Huawei: the company sanctions tried to break — and what it rebuilt
A neutral, evidence-first reading of Huawei, assembled from English and Chinese primary sources — both cases laid out, then weighed, with the reasoning shown.
69 sources · ~46% Chinese-languageAs of June 20269 analysis sections
In ~38 years Huawei went from a ¥21,000 Shenzhen switch reseller to the world's largest telecom-equipment maker — then absorbed the most aggressive technology sanctions ever aimed at one company, and clawed revenue back to within ~¥1bn of its all-time peak.
The genuinely open question is not whether Huawei survived — it plainly did — but how much of the comeback is durable advantage versus costly, China-bound substitution that remains capped at the technology frontier. The evidence cuts both ways on every major question below — but not equally far. Read together, it points to a comeback that is real at the level of survival and capped at the level of the frontier: revenue rebuilt to near-peak, silicon still a generation behind, and ~71% of sales at home. Each section lays out both cases and states where the weight of evidence sits; the full weighing closes the Risks & Challenges section.
The decisive questions
Each links to the section that lays out the evidence on both sides.
Group revenue in CNY bn (from audited annual reports). The 2020 peak, the 2021 sanctions collapse, and the grind back to near-peak by 2025 are the spine of the whole story.
Huawei group revenue, CNY bn (2019–2025)
⚖️
What reasonable people disagree about
Whether Huawei's self-reliance is a durable edge or sanction-forced inefficiency; whether its domestic 7nm/AI progress can ever reach the frontier without EUV and HBM; whether ~71% China reliance is resilience or fragility; and whether Western security concerns are well-founded or unproven. This study weighs each one rather than shrugging: capped-but-real substitution (high confidence), a genuine but lower-margin recovery (high confidence), a China-centred footprint (high confidence), and a moats question that is honestly contested. The reasoning, the counter-arguments and the tripwires that would flip each call are in Risks & Challenges.
How to read this
Nine sections, each built the same way: a neutral synthesis, a two-sided case-for / case-against ledger, dated quotes (with the original Chinese shown alongside any translation), and the sources used. Start with the question that interests you, or read in order from Overview.
🔍
Independent research artifact, not affiliated with or endorsed by Huawei. Huawei is private but publishes audited annual reports, so group figures are disclosed; segment and market-share figures mix disclosure with third-party estimates and are labeled. Where the research could not verify a claim, the relevant page says so. See Methodology & Limits.
Section 01
Overview & Timeline
From a Shenzhen switch reseller to the world's largest telecom-equipment maker and a full-stack ICT, devices, chips, cloud and automotive group — privately held, employee-owned, and reshaped by a decade under US sanctions.
6 sources1 Chinese-languageAs of June 2026
In ~38 years Huawei went from a **¥21,000 startup** to a group with **~208,000 employees in 170+ countries** and **¥880.9B (≈$119B) revenue (2025)** — but it remains **private, unlisted and employee-owned**, with a governance structure that is itself contested (ov-3, bm-1, ov-4).
The path in nine dates
A compressed timeline of the rise, the sanctions shock, and the engineered recovery that now defines Huawei. Each date links to its source.
1987Ren Zhengfei founds Huawei in Shenzhen with ~¥21,000, reselling phone switches [1][2]
1990s–2000sBuilds its own switches, expands across China then overseas; becomes a global carrier-equipment force [2]
2018CFO Meng Wanzhou detained in Vancouver on a US extradition request [45]
2019Added to the US Entity List (May); HiSilicon's "spare tire" chips go live [44][47]
2020Revenue peaks at ¥891.4B; sells Honor to survive the chip cut-off [62][16]
2021Revenue collapses −28.6%; Meng Wanzhou released and returns to China [62][45]
2023The Mate 60 Pro ships with an SMIC 7nm Kirin 9000S — the comeback [46]
2024HarmonyOS NEXT ("pure" HarmonyOS) launches; revenue rebounds +22.4%[54][58]
2025Revenue ¥880.9B — within ~¥1bn of the 2020 peak; R&D a record ¥192.3B [63][33]
Huawei Technologies (华为) was founded in September 1987 in Shenzhen by Ren Zhengfei (任正非), a former PLA engineer, reportedly with about ¥21,000 of capital, initially reselling telephone switches before building its own [1][2]. It is now the world's largest telecommunications-equipment maker, with ~208,000 employees, operations in 170+ countries, and five business groups — Carrier/ICT, Enterprise, Consumer, Huawei Cloud and Digital Power — plus an Intelligent Automotive unit [3]. Unusually for a global tech giant, it is private and unlisted, financed through retained earnings and bonds rather than public equity.
Ownership is Huawei's most contested governance feature. The holding company is ~99% held by a 'trade union committee' and ~1% by Ren Zhengfei, with employees participating through a virtual-restricted-stock scheme; Huawei presents this as genuine 100% employee ownership, and a union representing 60,000+ employees formally holds the 98.99% stake [6]. Critics — notably academics Balding & Clarke — argue that because the holding entity is a Chinese trade union, control may not truly rest with employees and could be state-linked, a claim that fuels Western distrust [4]. A detailed rebuttal counters that employee shares are legally not trade-union assets and that a separate Representatives' Commission, not the union, exercises shareholder rights [5]. The dispute is unresolved — and reputationally material abroad.
Operationally, Huawei runs a rotating-chairman system (six-month terms) currently shared by Eric Xu (徐直军), Ken Hu (胡厚崑) and Sabrina Meng Wanzhou (孟晚舟), who is also deputy chair and CFO — a design meant to avoid single-person dependency, though insiders concede founder Ren, now 80, is seen as irreplaceable[6]. The defining external fact is the decade of US pressure: the 2019 Entity List, the 2020 sale of Honor to ensure that brand's survival, and a pivot that pushed China to ~71% of revenue as overseas markets contracted [1][59].
Both sides of the ledger
The strongest sourced case on each side, held to the same standard. The synthesis above states where this study thinks the weight of evidence sits; this ledger is what that lean has to survive.
The case for
+Rare full-stack breadth: a single firm spans carrier networks, smartphones, in-house chips (HiSilicon), an OS (HarmonyOS), cloud and automotive — a vertical reach no Western peer matches [3].
+Private, employee-owned and debt-financed, so it can invest through downturns without public-market quarterly pressure; ~150,000 staff share in profits via virtual stock [5][6].
+Demonstrated resilience: revenue recovered to ¥880.9B in 2025, within ~¥1bn of the pre-sanction 2020 peak, after a −28.6% collapse in 2021 [12][62].
The case against
−Opaque, contested governance: the trade-union ownership structure and Ren's reported veto draw credible academic criticism that control may be state-linked — unresolved and a barrier in Western markets [4].
−Geographic concentration: ~71% of revenue now comes from China, leaving the group exposed to one slowing economy and to the sanctions that capped its once-global business [59].
−Succession risk: governance still effectively centres on an 80-year-old founder, and the rotating-chair system is untested by a true post-Ren transition [6].
Sources for this section
6 sources · en, zh · tiers shown. Full bibliography on the Sources page.
Huawei sits across telecom equipment (shrinking), China cloud (it is #2) and AI accelerators (the contested frontier) — markets where it usually leads at home but is politically boxed in abroad.
5 sources2 Chinese-languageAs of June 2026
Huawei's arenas are big but slowing or politically bounded: global telecom equipment **fell 11% in 2024**, it is a clear **#2 in China cloud (18%)** but invisible globally, and in **AI chips it leads China domestically** while the US Big Three and Nvidia dominate worldwide (mk-5, mk-1, mk-2, mk-3).
Its strongest position: global telecom equipment
2024 global telecom-equipment share [24]. Huawei's ~31% lead is stable, but the pie is shrinking (revenue fell 11% in 2024) [11] — a low-growth, share-fight industry.
Global telecom-equipment share, 2024
Huawei — 31%
Others — 31%
Nokia — 14%
Ericsson — 13%
ZTE — 11%
Five Forces: attractive to incumbents, hard against Huawei
Click each force for the rated pressure and the evidence. Integration and patents build a real barrier to entrants and substitutes — but rivalry, government buyers and (uniquely for Huawei) sanctioned inputs keep pressure high.
