS-L mass-produced on NT2.0, full SkyOS planned for NT3.0. Full QNX-kernel replacement is forward-looking. CnEVPost, Sept 21 2023; ResearchInChina via GlobeNewswire, Jun 20 2024. https://cnevpost.com/2023/09/21/nio-unveils-skyos/
[3] Huawei HarmonyOS/HongMeng microkernel: ISO 26262 ASIL-D certified (Mar 2020), Common Criteria EAL 6+ (Aug 2023); Huawei iDVP automotive platform (AOS/HOS/VOS). HongMeng in cars is predominantly cockpit/HMI; evidence it replaces QNX in the safety/vehicle-control domain specifically is not established. Huawei/HongMeng Kernel documentation; AutoTech News, 2024–2025.
[4] CAAM open-source automotive microkernel plan (Feb 2023), led by i-Soft (iSOFT): EasyAda microkernel under the Mulan Public License; backers include FAW, Changan, Geely, Li Auto; roadmap to mass-production verification 2025. The “gradually replace QNX by 2025” wording traces to a ResearchInChina report, not a CAAM primary statement. Gasgoo, Feb 20 2023; AutoTech News, 2024. https://autonews.gasgoo.com/icv/70022312.html
[5] iSOFT became the first Chinese company to contribute AUTOSAR CAPI baseline code (Jun 15 2026) and markets EasyXMen as an open-source mass-production automotive OS; domestic infotainment stacks (Xiaomi HyperOS on Android; BYD/ECARX/Zeekr cockpits on Qualcomm) target the cockpit domain, not QNX’s safety-RTOS niche. Gasgoo, Jun 15 2026. https://autonews.gasgoo.com/articles/news/isoft-infrastructure-software-becomes-first-chinese-company-to-contribute-autosar-capi-baseline-2066443629424656385
[6] China “Xinchuang” localization / SASAC Document 79 (Sept 2022): mandates replacement of foreign software across state organizations by 2027, but documented scope is office/ERP/OS/IT infrastructure: it does not name automotive or embedded safety-critical software; a Dec 2024 government-procurement draft grants ~20% price preference to domestic products (general, not auto-safety-specific). ginterfaces summary; WSJ “Delete America,” Mar 7 2024; KWM/Lexology, 2025.
[7] QNX China footprint: SemiDrive X9 cockpit SoC + QNX powers 50+ mass-production models (SAIC, Chery, Changan, FAW, GAC, Dongfeng and others); Dongfeng and Geely named as QNX-trusted OEMs; QNX + SemiDrive X10 next-gen AI-cockpit SoC (mass production 2026). QNX China blog “Intelligent Driving Future in China,” 2026; TMTPost, 2025. https://qnx.software/en/blog/2026/intelligent-driving-future-in-china
[8] Leapmotor selects QNX for the D19 premium electric SUV, production from April 2026; NVIDIA DRIVE Thor Chinese customers (Li Auto, Zeekr, BYD, GAC Hyper, XPeng) build on DriveOS, whose certified-safety foundation is QNX OS for Safety 8 (integrated at DRIVE AGX Thor GA, Aug 2025, ISO 26262 ASIL-D). Per-OEM disclosure of QNX (vs. Linux) as the safety layer is not public. QNX press release, 2026; NVIDIA/StockTitan, 2025. https://qnx.software/en/press-release/2026/leapmotor-selects-qnx-for-d19-premium-electric-suv-production-starts-april-2026
[9] BlackBerry Q2 FY2026 (Sept 2025): QNX revenue $63.1M (+15% YoY), “mid-eight-figure design win in the Chinese market with a leading global tier-one supplier” for ADAS. Q1 FY2027 (Jun 25 2026): total revenue $152.9M (+26%), QNX FY27 guidance raised to $295–312M; first cash-positive fiscal Q1 in nine years. Investing.com transcript, Sept 2025; StockTitan, Jun 2026.
[10] RBC Capital Markets (Paul Treiber): after BlackBerry’s Q1 FY2027 beat, RBC raised its price target to $9 (from $4.50) on Jun 26 2026 while maintaining Sector Perform; the flagged caution remains GEM revenue/backlog-conversion transparency, not China domestic-OS substitution. MarketScreener / StocksToTrade, Jun 2026; Finimize, Jun 2026. https://finimize.com/content/blackberrys-rally-needs-proof-in-its-gem-business
[11] After the fatal Xiaomi SU7 ADAS crash (Apr 1 2025), China’s MIIT/SAMR banned “smart/autonomous driving” ad claims, capped consumer systems at L2 with driver monitoring, and required regulatory approval before OTA ADAS updates; MIIT’s 2025 Automotive Standardization Work Plan advances mandatory safety standards. BlackBerry (CEO John Giamatteo, Q2 FY26 call) frames the resulting shift toward safety-critical software as a QNX tailwind. BusinessToday, Apr 17 2025; Repairer Driven News, Jul 28 2025; Investing.com transcript, Sept 2025.
0 comments