When Huawei’s HiSilicon division taped out the Kirin 9000 — a 5nm mobile processor manufactured by TSMC in 2020 — it marked a moment of quiet reckoning for the global semiconductor industry. Here was a Chinese company designing chips that could rival Apple’s A-series and Qualcomm’s Snapdragon at the cutting edge of process technology. Then US export controls cut HiSilicon off from its foundry partner, and the world paid attention to a question it had underestimated: just how capable is China’s chip design industry, and what happens when it is forced to accelerate?
The answer, playing out across laboratories in Shanghai, Shenzhen, Beijing, and Chengdu, is more consequential for global technology supply chains than most business leaders outside the industry appreciate. China’s integrated circuit design sector — the upstream discipline of creating the blueprints for chips before any silicon is ever etched — has grown from a cottage industry into a $50 billion ecosystem in less than two decades. Understanding who the key players are, what they can actually build, and what gaps remain is essential intelligence for any company with exposure to hardware, consumer electronics, telecommunications, or advanced manufacturing.
The Architecture of China’s Chip Design Ecosystem
China’s semiconductor design industry operates through three distinct tiers, each serving different market segments and operating under different competitive dynamics.
The first tier is dominated by the largest and most strategic players: HiSilicon (a subsidiary of Huawei), Unisoc (formerly Spreadtrum, now controlled by Tsinghua Unigroup), and Cambricon Technologies. These companies design chips for smartphones, artificial intelligence acceleration, and 5G infrastructure — the most technically demanding segments of the market.
The second tier includes a dense cluster of application-specific integrated circuit (ASIC) designers and system-on-chip (SoC) developers targeting IoT devices, automotive electronics, industrial control systems, and smart home products. Companies like Allwinner Technology, Rockchip Electronics, and Zhaoxin (a joint venture with Via Technologies) operate here.
The third tier is the most dynamic: over 3,000 early-stage and mid-size fabless design houses that have emerged since the early 2010s, many spun out of university research programs or founded by engineers returning from positions at Qualcomm, Broadcom, Intel, and Samsung. The Ministry of Industry and Information Technology estimated in its 2025 integrated circuit industry report that China now hosts more registered IC design enterprises than any other country, though the gap in average design complexity compared to leading US, South Korean, and Taiwanese firms remains significant.
HiSilicon: The Benchmark That Reset Expectations
No discussion of Chinese chip design begins anywhere other than HiSilicon. Founded in 2004 as Huawei’s internal semiconductor arm, HiSilicon spent its first decade quietly building competency in baseband modems and network processing chips for Huawei’s own telecom equipment. Most industry analysts dismissed it as a captive design shop with no competitive relevance outside its parent company.
That assessment was wrong. By 2018, HiSilicon’s Kirin 980 — manufactured on TSMC’s 7nm node — was competing directly with Qualcomm’s Snapdragon 845 in independent benchmarks. By 2020, the Kirin 9000 at 5nm was among the most advanced mobile processors in the world. HiSilicon had achieved in 16 years what most industry experts believed would require decades: genuine design parity with the top tier of global fabless companies.
The US Commerce Department’s Entity List restrictions, imposed in 2020, ended HiSilicon’s ability to procure manufacturing from TSMC or any other non-Chinese foundry using US technology. The practical consequence was that Huawei’s smartphone division collapsed from a global market share of roughly 20% to under 4% within 18 months. HiSilicon’s workforce largely remained intact, however, and Huawei has continued investing in its design capabilities — now oriented toward what can be manufactured domestically at SMIC’s available process nodes, primarily 7nm and above using older lithography equipment.
Unisoc and the Mass Market Battleground
While HiSilicon captures strategic attention, Unisoc operates in the segment that actually determines semiconductor volume: entry-level and mid-range mobile devices. Unisoc’s T series SoCs power hundreds of millions of smartphones priced below $200 — a market dominated by Chinese OEMs selling primarily in Africa, South Asia, and Southeast Asia.
Founded through the 2013 merger of Spreadtrum Communications and RDA Microelectronics under Tsinghua Unigroup, Unisoc has progressively moved up the performance ladder. Its T760 chip, introduced in 2022 on a 6nm process, brought genuine 5G connectivity to sub-$150 handsets — a segment where Qualcomm’s pricing had previously made 5G economics unworkable for budget device makers. By the end of 2025, Unisoc claimed to have shipped chips powering over 1.2 billion connected devices cumulatively.
The strategic importance of Unisoc extends beyond handsets. It has aggressively expanded into IoT, automotive-grade chips, and satellite communication modules — the latter becoming a high-priority segment following China’s BeiDou navigation satellite network achieving full global operational capability in 2020.
Cambricon and the AI Chip Race
Cambricon Technologies represents a different kind of ambition. Founded in 2016 by Chen Tianshi and Chen Yunji — brothers and researchers from the Chinese Academy of Sciences — Cambricon was among the first companies globally to develop a neural processing unit (NPU) architecture designed specifically for deep learning inference. Its early IP was licensed to Huawei, where it formed the basis of HiSilicon’s Kirin NPU blocks.