Telecom equipment & ICT infrastructure
Competitive rivalry — High. Worldwide telecom-equipment revenue fell 11% in 2024 — the steepest drop in 20+ years — so Huawei (31%), Ericsson, Nokia and ZTE fight over a shrinking pie, even as Huawei holds a stable lead (cm-1, mk-5).
Huawei's core market — telecom equipment — is mature and contracting: worldwide revenue fell 11% in 2024, the steepest annual drop in over 20 years, before a ~4% recovery in 1H2025 [11]. The main architectural challenger, Open RAN, has underwhelmed: it is projected to be under 5% of RAN deployments by 2030, and market concentration is actually higher now than before the O-RAN Alliance — which protects integrated incumbents like Huawei but also signals a low-growth, share-fight industry [10].
The growth arenas are cloud and AI. In China's cloud-infrastructure market (~$11.6B in Q1 2025, +16%), Huawei Cloud is the established #2 at 18% share, behind Alibaba Cloud's 33% and ahead of Tencent's 10% [7]. Globally the picture inverts entirely: AWS 29%, Azure 20%, Google 13% — 63% of a $390B market — leave Chinese hyperscalers, Huawei included, essentially off the global board [8].
The most contested arena is AI accelerators. Nvidia still leads China (~39% by one count) with Huawei a close second, and Bernstein projects Huawei could reach ~50% and Nvidia fall to ~8% by 2026 as US controls bite and domestic demand compounds ~74% a year [9]. The throughline across every market: Huawei is structurally strong where it can sell, but its addressable market is bounded by sanctions abroad and by maturing demand at home — so growth must come from share gains and monetization depth, not new geographies.
Both sides of the ledger
The strongest sourced case on each side, held to the same standard. The synthesis above states where this study thinks the weight of evidence sits; this ledger is what that lean has to survive.
The case for
+Leadership in the fastest-growing domestic slice: clear #2 in China cloud and domestic #1 in AI accelerators, with projections toward ~50% China AI-chip share by 2026 as Nvidia is curbed [7][9].
+Industry structure favours incumbents: Open RAN's failure to scale (<5% by 2030) keeps the integrated-vendor moat intact [10].
+Diversified across telecom, cloud, AI, devices, energy and auto, so a slowdown in one pool can be partly offset by another [7][11].
The case against
−The core telecom market is shrinking (−11% in 2024) — a structural headwind for Huawei's largest segment [11].
−Globally invisible in cloud: the Big Three hold 63% and Chinese players barely register outside China, capping the cloud growth story to one market [8].
−Its China AI-chip 'leadership' partly depends on US policy curbing Nvidia, not on out-competing it on the merits — a politically contingent advantage [9].
Sources for this section
5 sources · zh, en · tiers shown. Full bibliography on the Sources page.
A vertically integrated ICT group: two mature pillars (carrier/computing and consumer devices) fund three faster bets — digital power, intelligent auto, and a strategically vital but loss-leading cloud/AI stack.
12 sources8 Chinese-languageAs of June 2026
Financially, Huawei rests on **two big mature pillars** — ICT (¥375.0B) and Consumer (¥344.5B) — that together are ~82% of 2025 revenue, while **Digital Power (+12.7%), Auto (+72.1%)** grow fast and **Cloud actually shrank 3.5%** (bm-2).
Where the revenue comes from
FY2025 revenue by business segment, CNY bn [13]. Two mature pillars dwarf the three growth bets; hover a bar for its growth rate. Cloud is the only segment that shrank.
FY2025 revenue by segment (CNY bn)
ICT Infrastructure
¥375B
Consumer / devices
¥344.5B
Digital Power
¥77.3B
Intelligent Auto
¥45B
Huawei Cloud
¥32.2B
The revenue base is two mature pillars. ICT Infrastructure (carrier networks + enterprise computing) was the largest segment at ¥375.0B (+2.6%) in 2025 — base stations, 5.5G/5G-A, optical and storage (OceanStor serves 26,000+ customers in 150+ countries, including 53 of the world's top-100 banks) — but it is now China-centric and cyclical as global carrier capex shrinks [13][22]. The Consumer/devices business reached ¥344.5B (+1.6%), a sharp cooldown from +38.3% in 2024; the bull point is that Huawei reclaimed #1 in China smartphones on self-made Kirin silicon and HarmonyOS, the bear point that growth has stalled to low single digits [13][15]. The 2020 sale of Honor — to 30+ supply-chain and dealer firms, with Huawei keeping no equity — was the foundational move that let the budget brand escape the chip cut-off [16].
The growth narrative lives in the newer segments. Digital Power hit ¥77.3B (+12.7%) — solar inverters (~176 GW shipped in 2024, world #1), energy storage and EV charging — a genuinely fast, profitable grower [13][17]. Intelligent Automotive surged +72.1% to ¥45.0B, after first turning an annual profit in 2024 [13][14]. The model is deliberately asset-light: under the 'Huawei does not build cars' (华为不造车) pledge, signed in 2020 and extended five years in 2023, Huawei supplies smart-driving and cockpit tech while partners manufacture, mainly via the HarmonyOS Intelligent Mobility (鸿蒙智行 / HIMA) alliance [19]. HIMA delivered 589,107 vehicles in 2025 (+32%), with AITO/问界 ~70%+ of volume, and Huawei carved the unit into Yinwang (引望), selling 20% for ¥23B cash at a ¥115B valuation[18][20].
The clear soft spot is Huawei Cloud, which fell 3.5% to ¥32.2B, even as global cloud grew ~29% — a striking laggard despite the Pangu (盘古) models and Ascend-powered AI cloud [13]. The overarching logic: sanctions forced a pivot from a globalized carrier-equipment vendor into a China-anchored conglomerate where mature pillars fund fast new bets and a strategically vital — if still subscale — AI/chip/cloud stack.
Both sides of the ledger
The strongest sourced case on each side, held to the same standard. The synthesis above states where this study thinks the weight of evidence sits; this ledger is what that lean has to survive.
The case for
+New segments are real engines, not slideware: Digital Power ¥77.3B (+12.7%, world #1 in solar inverters) and Auto ¥45.0B (+72.1%, now profitable)[13][17].
+Devices comeback with full vertical control — retook #1 in China smartphones on in-house Kirin chips + HarmonyOS, a silicon-to-OS stack no sanctioned peer has reconstituted [15].
+Capital-efficient auto play: the asset-light HIMA model (589,107 deliveries, +32%) plus the Yinwang carve-out (¥23B raised, 80% retained) monetize the unit while binding automaker partners [18][20].
The case against
−The mature pillars are stalling — ICT +2.6%, Consumer +1.6% — dragging group growth to just +2.2%, so diversification has not yet offset legacy fatigue [13].
−Cloud/AI is losing share, the one segment that should be the future shrank 3.5% while rivals grew double digits [13][23].
−Auto growth is uneven and price-war-exposed: within HIMA, Luxeed/智界 collapsed from ~12,000 to 1,728 monthly units and other brands drew quality complaints [21].
In their words
“Huawei will not hold any shares or be involved in any business management or decision-making activities in the new Honor company.”
Huawei · official statement on the Honor sale · 17 Nov 2020 · source
Sources for this section
12 sources · en, zh · tiers shown. Full bibliography on the Sources page.
A leader where it can sell, boxed out where it cannot: #1 in global telecom equipment and China smartphones, yet absent from the global smartphone top 5 and well behind Nvidia on AI-chip performance.
9 sources2 Chinese-languageAs of June 2026
Huawei holds **31% of global telecom equipment** (now #1 even outside China) and **#1 in China smartphones (16.4%)**, yet is **absent from the global smartphone top 5** and its best AI chip is ~**5× behind** the best US chip (cm-1, cm-5, cm-6, cm-7).
