⚠️ Editorial note: The open source ecosystem in China operates under a distinct institutional framework — characterized by state-led initiatives, intranet-like boundaries, and top-down governance. Readers should be aware that this context differs from the community-driven open source model common in other regions. The term “open source” as used in Chinese media may refer to practices that diverge from the conventional definition.

China Open Source Daily — 2026-09-08

🏗️ Institutional Change — OpenHarmony’s Ex-Vendor Architect Documents the 14-Billion-Device Base Layer and the Vendor-Foundation Separation Pattern

1. 凤凰《AI 研究院》 (Phoenix AI Research Institute, WeChat, published 2026-09-07, by 徐熳 / Xu Man) — “对话深开鸿 CEO 王成录：为什么 76% 国产机器人依赖海外操作系统？”

2. OpenAtom Foundation official press releases (2025-04-17, 2025-07-25, 2025-11 donation, 2026-05-25) — M-Robots OS timeline on openatom.org/journalism

3. Baseline: September 5 briefing — Ren Xudong’s Silicon Time essay recommending Chinese AAIF participation; September 6 briefing — MIIT’s AI SME Startup Support Plan and AtomGit’s national-community designation; September 7 briefing — Huawei’s AAIF entry and Lu Shouqun’s joint-three-foundation lifetime-achievement award

A 凤凰《AI 研究院》(Phoenix AI Research Institute) dialogue published on the WeChat input feed for 2026-09-07 delivers, for the first time in this series’ observation, a vendor-official-agenda-layer reading of the OpenHarmony base layer from the ex-Huawei architect who runs the ecosystem’s second-largest contributor (深开鸿 / Deepin AI Labs, formerly 深圳开鸿数字产业发展有限公司). The article, by Phoenix AI Research Institute’s 徐熳 (Xu Man), is an interview-format institutional placement that pairs with the OpenAtom Foundation’s own official timeline for the M-Robots OS donation — the vendor-foundation separation pattern that prior briefings’ Western-neutral-foundation question documented at the AAIF / KubeCon-China layer, applied here to the OpenHarmony / robot-OS layer.

Four institutionally new claims in the Wang Chenglu dialogue

Claim 1 — OpenHarmony has crossed the 14-billion-device base-layer threshold, and only ~3 billion are Huawei phones. The article’s own text, in Wang Chenglu’s voice: “开源鸿蒙 14 亿台设备里，只有不到 3 亿台是华为手机，11 亿台都不是华为的设备” (Of OpenHarmony’s 14 billion installed devices, fewer than 3 billion are Huawei phones, and 11 billion are non-Huawei devices). The 14 billion figure is the article’s own institutional number, and the 11/3 split is the article’s own reading of what the base layer’s non-Huawei dominance means for Huawei’s own strategy.

From an institutional economics standpoint, this 14-billion-base-layer / 11-billion-non-Huawei pairing is a structural finding: the OpenHarmony ecosystem has crossed an installed-base threshold where the non-vendor device count (11 billion) exceeds the vendor’s own device count (3 billion) by more than three-to-one — a structurally new institutional form that the previous briefings’ AAIF/KubeCon-China observation did not document at the OpenHarmony layer. The article’s own claim about the implication is that these 11 billion non-Huawei devices become a connectivity asset for Huawei’s own phones — Wang Chenglu’s own framing — which is a structurally new placement of a Chinese vendor’s own strategic rationale for running an OpenHarmony base layer that is not owned by the vendor.

Claim 2 — Huawei’s contribution share inside the OpenHarmony community has declined from ~90% to ~70%, and the article’s own diagnostic is a “natural process.” Wang Chenglu’s own words: “华为在开源鸿蒙的贡献占比，最早是 90% 多，现在已经到了 70% 几。这是一个自然的过程” (Huawei’s contribution share inside the OpenHarmony community started above 90% and is now in the low-70s. This is a natural process.) The article adds an institutional-comparison framing the previous briefings’ AAIF/KubeCon-China observation did not document: “任何一个大的社区从开始运营，一定要有一个产业的主导方。就像 Linux 社区，早期主导是 IBM，后来微软和谷歌才加入” (Any large community, at the start of operation, must have an industry-leading party. Like the Linux community, whose early lead was IBM, then Microsoft and Google later joined.)