Cambricon went public on Shanghai’s STAR Market in 2020 at a valuation of approximately 29 billion RMB ($4 billion). Its MLU series of data center AI accelerators targets the server-side inference market where Nvidia’s H100 has faced export restrictions into China since late 2022. This export control dynamic — covered in detail in our analysis of China’s technology supply chain leverage — has paradoxically accelerated domestic adoption of Chinese AI chips. Baidu’s ERNIE, Alibaba’s Tongyi Qianwen, and dozens of smaller large language model deployments now run at least partially on domestic accelerators including Cambricon’s MLU590.
Cambricon is not alone. Biren Technology, Moore Threads, Enflame, and Iluvatar CoreX have all emerged as serious contenders in the GPU-adjacent AI accelerator space, collectively attracting billions in venture capital from investors including Sequoia China (now HongShan), Hillhouse Capital, and strategic investment vehicles connected to state-backed funds.
The EDA Software Gap: China’s Critical Dependency
If manufacturing is the headline constraint on China’s semiconductor ambitions, electronic design automation (EDA) software is the structural constraint that attracts far less attention but may prove more durable.
EDA tools — the software used to design, simulate, verify, and optimize semiconductor layouts — are dominated globally by three US companies: Synopsys, Cadence Design Systems, and Siemens EDA (formerly Mentor Graphics). All three operate under US export control jurisdiction. Chinese chip designers working on advanced nodes below 10nm are heavily dependent on these tools, and their continued access is subject to ongoing policy uncertainty.
China has recognized this vulnerability for years. The Semiconductor Industry Association has documented that China’s Big Fund (National Integrated Circuit Industry Investment Fund) has channeled significant capital toward domestic EDA development. Companies including Empyrean Technology (EDA) and Primarius Technologies have emerged as credible alternatives for specific design flows, particularly in older-node analog and mixed-signal design. But no Chinese EDA company has yet delivered a full-stack solution competitive with Synopsys or Cadence for advanced digital design at sub-7nm nodes.
This gap matters enormously for foreign companies evaluating Chinese chip partners. A Chinese fabless company designing on mature nodes for automotive, industrial, or consumer IoT applications may operate with genuine design independence. One claiming to design cutting-edge mobile or AI processors is, for now, still dependent on Western software infrastructure in ways that create both business continuity risk and potential compliance exposure for foreign customers.
What This Means for Foreign Technology Companies
The growth of China’s chip design industry creates a complex matrix of opportunity and risk for non-Chinese technology companies.
On the opportunity side: Chinese fabless companies are viable, cost-competitive suppliers for a wide range of mature-node products. Automotive sensors, industrial microcontrollers, power management ICs, Wi-Fi and Bluetooth modules, and display drivers are categories where companies like Chipsea Technologies, ICCAD, and Goodix Technology offer products at prices that are difficult to match with fully Western supply chains. Foreign OEMs sourcing from Chinese chip designers for non-sensitive applications face no inherent regulatory barrier, though prudent practice requires vendor evaluation against current US Bureau of Industry and Security entity lists and any applicable EU, Japanese, or UK screening requirements.
On the risk side: companies that have become dependent on Chinese chip suppliers for more advanced applications face concentration risk that has crystallized uncomfortably in recent years. The whipsaw of Huawei’s collapse as a smartphone platform — from a dominant partner position for many European and Asian OEMs to an effectively sanctioned entity in roughly 12 months — is the case study every supply chain professional should study.
The US Bureau of Industry and Security has continued to refine and expand semiconductor-related export controls, most recently through the October 2022 and October 2023 rule updates. Foreign companies selling into US markets, using US-origin technology in their products, or operating within US-linked financial systems need legal counsel experienced in Export Administration Regulations before deepening semiconductor supply chain relationships with Chinese design companies on the Entity List.
For Chinese companies, the strategic imperative is equally clear: the path to genuine design independence runs through domestic EDA, domestic IP libraries, and increased reliance on SMIC and emerging domestic foundries — a multi-decade project that government policy and capital allocation are actively supporting. Understanding this trajectory matters for any Western company evaluating China’s technology ecosystem as either a market, a supply base, or a competitive threat.
The Bottom Line
China’s chip design industry has graduated from a domestic supply supplement to a genuine competitive force across large segments of the global semiconductor market. HiSilicon demonstrated what is designable; Unisoc demonstrated what is scalable; Cambricon and its peers are demonstrating what is possible in AI silicon. The EDA dependency and advanced manufacturing constraints are real and consequential, but they are not static — they are closing at a pace that consistently surprises outside observers.
For business leaders, the practical implication is precision: neither blanket avoidance nor uncritical engagement serves well in this landscape. Mapping your specific product’s technical requirements against the actual capability profile of Chinese design partners — and understanding the regulatory environment that governs those relationships on both sides of the Pacific — is the work that separates companies that navigate this era successfully from those that get caught off-guard.
The semiconductor design race between the US and China will be one of the defining economic competitions of the next decade. It is already reshaping the economics of every industry that depends on silicon.