Integration vs. market access, mapped
Huawei and its rivals on two axes that define its position: vertical integration / R&D depth and market access. Hover a point for the sourced basis. Huawei's top-left placement — deepest stack, most-constrained access — captures its core tension in one chart.
Hover a point to see the basis for its placement.
In telecom equipment — its fortress — Huawei held 31% of the global market in both 2024 and 1H2025, far ahead of Nokia (~14%/13%), Ericsson (~13%/12%) and ZTE (~11%/10%) [24][25]. Crucially the dominance is no longer just a home-market artifact: Huawei passed Nokia to become #1 even outside China, holding ~21% ex-China and ~40% of the non-North-America market [25][26]. In RAN specifically the worldwide top five are Huawei/Ericsson/Nokia/ZTE/Samsung, but outside China the order flips to Ericsson/Nokia/Huawei — Western markets still favour the European pair [27].
In smartphones the story is starkly two-sided. Huawei reclaimed #1 in China for full-year 2025 — 46.7M units, 16.4% share — narrowly ahead of Apple and vivo, powered by self-made Kirin silicon [28]. But stripped of Google Mobile Services since 2019, Huawei does not appear in the global top 5 (Apple ~19.7%, Samsung, Xiaomi, vivo, OPPO): the comeback is real but geographically trapped at home [29].
In AI accelerators the gap is the sharpest. Huawei Ascend is China's clear domestic #1, but the CFR estimates US chips are ~5× more powerful (forecast ~17× by 2027), SMIC is stuck at 7nm, and an HBM shortage caps Huawei at ~200,000–300,000 completed AI chips a year versus Nvidia's ~1M H20s sold in China in 2024 [30][31]. And the addressable market is politically bounded: 11 EU states restrict Huawei 5G and the US runs a $1.9B rip-and-replace program [32]. Where Huawei can sell, it usually wins; where geopolitics intervenes, rivals keep the West.
Both sides of the ledger
The strongest sourced case on each side, held to the same standard. The synthesis above states where this study thinks the weight of evidence sits; this ledger is what that lean has to survive.
The case for
+Undisputed #1 in the core market, now winning abroad too: 31% of global telecom equipment, having passed Nokia to lead even outside China [24][25][26].
+China smartphone crown reclaimed (46.7M units, 16.4%), beating Apple and vivo despite the GMS handicap [28].
+Domestic #1 in AI accelerators and the only credible Ascend-based alternative to Nvidia inside China [30][31].
The case against
−Absent from the global smartphone top 5 — the China win has not translated into global scale post-GMS [29].
−The AI-chip hardware gap is real and may widen: ~5× behind today, ~17× by 2027, with output throttled by HBM scarcity [30][31].
−Geographically capped: 11 EU states restrict Huawei 5G and the US funds rip-and-replace — large premium markets are structurally closed [32].
Sources for this section
9 sources · en, zh · tiers shown. Full bibliography on the Sources page.
Out-invest in R&D and vertically integrate to the point of self-sufficiency (自主可控) — a doctrine forced by sanctions that built real moats (chips, OS, patents) at the cost of efficiency and a reliance on the home market.
10 sources8 Chinese-languageAs of June 2026
Huawei spends **¥192.3B — 21.8% of revenue — on R&D** (≈$26B, a record) and holds **~165,000 patents**, funding a self-sufficiency stack (HiSilicon, HarmonyOS, openEuler, MetaERP) that is a real moat — but one that is **substitution forced by sanctions**, not a freely chosen edge (st-1, st-2, st-7).
🧩
The moat, in one line
Huawei's advantage is R&D depth plus end-to-end vertical integration — chips, OS, network, devices and cloud under one roof. The open question is whether that is a chosen, durable edge or a sanctions-forced substitution that is costly, inefficient, and dependent on the China market and state support.
R&D through the cycle
R&D spend, CNY bn [33][58][61]. Huawei held R&D at 20–25% of revenue even through the 2021–22 collapse — the clearest evidence of its invest-through-the-cycle doctrine.
Huawei R&D investment, CNY bn
The defining strategy is 'investing through the cycle' in R&D. In 2025 Huawei spent ¥192.3B — 21.8% of revenue (≈$26B) — lifting cumulative 10-year R&D past ¥1.382 trillion, with ~114,000 staff (53.7%) in R&D [33]. That depth underwrites a patent moat (~165,000 valid patents; ~$630M of 2024 licensing revenue across 2.7B+ 5G devices) and, crucially, the self-sufficiency (自主可控) that let Huawei survive sanctions [33][34]. The 'spare tire' (备胎) doctrine extended into a full domestic stack: HiSilicon chips, HarmonyOS, the openEuler OS, GaussDB, and a self-built MetaERP that replaced Oracle/SAP in 'the most extensive and complex project in Huawei's history' [36].
Strategically Huawei is revealed-versus-stated: publicly it pivoted to 'survival as the main guideline' (Ren Zhengfei, 2022: make survival the priority through 2025, 'transmit the sense of crisis to everyone,' shrink edge businesses), while in practice it reallocated into higher-value growth — auto, digital power, cloud — and now frames the future as 'All Intelligence' (全面智能化): supply sustainable AI compute (Ascend + Pangu) and embed AI across its products [35][37]. A provocative outside reading comes from ITIF, which argues US export controls backfired — they 'helped create and accelerate Huawei's OS,' forced replacement of 13,000+ components, and cost US suppliers $33B+ in lost Huawei sales [39].
The moats — and what could erode them — cut both ways. R&D depth and patents are durable, but licensing income (~$630M) is tiny relative to revenue. Vertical integration is a genuine differentiator no Western peer matches, but it is costly and inefficient by design — a sanctions-forced substitution, not a chosen advantage. HarmonyOS lock-in is growing (17M+ HarmonyOS 5 devices, a ¥1B 'Tiangong' developer fund) but its app catalog still depends on subsidy [40]. And national-champion status brings an estimated large state-support tailwind that is simultaneously Huawei's strongest domestic moat and the central plank of the Western 'state-linked' critique [39]. On culture, Ren defends the 'wolf culture' (狼性文化) as sensitivity, teamwork and perseverance and denies an enforced 996 — though critics cite overwork and a militarized 'striver' ethos [38].
Both sides of the ledger
The strongest sourced case on each side, held to the same standard. The synthesis above states where this study thinks the weight of evidence sits; this ledger is what that lean has to survive.
The case for
+Sustained 21.8%-of-revenue R&D (¥192.3B; ¥1.382tn cumulative) and ~114,000 R&D staff create technical depth no peer matches [33].
+Unmatched vertical integration — HiSilicon chips, HarmonyOS, openEuler, GaussDB, MetaERP — delivering real sanctions resilience and a full domestic stack [36].
+The new-growth pivot is working and a coherent 'All Intelligence' AI strategy now anchors the next decade [37].
The case against
−The moats are substitution forced by sanctions, not free choice — costly and inefficient; ITIF frames self-sufficiency as a backfire by-product, and critics tie Huawei's scale to large state support and ~71% home-market reliance [39][41][42].
−Patent licensing income (~$630M) is small relative to revenue, and HarmonyOS lock-in still leans on a ¥1B subsidy rather than self-sustaining network effects [34][40].
−Culture risk: the 'wolf'/'striver' model and reliance on national-champion state support are both contested — celebrated at home, distrusted abroad [38][39].
In their words
“From 2023 and even to 2025, we must make survival the main guideline... transmit the sense of crisis to everyone.”
original · zh“2023年甚至到2025年，一定要把活下来作为最主要的纲领……把寒气传递给每个人。”
Ren Zhengfei (任正非) · Founder, Huawei — internal speech · 2022 · English is a translation from zh · source
“Wolves have three traits: sensitivity, teamwork, perseverance... We don't have a '996' rule; even our lowest-level employees can't work excessive overtime.”
original · zh“狼有三个特点：敏感性、团队性、不屈不挠性……我们没有996的说法……我们最基层的员工想多加一点班也不行。”
Ren Zhengfei (任正非) · Founder, Huawei · 2020 · English is a translation from zh · source
Sources for this section
10 sources · zh, en · tiers shown. Full bibliography on the Sources page.