From an institutional economics standpoint, this IBM-Linux-precedent / Huawei-OpenHarmony-precedent pairing is a structural finding: Wang Chenglu is placing Huawei’s current position in the OpenHarmony community in the same lineage as IBM’s historical position in the Linux community — the article’s own framing — as an industry-leading-party position at a foundation-owned base layer. This is a structurally new vendor-official claim that prior briefings did not document at the OpenHarmony layer. The article’s own framing — 90% → 70% as a “natural process” — is a specific institutional-economics claim: the vendor’s contribution share is expected to decline as the community matures, and the “natural process” framing is a vendor-official institutional reading of what community-maturity means for a vendor-originated base layer.

Claim 3 — 哈勃投资 (Hubble Investment, Huawei’s venture arm) held 20% of 深开鸿 at founding — a structurally new institutional fact about the vendor-foundation separation. The article’s own institutional fact: “深开鸿初创时，哈勃投资占了 20% 的股份。哈勃对所有产业链的投资，很少超过 10%，20% 的占股，是坚定的看好，更是战略的笃定” (At 深开鸿’s founding, Hubble Investment held a 20% stake. Hubble rarely exceeds 10% across its industrial-chain investments; a 20% stake is firm conviction and strategic determination.) And from Ren Zhengfei’s side: “开源鸿蒙不单对各行各业有意义，对华为本身就是一条很宽的护城河” (OpenHarmony is not only meaningful to every industry — for Huawei itself, it is a very wide moat.)

From an institutional economics standpoint, this Hubble-20%-stake / moat-for-Huawei pairing is a structural finding: the vendor-foundation separation that prior briefings’ AAIF/KubeCon-China observation documented as an institutional aspiration is being implemented here with a specific equity structure (Hubble’s 20% stake — an outlier for Hubble’s own <10% rule) and a specific vendor-official strategic rationale (a “wide moat” for the vendor itself, in Ren Zhengfei’s own framing). The institutional-form object documented here — a vendor-owned stake in a vendor-separating foundation-backed entity, justified as a moat rather than a dividend — is a structurally new institutional form at the vendor-foundation separation layer that the previous briefings did not document.

Claim 4 — 深开鸿’s 2028 target is to become “a pure ecosystem-platform enterprise, like today’s Google or Microsoft,” and the base layer for the Internet of Everything is “most likely to be realized in China.” The article’s own institutional claim in Wang Chenglu’s voice: “我们的目标是到 2028 年成为一家纯的生态平台型企业，像今天的谷歌或微软” (Our goal is to become a pure ecosystem-platform enterprise by 2028, like today’s Google or Microsoft). And: “万物互联最可能实现的地方就是中国。场景、规模、硬件、芯片、模组、网络，全要素最全的就是中国。一旦中国把它做成，就相当于工业时代的英国” (The most likely place for the Internet of Everything to be realized is China. Scenarios, scale, hardware, chips, modules, networks — the most complete element set is in China. Once China makes it happen, it will be like the UK of the industrial era.)

From an institutional economics standpoint, this 2028-ecosystem-platform / Internet-of-Everything-China-base-layer pairing is a structural finding: a Chinese vendor is placing a specific 2028 ecosystem-platform target (in the Google/Microsoft analogy) inside a national-strategy framing (Internet of Everything as China’s industrial-era-UK claim). The institutional-form object documented here — a Chinese vendor explicitly claiming the base-layer realization of the Internet of Everything with a named 2028 enterprise-form target — is a structurally new vendor-official claim that prior briefings did not document at the OpenHarmony layer.