Huawei's defining decade: the 2019 Entity List and 2020 chip rule cut it off from Google and TSMC; the response — the 'spare tire,' the Mate 60, Ascend AI super-nodes and HarmonyOS — shows genuine substitution, capped by a still-contested security debate.
16 sources5 Chinese-languageAs of June 2026
Huawei has demonstrably **survived and partially substituted** for what sanctions removed — domestic **7nm silicon (Mate 60)**, a competitive AI **super-node**, a from-scratch **OS** — yet remains **structurally constrained** by EUV access, HBM supply and an app/CUDA gap, while the underlying security concerns stay **contested, not resolved** (sx-4, sx-6, sx-12, sx-8).
The pressure, in seven dates
How the sanctions campaign escalated, and where Huawei pushed back. Each date links to its source.
Dec 2018CFO Meng Wanzhou detained in Vancouver on a US fraud / Iran-sanctions request [45]
May 2019Huawei added to the US Entity List; Google strips GMS from new phones [43][44]
Feb 2020US DOJ files a 16-count superseding indictment charging Huawei and two US subsidiaries with conspiracy to violate RICO (racketeering) and to steal trade secrets, plus business in Iran and North Korea in breach of US sanctions [74]
May 2020The Foreign-Produced Direct Product Rule cuts Huawei off from TSMC's leading-edge chips [43]
Sep 2021Meng released under a deferred-prosecution deal; two detained Canadians freed [45]
Aug 2023The Mate 60 Pro ships with an SMIC-built 7nm Kirin 9000S — without EUV [46]
Oct 2024HarmonyOS NEXT ("pure-blood" 纯血鸿蒙) drops Android/AOSP entirely [54]
May 2025US BIS: using Ascend chips "anywhere in the world" violates export controls [53]
Where the self-reliance bet still hits a wall: AI chips
The clearest measure of the gap. In China in 2024, Nvidia moved roughly five times more AI accelerators than Huawei's Ascend, and an HBM-memory shortage caps Huawei at an estimated ~200,000–300,000 completed AI chips a year. Substitution is real — but capped, for now, well below the volume the frontier vendor ships into the same market.
AI accelerators shipped in China, 2024 (thousands of chips)
Nvidia H20
1,000k
Huawei Ascend 910B
200k
China-market unit volumes, 2024: Nvidia ~1,000k H20 vs Huawei ~200k Ascend 910B [31]; an HBM-memory shortage is estimated to cap Huawei at ~200,000–300,000 completed AI chips per year, and US chips remain ~5× (forecast ~17× by 2027) more powerful [30]. Apples-to-apples on volume; capability gap is larger still.
⚖️
The security debate — both sides, held at once
The US case is structural: Huawei is seen as a potential extension of the Chinese state under the 2017 National Intelligence Law [57], the logic behind Australia's 2018 5G ban and a 2025 Belgian bribery raid [52][56]. The counterpoint: a decade of allegations has not produced published proof of a deliberate backdoor, and Huawei flatly denies espionage [51]. Our weighing: contested on the underlying espionage facts — no public forensic proof has surfaced in a decade, and the denial has never been falsified — but effectively settled on the consequences, because Western governments demonstrably price the structural risk whether or not it is proven [52][32]. What would move it: a published, technically attributed backdoor finding, or a Five Eyes government reversing a 5G ban.
The defining decade began on 16 May 2019, when the US placed Huawei on the Entity List, forcing Google to strip Google Mobile Services from its phones [44]. The decisive blow came in May 2020, when the Foreign-Produced Direct Product Rule was amended to bar any foreign foundry — notably TSMC — from making Huawei-designed chips with US technology, severing Huawei from leading-edge silicon [43]. Layered on top was the arrest of CFO Meng Wanzhou in Vancouver in December 2018 on US fraud charges tied to alleged Iran-sanctions evasion; she was released in September 2021 under a deferred-prosecution agreement, after which China freed two detained Canadians — a sequence Western governments cast as 'hostage diplomacy' and Beijing rejects [45].
Huawei's answer was the 'spare tire' (备胎) strategy run by HiSilicon — years of building in-house fallbacks for 'extreme survival.' The payoff arrived in August 2023: the Mate 60 Pro, whose Kirin 9000S TechInsights confirmed was fabbed by SMIC on a 7nm process without EUV — proof China could reach 7nm under sanctions, even if the die is coarser than true 5nm [47][46]. The same playbook scaled to AI with Ascend: unable to match Nvidia per-chip, Huawei went system-level, and SemiAnalysis found the CloudMatrix 384 super-node delivers ~300 BF16 PFLOPs — nearly double Nvidia's GB200 NVL72 — but at ~4.1× the power[48]. Chinese coverage frames this as deliberate doctrine: 'use mathematics to supplement physics, systems to supplement single points' [49]. Tellingly, Ren himself conceded the US 'exaggerated' Huawei and that its single chips 'still lag the US by one generation' [50].
Two hard ceilings remain. Washington's May 2025 BIS guidance declared using Ascend 910B/C/D chips anywhere in the world a violation of US export controls, and HBM memory plus SMIC capacity throttle output [53]. On software, HarmonyOS NEXT ('pure-blood' 纯血鸿蒙) shed Android/AOSP entirely in October 2024 — a genuinely independent OS — but Huawei's own 2025 goal of 'sprinting to 100,000 native apps' concedes how thin the ecosystem still is versus Android/iOS [54][55].
On the security question, neutrality requires holding both sides in view. The US case is structural: policymakers view Huawei as a potential commercial extension of the CCP, citing China's 2017 National Intelligence Law and large state support, a logic that drove Australia's 2018 5G ban and a 2025 Belgian bribery raid on Huawei's EU office [57][52][56]. The counterpoint is that a decade of allegations has not been accompanied by published technical proof of a deliberate backdoor, and Ren has flatly denied espionage: 'we absolutely never install backdoors. Even if we were required by Chinese law, we would firmly reject that' [51]. The honest read: the concerns rest on real structural facts (Chinese law, opacity), while the denial rests on the absence of disclosed forensic evidence — both true at once.
Both sides of the ledger
The strongest sourced case on each side, held to the same standard. The synthesis above states where this study thinks the weight of evidence sits; this ledger is what that lean has to survive.
The case for
+Domestic 7nm silicon under full sanctions: TechInsights independently confirmed the SMIC-built Kirin 9000S in the Mate 60 — a milestone reached without EUV [46].
+System-level AI competitiveness: CloudMatrix 384 out-aggregates Nvidia's flagship rack on raw BF16 compute by going wide instead of deep [48][49].
+Security concerns remain unproven in public: no disclosed forensic backdoor, and Huawei consistently denies espionage [51].
The case against
−Stuck near 7nm at punishing economics, with the frontier (EUV, leading-edge nodes) still out of reach [46][50].
−US rules cap AI output: the May-2025 guidance makes using Ascend chips anywhere a claimed export-control violation, and HBM/SMIC capacity throttle volume [53].
−Huawei's own founder concedes chips lag the US 'by one generation' [50].
−Security/governance distrust is live and structural — from Australia's ban to the 2025 Belgium bribery raid — even absent published backdoor proof [52][56][57].
In their words
“Today is history's choice; all the backup tires (备胎) we once built have, overnight, all turned into the 'real thing.'”
original · zh“今天，是历史的选择，所有我们曾经打造的备胎，一夜之间全部转'正'。”
He Tingbo (何庭波) · President, HiSilicon — letter the night of the Entity-List designation · 17 May 2019 · English is a translation from zh · source
“We absolutely never install backdoors. Even if we were required by Chinese law, we would firmly reject that.”
Private but audited: the revenue line traces the sanctions arc — a 2020 peak, a 2021 collapse, and a grind back to near-peak by 2025 — while profit has been distorted by one-off divestiture gains.
8 sources4 Chinese-languageAs of June 2026
2025 revenue of **¥880.9B was ~¥1bn short of the 2020 peak**, with net profit recovering **+8.6% to ¥68B** — but recent profit numbers are noisy: 2024 fell **−28%** as one-off income dried up, and the 2021/2023 spikes were flattered by **Honor/xFusion divestiture gains** (fn-6, fn-1, fn-5).
The arc that defines Huawei
Group revenue, CNY bn [58][61][62][63]. The 2021 collapse (Entity List + Honor sale) and the grind back to within ~¥1bn of the 2020 peak are the spine of the revenue arc. Hover a point for the year's note.
Huawei group revenue, CNY bn
⚖️
Read the profit line carefully
The headline profit jumps of 2021 (+75.9%) and 2023 (+144%) were flattered by one-off Honor / xFusion divestiture gains[62], and 2024 net profit then fell ~28% when that "other income" dried up [60]. The cleaner read is 2025's underlying operating profit, up +22% to ¥96.9B[64].
The expectations bar: what the only market price implies
Huawei is unlisted, so there is no group valuation to argue with — but one piece of it has been priced by outside money. Huawei carved its car unit into Yinwang (引望) at a ¥115B valuation, selling 20% to Avatr and Seres for ¥23B cash, on H1-2024 revenue of ¥10.4B and net profit of ¥2.2B [20]. Annualize those half-year figures (illustrative — it assumes no seasonality) and the buyers paid roughly ¥115B ÷ ~¥4.5B ≈ 26× earnings, or ~5.5× revenue. That is a growth-supplier multiple, and it is the bar both readings are measured against: for the price to look cheap, the bulls need the auto segment's +72% surge to ¥45.0B in 2025 [13] to keep compounding; the bears need only note that the buyers are also Huawei's dependent car partners [72], so the one external price on a Huawei asset is not a fully arm's-length one. Not a recommendation — just the only disclosed bar there is.
The trajectory is the story. Revenue peaked at ¥891.4B in 2020, then collapsed −28.6% to ¥636.8B in 2021 after the Entity List and the Honor sale, bottomed around ¥642.3B in 2022, and climbed back: ¥704.2B (2023), ¥862.1B (2024, +22.4%) and ¥880.9B (2025, +2.2%) — within ~¥1bn of the all-time peak [62][61][58][63]. Throughout, Huawei kept R&D at 20–25% of revenue, even in the trough — the clearest signal of its invest-through-the-cycle doctrine [58][61].
Profit, though, is noisy and warrants scrutiny. The headline +75.9% jump in 2021 and +144% in 2023 were flattered by one-off gains from selling Honor and xFusion — roughly half of 2021 operating profit came from such disposals [62]. Conversely, 2024 net profit fell ~28% to ¥62.6B precisely because that 'other net income' dried up (from ¥62.7B to ¥8.3B) and operating margin dropped to 9.2% from 14.8%[59][60]. The cleaner read is 2025: operating profit rose +22% to ¥96.9B at an 11% margin, with operating cash flow up 44% — suggesting the *underlying* business, not disposals, is now driving results [64].
The forward bet is explicit and capital-hungry. Management calls AI 'possibly the last technological revolution in human history' and is pouring record R&D into Ascend/Pangu — even as Huawei Cloud revenue fell 3.5% in 2025, the one segment that shrank, a reminder that the AI bet is not yet paying off in cloud [64][65]. As a private company Huawei has no public valuation; the financials are best read as a resilient, high-reinvestment hardware group whose reported profitability has been harder to interpret than its revenue line.
Both sides of the ledger
The strongest sourced case on each side, held to the same standard. The synthesis above states where this study thinks the weight of evidence sits; this ledger is what that lean has to survive.
The case for
+Recovery to near-peak scale despite sanctions: ¥880.9B in 2025, within ~¥1bn of the 2020 record, with profit growing again [63].
+Clean-core profitability is improving: 2025 operating profit +22% to ¥96.9B (11% margin) and operating cash flow +44% — driven by the business, not disposals [64].
+Extraordinary, counter-cyclical R&D at 20–25% of revenue sustained even through the 2021–22 trough [58][61].
The case against
−Profit quality is muddy: 2021/2023 spikes were one-off Honor/xFusion gains, and 2024 fell ~28% when those gains vanished [62][60].
−Margins compressed hard in 2024 (op margin 9.2% vs 14.8%), showing how thin the underlying core can be [60].
−The AI growth bet isn't yet showing up where it should — Cloud revenue fell 3.5% in 2025 [65].
In their words
“AI may be the last technological revolution in human history... this will be the biggest development opportunity over the next decade or longer.”
original · zh“人工智能可能是人类历史上最后一场技术变革……这将是未来十年、甚至更长周期里最大的发展机会。”
Huawei management · FY2025 annual report · 31 Mar 2026 · English is a translation from zh · source
Sources for this section
8 sources · en, zh · tiers shown. Full bibliography on the Sources page.
Huawei against the telecom-equipment peers it competes with most directly. Scopes differ — Huawei is a whole-group conglomerate, the others are narrower — so read the table as directional, not like-for-like.
5 peersAs of June 2026
⚠️
Not strictly like-for-like
Huawei's ¥880.9B is a whole-group figure spanning networks, devices, chips, cloud, power and auto; Ericsson, Nokia and ZTE are far more focused on carrier equipment, and Cisco is enterprise-IP-led. Different fiscal years and currencies (converted at ~7.4 RMB, ~10.7 SEK, ~0.92 EUR per USD). Treat as scale context, not a precise ranking.
Revenue, to scale
Latest-FY revenue, US$B [66][67][68][69]. Huawei's group revenue dwarfs the pure-play vendors — but largely because its scope is far broader.
Latest-FY revenue, US$B (Huawei = whole group)
Huawei
$119B
Cisco
$57B
Ericsson
$23B
Nokia
$21B
ZTE
$17B
Company
Revenue
Growth
R&D % rev
Ownership
Positioning
Huawei
¥880.9B FY25 (≈$119B)
+2.2%
21.8%
Private, employee-owned
World #1 telecom equip (31%); full chips→devices→network→cloud stack
Cisco
$56.7B FY25
+5% (rev recovery)
~13%
Public (NASDAQ:CSCO)
Enterprise IP networking, not mobile RAN; adjacent rather than head-to-head
Ericsson
SEK 247.9B FY24 (≈$23B)
−6%
~17%
Public (NASDAQ:ERIC)
#1 RAN outside China; network-centric, no captive silicon/devices
Nokia
€19.2B FY24 (≈$21B)
−9%
~17%
Public (HEL/NYSE:NOK)
#2 telecom equip (14%); strong fixed/optical/IP
ZTE
¥121.3B FY24 (≈$16.7B)
by segment
~20%
Public, state-linked
China's #2 vendor; similar sanctions exposure to Huawei, smaller scale
The pattern
On the shared battlefield of telecom equipment, Huawei is the clear leader — ~31% global share vs Nokia's ~14% and Ericsson's ~13% [24] — and it spends more on R&D, both in absolute terms and as a share of revenue, than any of them [66]. The European pair's edge is the mirror image of Huawei's weakness: open access to Western markets that sanctions deny Huawei, which is why Ericsson and Nokia still lead RAN outside China [25]. ZTE shares Huawei's integration and sanctions exposure at smaller scale; Cisco is larger overall but competes in enterprise IP, not mobile networks. Fuller competitive evidence is on the Competitive Landscape page.
Section 09
Risks & Challenges
Mostly structural, not cyclical: a chip-supply ceiling, deeper extraterritorial sanctions, heavy domestic-market reliance, a subsidy-dependent OS ecosystem, auto-partner dependence, and founder succession.
4 sources2 Chinese-languageAs of June 2026
The honest bear case is structural, not cyclical: a **chip ceiling** (~5% of Nvidia's AI compute, stuck at 7nm), **71% China reliance**, extraterritorial **Ascend bans**, a subsidy-dependent OS ecosystem, and an **80-year-old founder** at the centre of governance (rk-1, rk-2, sx-11, ov-6).