The M-Robots OS donation — the vendor-foundation separation pattern applied to a new domain

The article pairs the OpenHarmony base-layer discussion with a new institutional move in the robotics domain: Wang Chenglu confirms that 深开鸿 has donated M-Robots OS to the OpenAtom Foundation. The OpenAtom Foundation’s own official press releases document the timeline that the dialogue references:

  • April 17, 2025 — M-Robots OS 1.0 released by 深开鸿 as “the country’s first distributed heterogeneous multi-machine collaborative robot operating system” (OpenAtom Foundation)
  • July 25, 2025 — M-Robots OS fully open-sourced at the OpenAtom Open Source Ecosystem Conference, with 21 founding members including 深开鸿, iSoftStone, Digital Huaxia, Leju Robotics, Harbin Institute of Technology Chongqing Institute, Beijing University of Technology, and Beijing Institute of Technology (OpenAtom Foundation)
  • November 2025 — 深开鸿 donates the M-Robots project to the OpenAtom Foundation with an independent PMC (Project Management Committee) and 16 professional SIGs covering kernel, BSP/drivers, and system services / communication middleware (OpenAtom Foundation)
  • May 25, 2026 — M-Robots OS 2.0 released; six core capabilities including block-style framework, sub-1μs interrupt / task-switch latency, self-developed M-DDS low-latency collaboration technology (4 ms inter-body audiovisual latency, 42% better than Fast-DDS), ROS1/ROS2 + Dora-rs middleware compatibility reducing migration cost by 80% (OpenAtom Foundation)

The article’s own diagnostic for the institutional need: 76% of Chinese domestic robots still depend on foreign ROS and core algorithm autonomy is below 40% (2026 World Robot Conference context, per the article’s own framing). The article’s own institutional comparison — ROS vs. M-Robots OS — is framed not as an incremental improvement but as a full replacement (“完全换道，不一样的路” / “a completely different track, a different logic”): ROS is a middleware that runs on Linux for single robots; M-Robots OS is a “complete operating system” (kernel to upper layers) built on OpenHarmony for robot swarms.

From an institutional economics standpoint, this 76%-ROS-dependency / 40%-algorithm-autonomy / full-track-replacement framing is a structural finding: the OpenAtom Foundation is being used as the institutional venue for a domain-shift move (single-robot middleware → swarm-native full OS) inside a state-coordinated foundation, with a vendor-originator (深开鸿) as the donation source and 21 founding members drawn from industry + research + government-adjacent institutions — an institutional-form pattern that prior briefings’ AAIF/KubeCon-China observation did not document at the OpenAtom Foundation layer.

The 工信部 unified-OpenHarmony-IoT-standards claim — an administrative-industrial-policy placement

The article’s own statement that “工信部正在推进开源鸿蒙统一互联安全接入的行业标准” (MIIT is advancing the industry standard for unified, secure interconnection access for OpenHarmony) is a structurally new institutional-form placement that pairs with the September 6 briefing’s documentation of MIIT’s AI SME Startup Support Plan. The article’s own framing — “人类社会最大的效率提升就是标准化” (Standardization is the greatest efficiency gain in human society) — is the vendor-official reading of what the administrative standard means for the ecosystem.

From an institutional economics standpoint, this MIIT-OpenHarmony-standards / MIIT-AI-SME-Startup-Plan pairing is a structural finding: MIIT is being placed inside the same administrative-industrial-policy frame as both (a) the AI-SME-Startup-Plan (September 6) and (b) the OpenHarmony unified-standards claim (this briefing) — two administrative-industrial-policy placements by the same ministry inside the same one-week window on two different base-layer objects. The institutional-economics reading is that MIIT’s industrial-policy surface is now being documented at two distinct base-layer objects (AI-SME-startups and OpenHarmony-IoT-standards) within the same week — a structurally new observation that prior briefings did not document together.