⚠️
The standing terms, not passing storms
These are conditions Huawei operates within, unlikely to resolve inside the as-of window: a chip ceiling (~5% of Nvidia's AI compute, stuck at 7nm) [70], an extraterritorial Ascend ban[53], ~71% China reliance[71], a subsidy-leaning HarmonyOS ecosystem [55], auto-partner cannibalization [72], and an 80-year-old founder still central to governance [6].
None of this is a forecast of failure: Huawei has already absorbed the harshest technology sanctions ever aimed at one company and rebuilt revenue to within ~¥1bn of its peak [73]. But these standing terms are what the study's four questions get answered against — so here, in closing, is where the evidence lands on each.
The weighing
On whether the self-reliance bet beat the sanctions: the evidence leans toward survival won, frontier not — real substitution, hard-capped below the leading edge (high confidence). The controlling evidence is Ren Zhengfei's own concession that Huawei's single chips lag the US "by one generation" [50] and CFR's estimate that Huawei produced only ~5% of Nvidia's aggregate AI compute in 2025 [70], which outweighs the Mate 60's SMIC-7nm breakthrough [46] because the breakthrough is a node, not a volume — an HBM shortage caps output at ~200,000–300,000 AI chips a year [30]. The strongest surviving counter-argument: ITIF's "backfire" case that controls accelerated Huawei's own OS and 13,000+ component replacements while costing US suppliers $33B+ in sales [39]. What would flip this reading: a verified sub-7nm SMIC node in a shipping Kirin (a TechInsights-class teardown), or credible reporting of Ascend output above ~500,000 units a year. Pre-mortem: if this looks wrong in two years, the most likely reason is that CloudMatrix-style system engineering [48] closed more of the effective-compute gap than per-chip metrics capture — or, on the other side, that the 2026 Ascend generation regressed exactly as CFR projects [70].
On whether the recovery is as strong as the revenue line: the evidence leans yes on revenue, more guardedly on profit (high confidence on the top line, medium on earnings quality). The controlling evidence is FY2025 operating profit up 22% to ¥96.9B with operating cash flow up 44% to ¥127.4B [64] and revenue back within ~¥1bn of the 2020 peak [63], which outweighs the one-off-flattered profit history [62] because operating profit and cash flow are precisely the lines a divestiture gain cannot inflate. The strongest surviving counter-argument: the operating margin, ~11%, is still below 2023's 14.8% [60], and cloud — the strategic on-ramp to AI — shrank 3.5% while Alibaba Cloud grew ~36% [65]. What would flip this reading: FY2026 operating margin below ~10% in the annual report Huawei publishes each March [12], or a second consecutive year of Huawei Cloud decline. Pre-mortem: if this looks wrong in two years, the bull-side miss is most likely the auto and AI-infrastructure mix re-rating margins faster than expected — the bear-side miss, a smartphone comeback that stalls on the cost of 7nm-era silicon.
On global leader versus China champion: the evidence leans China champion with one global stronghold (high confidence). The controlling evidence is ~71% of revenue from China, with the overseas share down from ~65% in 2010 to under 30% [71], and absence from the global smartphone top 5 [29], which outweighs the genuine 31% global telecom-equipment lead [24] because that lead sits in a market that shrank 11% in 2024 [11] while 11 EU states restrict Huawei 5G and Germany must strip it from 5G cores by end-2026 [32]. The strongest surviving counter-argument: Huawei still holds ~40% of the non-North-America market [26] — the wall is Western, not global. What would flip this reading: the overseas revenue share back above ~35% in an annual report, or a major EU state loosening rather than tightening its 5G restrictions. Pre-mortem: if this looks wrong in two years, the likeliest reason is Global-South telecom and digital-power growth quietly rebuilding the overseas mix — or, on the other side, a Chinese capex downturn proving that 71% concentration cuts both ways.
On whether the moats are a chosen edge or forced substitution: contested — and the deadlock is specific. On one side, ¥192.3B of R&D (21.8% of revenue; ¥1.38T over a decade) [33] and ~$630M of patent licensing that rivals pay regardless of politics [34]; on the other, estimates of up to ~$75B in state support [41], home-market reliance that doubles as dependence [42], and an OS ecosystem Huawei itself says needs 100,000 native apps to be mature [55]. The cleanest test is what happens where customers can freely choose, and it splits: Huawei Cloud shrank in a growing market [65] (the forced-substitution reading), while the Yinwang car unit fetched a ¥115B third-party price [20] (the chosen-edge reading). What would flip this to a lean: HarmonyOS native apps clearing Huawei's own 100,000 target [55], or Huawei Cloud returning to growth in the FY2026 report. Pre-mortem: if this looks wrong in two years, the bull-side miss would be treating sanction-era loyalty as a permanent moat — the bear-side miss, underestimating how long ¥1.38T of compounded R&D outlasts the subsidies that helped fund it.
The hardest ceiling is silicon. CFR estimates Huawei produced only ~5% of Nvidia's aggregate AI compute in 2025, with SMIC stuck at 7nm, and warns its next-generation 2026 chip could be less powerful than its best chip today because of yield and tooling limits [70]. The constraint is compounded by policy: the May 2025 US guidance makes using Ascend chips anywhere a claimed export-control violation, an extraterritorial reach that chills overseas buyers [53].
Concentration and ecosystem risks come next. Huawei is now ~71% reliant on China (¥615.3B in 2024), with the overseas share down from ~65% in 2010 to under 30% — little geographic cushion if the domestic economy slows [71]. HarmonyOS lock-in still leans on subsidy and a catalog far thinner than Android/iOS [55]. And the auto business depends on partners and faces internal cannibalization: roughly ¥136,000 reportedly flows to Huawei per AITO vehicle, but 问界's share of the HIMA alliance fell to ~72% as Huawei spread across five marques while some sub-brands stumbled [72][21].
Finally, governance and geopolitics. The rotating-chairman system is meant to institutionalize succession, but insiders concede an 80-year-old founder remains irreplaceable, and Western 5G bans plus National-Intelligence-Law concerns keep premium markets largely closed [6][57]. None of these are likely to be 'solved' inside the as-of window — they are the standing terms on which Huawei operates.
Both sides of the ledger
The strongest sourced case on each side, held to the same standard. The synthesis above states where this study thinks the weight of evidence sits; this ledger is what that lean has to survive.
The case for
+Many risks are ones Huawei has already partly absorbed: it survived the worst sanctions shock and rebuilt revenue to near-peak, so execution risk is demonstrated as manageable [63][73].
+Deep R&D and a full domestic stack give it more tools than any sanctioned peer to keep substituting around constraints [33][36].
+Diversification into power and auto reduces reliance on the most sanction-exposed lines [13].
The case against
−The chip ceiling is structural — ~5% of Nvidia's compute, stuck at 7nm, possibly regressing in 2026 [70].
−71% China reliance plus extraterritorial Ascend bans leave little diversification and a politically capped market [71][53].
−Ecosystem and partner dependence (subsidized HarmonyOS apps; HIMA cannibalization) plus founder-succession risk are unresolved [55][72][6].
Sources for this section
4 sources · en, zh · tiers shown. Full bibliography on the Sources page.
What this study is, how it was researched, what is disclosed versus estimated, and where it could be wrong.
74 sources32 Chinese · 43%As of June 2026
This is an independent study, built evidence-first. It presents the strongest evidence on both sides of each question — including a politically charged security debate — and then weighs it, stating each lean, its confidence level, and the tripwires that would flip it (the full weighing closes Risks & Challenges).
Method
Research proceeded by fanning out web searches in both English and Chinese and then directly fetching the underlying primary and reputable secondary sources; every URL cited here was opened and read, never cited from a snippet alone. Each claim was transcribed into a structured manifest that tags it with a source tier (1 primary, 2 reputable press, 3 forum/soft), a confidence level, and a stance (supporting, critical, or neutral), which is what lets the case-for and case-against in each section be sourced to the same standard. Because Huawei's home market is China, a substantial share of the 74 sources (32, about 43%) is in Chinese — Huawei's own annual reports, domestic press such as 观察者网, 第一财经, 证券时报, 华尔街见闻 and 新浪财经, founder and executive remarks (Ren Zhengfei, He Tingbo), and skeptical domestic commentary that rarely surfaces in English coverage. The load-bearing figures here are Huawei's disclosed annual-report numbers (revenue ¥862.1B in 2024, ¥880.9B in 2025; R&D 20–25% of revenue), the SMIC 7nm Mate-60 confirmation, the Ascend/CloudMatrix performance claims, and the telecom/cloud/AI-chip market shares.
Frameworks used
The study leads with an answer-first executive summary (Pyramid Principle), uses Porter's Five Forces to read the telecom-equipment and ICT industry, places Huawei against its rivals on a 2×2 integration-versus-market-access positioning map, benchmarks it through peer comparables, and shows revenue- and R&D-trajectory charts and a segment breakdown. These frameworks are used only to organize evidence and never to render a verdict; a conventional DCF or sum-of-the-parts valuation was deliberately skipped because Huawei is unlisted, has no public share price, and does not disclose the per-segment economics such a model would require.
Disclosed vs. estimated
Huawei is private but, unlike many private firms, publishes audited annual reports, so its group-level revenue, net profit and R&D are disclosures rather than guesses. The softer figures are third-party estimates and are labeled inline: market shares (Dell'Oro, Canalys, IDC, Bernstein, MERICS, CFR), AI-chip performance and output (SemiAnalysis, CFR), and the chip-process details (TechInsights). Currency conversions (RMB→USD at ~7.4, SEK and EUR for peers) are approximate and stated as such. Of the 74 sources, 18 are Tier 1 (primary), 40 Tier 2 (reputable press) and 16 Tier 3 (forum/soft), split 24 supporting / 24 critical / 26 neutral by stance.
🚧
Where this case study may be wrong
The security debate is genuinely unresolved. We present the US/Western structural concern and Huawei's denial side by side; we do not — and cannot — adjudicate whether espionage risk is real, and no public forensic proof of a deliberate backdoor was located.
Market shares and AI-chip figures are third-party estimates that vary by methodology (e.g. Nvidia's China AI-chip share is cited as both ~39% and ~55% depending on revenue vs units).
Profit is hard to interpret: 2021/2023 spikes were inflated by one-off Honor/xFusion gains, so year-on-year profit comparisons can mislead. We flag this rather than smooth it over.
Some figures are point-in-time: the chip roadmap (Ascend output, SMIC nodes), HarmonyOS app counts, and HIMA vehicle sales move fast and may be overtaken after the as-of date.
A few primary pages (Cisco's SEC filing, some government pages) bot-walled the automated checker and were verified manually; those rows are noted.
Neutrality & independence
Every section pairs the case for and the case against and attributes critical and positive claims to the same standard; where the study states a lean, the controlling evidence, the strongest surviving counter-argument and the tripwires that would flip it are shown, so the weighing is something you can audit rather than take on trust. Huawei is a uniquely politicized company, and domestic-Chinese and Western framings of it legitimately differ — both are presented rather than treating either as the neutral default. It is not affiliated with, sponsored by, or endorsed by Huawei or any other party named here, and it is not investment advice — no rating, price target, or recommendation to buy or sell any security. Everything is a point-in-time snapshot as of June 2026.
Bibliography
Sources
Every cited source was fetched during the research run. Tiers: 1 = primary/official, 2 = reputable press, 3 = forums/sentiment.
Huawei was founded in September 1987 in Shenzhen by Ren Zhengfei; sold its Honor smartphone brand in 2020 'to ensure survival' under US sanctions; had built ~70% of the world's 5G base stations.
Ren Zhengfei was a former PLA engineer; Huawei grew into the world's largest telecom-equipment maker; the Mate 60 Pro launched 6 Sep 2023 with an SMIC 7nm Kirin 9000S.
Huawei has ~208,000 employees, operates in 170+ countries serving 3 billion+ people, and runs five business groups (Carrier, Enterprise, Consumer, Cloud, Digital Power).
Huawei's holding company is ~99% owned by a 'trade union committee' and ~1% by Ren Zhengfei; academics Balding & Clarke argue the union structure could mean the firm is effectively state-controlled.
A rebuttal argues employee shares held via the union are not legally trade-union assets and that a separate Representatives' Commission, not the union, exercises shareholder rights — making state control unlikely.
Huawei runs a rotating-chairman system; Ren Zhengfei holds ~1.01% and a union representing 60,000+ employees holds 98.99% via virtual restricted shares; Ren is 80 and seen as irreplaceable.
Globally the cloud Big Three dominate: AWS 29%, Azure 20%, Google Cloud 13% (63% combined) of a $390B TTM market growing ~30% — Chinese hyperscalers are barely present globally.
Nvidia leads China's AI-accelerator market (~39%) with Huawei a close second; Bernstein projects Huawei could reach ~50% and Nvidia fall to ~8% by 2026 as controls tighten; domestic AI-chip demand compounding ~74%.
Open RAN — the main architectural substitute to integrated vendors — is projected to be under 5% of total RAN deployments by 2030, and RAN market concentration is higher now than before the O-RAN Alliance.
Global telecom-equipment revenue fell 11% in 2024 — the steepest annual decline in 20+ years — before recovering ~4% in 1H2025; carrier capex is a structural headwind.
Huawei reclaimed #1 in China smartphones in 2025 (46.7M units, 16.4%), narrowly ahead of Apple — driven by in-house Kirin silicon (Kirin 9030 in the Mate 80 Pro Max).
Huawei sold the Honor brand on 17 Nov 2020 to Shenzhen Zhixin (a buyer proposed by 30+ Honor agents/dealers), retaining no equity or management role, to let Honor escape the US chip and Google cut-off.
Huawei Digital Power FY2024 revenue CNY68.7B (+24.4%, record); inverter shipments ~176 GW in 2024 (>405 GW cumulative, world #1); 50,000+ EV charging stalls.
HarmonyOS Intelligent Mobility (鸿蒙智行 / HIMA) delivered 589,107 vehicles in 2025 (+32%), with the AITO/问界 (Seres) line ~70%+ of volume across its 'five realms' brands.
Ren Zhengfei signed a 'Huawei does not build cars' (华为不造车) decree in 2020 and extended it five years in March 2023; Huawei supplies smart-driving and cockpit tech via three modes (component, HI, and the 智选车/HIMA co-design model).
Huawei carved its car unit into Yinwang (引望) at a CNY115B valuation, selling 20% to Avatr and Seres for CNY23B cash; Yinwang was already profitable (H1 2024: CNY10.4B revenue, CNY2.2B net profit, 21.4% margin).
Within HIMA, brand performance is uneven: Luxeed/智界 fell from ~12,000 monthly units (Jan 2025) to 1,728 (Aug), and Stelato/享界 drew quality and delivery complaints amid an intensifying 2025 EV price war.
Huawei Cloud revenue fell 3.5% in 2025 (to ~US$4.6B) even as global cloud infrastructure grew ~29% and Alibaba Cloud grew ~36% — the one segment that shrank.
In 2024 Huawei held 31% of the global telecom-equipment market — #1 by far — ahead of Nokia (14%), Ericsson (13%) and ZTE (11%); it passed Nokia to become #1 even outside China.
In 1H2025 Huawei held 31% of global telecom equipment (Nokia 13%, Ericsson 12%, ZTE 10%, Cisco 4%); ~21% excluding China and ~40% of the non-North-America market.
Chinese coverage confirms the 1H2025 Dell'Oro shares and notes that in the non-North-America market Huawei holds ~40%, ZTE 14%, Nokia 12%, Ericsson 9%.
In 2024 RAN, the worldwide top five were Huawei, Ericsson, Nokia, ZTE and Samsung; outside China the order is Ericsson, Nokia, Huawei, Samsung, ZTE — Western markets still favour Ericsson/Nokia.
For full-year 2025 IDC ranked Huawei #1 in China smartphones (46.7M / 16.4%), ahead of Apple and vivo (both 16.2%); the China market was 285M units (−0.6%).