Why this story matters for the Great Divergence 2.0 framework

Reading the Wang Chenglu dialogue and the OpenAtom M-Robots timeline together documents the ninth institutional surface on which the governance-layer-deficit question has been documented in the past week-plus:

  • Xie Lan’s Great Divergence discourse (August 31) — policy-discourse layer
  • COPU’s KubeCon China address (September 2) — national-coordination layer (drafted, video invitation)
  • Chongqing’s 15th-Five-Year-Plan (September 2) — municipal-planning layer
  • CAC’s Qinglang Phase 2 (September 3) — administrative-control layer
  • Ren Xudong’s essay (September 5) — vendor-official analytical layer (AAIF-seat recommendation)
  • MIIT’s AI SME Startup Support Plan (September 6) — administrative-industrial-policy layer (AI-SME-startups)
  • Huawei’s AAIF entry at AGNTCon China (September 6) — vendor-official agenda layer (actual AAIF entry)
  • Lu Shouqun’s KubeCon-China lifetime-achievement award (September 6) — Western-foundation-formal-recognition layer
  • Wang Chenglu’s OpenHarmony base-layer claim + M-Robots OS donation (September 7) — vendor-foundation-separation + national-base-layer-claim layer

Sources

🔍 WeChat Monitor — 16 Official Accounts Tracked

The 16 monitored WeChat official accounts were checked against the September 7, 2026 input file. One of the 16 had a non-empty entry on September 7; the remaining 15 had no new articles that met the institutional-threshold filter.

Active on September 7, 2026:

#AccountSource URLSignal
(其他)凤凰《AI 研究院》 (Phoenix AI Research Institute)mp.weixin.qq.com/s/sAtEmQLg1r27a0P_RAVZ_AWang Chenglu’s Phoenix AI dialogue: 14 billion OpenHarmony device installs (11 billion non-Huawei), 90%→70% Huawei contribution share, Hubble 20% stake in 深开鸿, M-Robots OS donated to OpenAtom, 76% robotics ROS dependency, 工信部 unified-OpenHarmony-IoT-standards, 2028 Google/Microsoft ecosystem-platform target, Internet-of-Everything-China base-layer claim

No new institutional content on September 7 (no entries in input file):

  1. 开放原子开源基金会 (OpenAtom Foundation)
  2. CCF 开源发展技术委员会 (CCF)
  3. 华为开源 (Huawei Open Source)
  4. 木兰开源社区 (Mulan)
  5. COPU 开源联盟
  6. 天工开物开源基金会 (Tiangong Kaiwu)
  7. 明说开源 (Ming Says)
  8. 智源 FlagOpen (BAAI FlagOpen)
  9. 开源社 KAIYUANSHE
  10. 硅基时间 (Silicon Time)
  11. AtomGit
  12. Linux 内核开发者大会 (LKDC)
  13. GOSIM / CSDN
  14. 黄大年茶思屋 (Huang Danian Tea House)
  15. 夜天之书 (Yetian’s Book)

Deduplication note: Wang Chenglu’s 凤凰《AI 研究院》 dialogue was not documented in any prior briefing — this is the first-documented appearance of the ex-Huawei HarmonyOS architect’s current institutional position at 深开鸿 in this series. The M-Robots OS donation itself (to OpenAtom Foundation, November 2025) was not documented in any prior briefing — the OpenAtom Foundation’s own press-release timeline is being cited here for the first time as an institutional source. The 14 billion OpenHarmony installed-base figure was not documented in any prior briefing — the article’s own claim about the 11 billion non-Huawei / 3 billion Huawei split is the structurally new institutional fact of the day. The 哈勃 20% stake was not documented in any prior briefing — the Hubble-20%-stake / moat-for-Huawei framing is the vendor-foundation-separation institutional fact of the day.

Organizational-tracking update: The Wang Chenglu / 深开鸿 institutional-form object is a structurally new placement at the vendor-foundation-separation layer — one of the four organizationally-tracked entities (BAAI, DeepSeek, Moonshot AI / Kimi, and now 深开鸿 as the ex-vendor architect running the vendor-separating foundation-backed entity) that had no documented move this week. The M-Robots OS donation to OpenAtom Foundation is a structurally new institutional-form object on the OpenAtom Foundation layer — an institutional object that the September 7 briefing’s AAIF/KubeCon-China observation did not document.

🔍 Commentary

One institutionally consequential move on the same unsolved object — the governance layer — from a different institutional surface than the prior seven briefings.

The Wang Chenglu / Phoenix AI dialogue and the OpenAtom M-Robots timeline documented here sit on the same unsolved institutional object as the previous seven briefings — the governance layer of China’s open-source ecosystem — but on a structurally distinct institutional surface: the vendor-foundation-separation + national-base-layer-claim layer, not the AAIF / KubeCon-China layer documented in the September 7 briefing.

One structurally new reading: the ex-Huawei architect as the institutional voice for a vendor-foundation separation that the vendor itself is running as its own moat.

Wang Chenglu’s own institutional position — a 24-year Huawei veteran, one of the original co-creators of HarmonyOS, who left Huawei in 2022 to run 深开鸿 with a 20% Hubble stake — is the article’s own institutional fact. His own diagnostic about the “natural process” of Huawei’s contribution share declining from 90% to 70% is the article’s own reading of what community maturity means for a vendor-originated base layer. His own claim that OpenHarmony is “not Huawei’s, it is everyone’s” — 严格意义上讲不是了 — is the article’s own vendor-foundation-separation claim, delivered by the ex-vendor architect.

The institutionally consequential fact that none of the prior seven briefings of the week documented is that a Chinese vendor’s own architectural heir is now, institutionally, running the vendor-foundation separation as the vendor’s own strategic moat — Ren Zhengfei’s own framing in the article — rather than as a foundation-owned asset separate from the vendor. This is a structurally new institutional reading of the vendor-foundation separation that the September 7 AAIF/KubeCon-China observation did not document at the OpenHarmony layer.

One structurally new pair of institutional moves on the OpenAtom Foundation layer.

The M-Robots OS timeline (April 2025 1.0 → July 2025 full open-source → November 2025 donation with independent PMC → May 2026 2.0) is an OpenAtom Foundation layer event that pairs with the September 6 briefing’s documentation of AtomGit’s national-community designation (also OpenAtom Foundation layer). Reading the two together documents the OpenAtom Foundation layer as an institutional venue for two distinct base-layer objects — AtomGit (code-hosting, documented September 6) and M-Robots OS (robot operating systems, documented this briefing) — in the same one-week window. From an institutional economics standpoint, this AtomGit / M-Robots OS pairing is a structural finding: the OpenAtom Foundation is being placed inside the same institutional frame as two distinct base-layer objects in the same week, and the institutional-form pattern being applied — vendor-originated project, foundation-donated, independent PMC, multi-member community — is being repeated across two distinct domains. The institutional-economics question the two moves raise together — and do not answer — is whether the OpenAtom Foundation is being institutionalized as the primary Chinese institutional venue for vendor-originated base-layer projects across multiple domains, or whether the vendor-foundation-separation pattern documented at OpenAtom will converge with the Western-neutral-foundation pattern documented at AAIF/KubeCon-China in the September 7 briefing.

One perspective, not a verdict.

The Wang Chenglu / 深开鸿 institutional reading and the M-Robots OS donation timeline are best read as observations of institutional movement in progress, not as verdicts on institutional direction. The 14 billion device install figure is an article’s own claim, not an audited institutional fact. The 2028 ecosystem-platform target is an article’s own claim, not a settled institutional form. The 76% robotics ROS dependency figure is a 2026 World Robot Conference context, not a settled policy outcome. What this briefing documents is that the governance-layer-deficit question has crossed a first-documented threshold on the ninth institutional surface — and that this ninth surface sits on the same unsolved object as the eight documented in the prior eight briefings, and that the vendor-foundation-separation pattern documented at the OpenHarmony / M-Robots OS layer is a structurally new institutional form that the AAIF/KubeCon-China layer of the September 7 briefing did not document.

Editorial note on perspective: This briefing presents one institutional-economics reading of Chinese open-source developments, not a verdict. The “institution” in these stories — the vendor-foundation-separation layer, the OpenAtom Foundation as a multi-domain base-layer venue, the national-base-layer-claim layer — is treated as an object of observation, not a target of critique. The Great Divergence 2.0 framework (FLOSS vs. State-Chartered Codebase vs. Intranet Shared Source vs. Cyber-Estate) is a lens, not a universal answer. One perspective, not a verdict.