Stripped of Google Mobile Services since 2019, Huawei does not appear in the global smartphone top 5 (Apple ~19.7%, Samsung, Xiaomi, vivo, OPPO) — its comeback is China-only.
Council on Foreign Relations: US AI chips are ~5× more powerful than the best Chinese chips (forecast ~17× by 2027); SMIC cannot exceed 7nm; an HBM shortage caps Huawei at ~200,000–300,000 completed AI chips/year.
MERICS: Nvidia sold ~1M H20 chips in China in 2024 versus ~200,000 Ascend 910B from Huawei; the 910B is inference-leaning with weaker interconnect/memory, and Nvidia likely retains dominance near-term.
11 EU states have legal restrictions on Huawei/ZTE 5G; Germany must remove their gear from 5G cores by end-2026; the US runs a $1.9B 'rip-and-replace' program (carriers requested $5.6B).
Huawei FY2025 R&D was CNY192.3B (21.8% of revenue, a record), bringing cumulative 10-year R&D to over CNY1.382 trillion; ~114,000 employees (53.7%) work in R&D.
Ren Zhengfei's 2022 internal speech made survival the priority through 2025: 'make survival the main guideline... transmit the sense of crisis to everyone... comprehensively shrink and close edge businesses.'
Huawei built a self-controllable MetaERP on its own openEuler OS and GaussDB, replacing Oracle/SAP — described as 'the most extensive and complex project in Huawei's history' — part of a broader self-sufficiency (自主可控) push.
Ren Zhengfei defends Huawei's 'wolf culture' as sensitivity, teamwork and perseverance and denies an enforced 996, saying even junior staff cannot work excessive overtime.
ITIF argues US export controls 'backfired': they accelerated Huawei's own OS, forced replacement of 13,000+ components and 4,000+ redesigned boards, and cost US suppliers $33B+ in lost Huawei sales (2021–24).
HarmonyOS 5 terminal devices passed 17 million by Sept 2025; Huawei committed a CNY1 billion 'Tiangong' fund to grow the HarmonyOS AI ecosystem (10,000+ AI services, 5,000+ agents).
The central Western critique of Huawei's strategy is that its scale rests on state backing: CFR cites estimates of up to US$75B in cumulative state support, framing the 'national champion' moat as a subsidy-dependent one.
Self-reliance also concentrated Huawei on its home market: China rose to ~71% of revenue while the overseas share fell from ~65% (2010) to under 30% (2024) — a strategic dependence as much as a strength.
The US placed Huawei on the Entity List in May 2019, then in May 2020 amended the Foreign-Produced Direct Product Rule to bar foreign foundries (e.g. TSMC) from making Huawei-designed chips with US technology.
Chinese coverage of the May 2019 listing notes US firms could supply Huawei only by special license and that Huawei phones would lose Google's Android (GMS) and Qualcomm chips.
Huawei CFO Meng Wanzhou was detained in Vancouver on 1 Dec 2018 on a US fraud request tied to alleged Iran-sanctions evasion; she was released on 24 Sep 2021 under a deferred-prosecution agreement, after which China freed two detained Canadians.
TechInsights confirmed the Mate 60 Pro's Kirin 9000S was fabricated by SMIC on a 7nm (N+2) process without EUV — proof China could reach 7nm under sanctions, though the die is coarser than true 5nm.
HiSilicon's 'spare tire' (备胎) doctrine: on the night of the 2019 listing, chief He Tingbo wrote that all the backup chips Huawei had quietly developed had 'overnight, all turned into the real thing.'
SemiAnalysis found Huawei's CloudMatrix 384 (384 Ascend 910C) delivers ~300 BF16 PFLOPs — nearly double Nvidia's GB200 NVL72 — but at ~4.1× the power and ~2.5× worse power efficiency per FLOP.
Chinese coverage frames Huawei's AI approach as deliberate doctrine — 'use mathematics to supplement physics, non-Moore to supplement Moore's Law, systems to supplement single points' — to offset the per-chip gap.
Ren Zhengfei conceded in June 2025 that the US 'exaggerated' Huawei's achievements and that its single chips still lag the US 'by one generation,' compensated for by clustering and mathematics.
Ren Zhengfei has flatly denied espionage and backdoors: 'we absolutely never install backdoors. Even if we were required by Chinese law, we would firmly reject that,' and called Meng's arrest 'politically motivated.'
Western security concerns predate the trade war: Australia banned Huawei from 5G in 2018, citing the CCP's tightening grip on Chinese tech firms and telecom espionage vulnerability.
On 13 May 2025 US BIS guidance declared that using Huawei Ascend 910B/910C/910D chips anywhere in the world violates US export controls; the Jan-2025 Nvidia H20 ban (reversed July 2025) had briefly boosted Ascend demand.
HarmonyOS NEXT ('pure-blood' 纯血鸿蒙), launched 22 Oct 2024, has no Android/AOSP layer and cannot run Android apps; the native app/meta-service count grew from ~10,000 (Sep 2024) toward 300,000+ (Nov 2025).
Huawei's own stated 2025 goal was to 'sprint to 100,000 native HarmonyOS apps' — implicitly conceding how much of the app ecosystem still has to be rebuilt to escape Android dependence.
Western distrust persists: in March 2025 Belgian police raided Huawei's Brussels EU office in a bribery probe alleging payments to members of the European Parliament.
US policymakers view Huawei as a potential commercial extension of the CCP, citing China's 2017 National Intelligence Law (firms must 'support, assist and cooperate with' intelligence work) and up to $75B in estimated state support.
On 13 February 2020 the US Department of Justice filed a 16-count superseding indictment charging Huawei Technologies and two US subsidiaries with conspiracy to violate the Racketeer Influenced and Corrupt Organizations Act (RICO) and conspiracy to steal trade secrets, and alleging Huawei's involvement in business and technology projects in Iran and North Korea in violation of US sanctions.
The 2024 net-profit drop was driven by 'other net income' collapsing from CNY62.7B (2023) to CNY8.3B (2024); operating margin fell to 9.2% from 14.8% — a quality-of-earnings concern.
FY2021 revenue collapsed −28.6% to CNY636.8B after the Honor sale and sanctions; reported net profit +75.9% to CNY113.7B was flattered by a one-off Honor/xFusion divestiture gain (~half of operating profit).
FY2025 revenue CNY880.9B (+2.2%) was just CNY1.05B short of the 2020 all-time peak (~CNY891.4B); net profit recovered +8.6% to CNY68B and operating margin rose to ~11%.
FY2025 operating profit rose 22% to CNY96.9B (11.0% margin) and operating cash flow rose 44% to CNY127.4B; management called AI 'possibly the last technological revolution in human history.'
Huawei Cloud revenue fell 3.5% in 2025 (to ~US$4.6B) even as global cloud infrastructure grew ~29% and Alibaba Cloud grew ~36% — the only Huawei segment to shrink.
Cisco fiscal-year 2025 (ended 26 Jul 2025): full-year revenue $56.7B; it leads enterprise IP networking rather than mobile RAN, and is a ~4% telecom-equipment player.
CFR estimates Huawei produced only ~5% of Nvidia's aggregate AI compute in 2025, with SMIC stuck at 7nm; its next-generation 2026 chip could be less powerful than its best chip today.
Huawei is heavily reliant on its home market: China was CNY615.3B (~71% of 2024 revenue), with the overseas share falling from ~65% (2010) to under 30% (2024).
The auto business depends on partners and faces internal cannibalization: ~CNY136,000 reportedly flows to Huawei per AITO vehicle, and 问界's share of HIMA fell from the absolute mainstay in 2023 to ~72% in 2025 as Huawei spread across five marques.
Mitigant: Huawei has already absorbed the worst sanctions shock and rebuilt revenue to within ~¥1bn of its 2020 peak by 2025 — evidence its execution risk is, so far, manageable.
Cross-checked at build time by an automated link checker; a few primary sources may be paywalled or bot-walled and were verified manually. See Methodology & Limits.