The US-China Tech War Escalates: How Export Controls Are Reshaping the Global Technology Order
The technology conflict between the United States and China is no longer a trade dispute dressed in tariff language — it has become a full-spectrum strategic competition to control the foundational technologies that will determine economic and military power for the next generation. The Biden-era export controls on advanced semiconductors and chipmaking equipment, far from being rolled back under the Trump administration, have been extended, tightened, and supplemented with new restrictions on AI software, quantum computing technology, and the advanced manufacturing equipment that produces frontier technology hardware. China's response — accelerated domestic investment in semiconductor manufacturing, AI infrastructure, and the full stack of technologies subject to restriction — has transformed what began as defensive Western industrial policy into an arms race dynamic where both sides are racing to achieve capability independence from the other. The global technology industry is being forced to navigate a bifurcating world in which the same products, services, and supply chains that connected the global economy for three decades are now weapons in a strategic competition that has no obvious resolution timeline.
What the Export Controls Actually Cover — and What They Are Trying to Achieve
The US export control architecture targeting China's advanced technology capabilities has been built layer by layer since 2019, with each successive round expanding the scope of restricted technologies and the extraterritorial reach of US enforcement. The October 2022 rules — which restricted the sale of advanced AI training chips (above certain performance thresholds), chipmaking equipment capable of producing sub-14nm chips, and the software and technical expertise to operate them — were the most consequential single technology policy action since the Cold War-era COCOM controls on technology exports to the Soviet bloc. The October 2023 expansion closed the loopholes that had allowed modified chip variants to reach Chinese customers; the 2024 controls added more chipmaking equipment suppliers, more countries to the restricted list, and tightened the rules around third-country re-export that China had been exploiting through intermediaries in Singapore, Malaysia, and the Middle East. The stated objective of the controls is to prevent China from using American technology to develop military capabilities — specifically, the AI-enabled autonomous weapons, surveillance infrastructure, and cyberattack tools that advanced compute enables. The operational objective is to widen the technology capability gap between the US and China by denying China access to the hardware at the technology frontier, buying time for American-led technology development to compound its advantage before Chinese domestic alternatives mature.
Whether this strategy is succeeding is contested and deeply uncertain. The evidence for partial success is real: China's leading semiconductor manufacturer SMIC cannot yet manufacture chips below 7nm at commercial yield rates, compared to TSMC's volume production at 3nm and development of 2nm technology. Huawei's Ascend AI chips — manufactured by SMIC at 7nm using deep ultraviolet lithography rather than the extreme ultraviolet lithography that TSMC uses — perform at approximately 50–70% of H100 efficiency on transformer training workloads. The evidence for the strategy's limitations is equally real: China is spending at historically unprecedented scale to close the gap. The Chinese government's semiconductor investment programmes — the National Integrated Circuit Industry Investment Fund's USD 47.5 billion third tranche announced in 2024, provincial programmes that collectively add tens of billions more, and mandatory state-backed research spending — represent the most concentrated technology development investment programme since the Manhattan Project.
How Industry Is Adapting — and Who Is Paying the Price
The global technology industry's adaptation to the US-China export control regime is creating a structural bifurcation that will take years to fully manifest but is already visible in the strategic decisions of the world's most influential technology companies. NVIDIA — whose H100, H200, and Blackwell AI chips are the primary target of chip export controls — has developed China-specific chip variants (H20, L20) that comply with export control performance thresholds while delivering meaningfully lower AI training performance than the restricted products. The H20 chip, which Chinese hyperscalers including Alibaba, Baidu, Tencent, and ByteDance have purchased in substantial volumes, represents approximately 30–40% of H100 performance — sufficient for inference workloads at current model sizes but inadequate for frontier model training. ASML, the Dutch monopolist in extreme ultraviolet lithography equipment, has been prohibited from exporting EUV systems to China since 2019 and has since been prohibited from servicing the deep ultraviolet systems it previously sold to Chinese customers — a decision with profound implications for ASML's Chinese revenue (approximately 15% of total sales) and for China's ability to maintain existing chip production equipment.
The price of bifurcation is being paid unevenly. American semiconductor equipment companies — Applied Materials, Lam Research, KLA — have collectively written off several billion dollars in Chinese revenue since the 2022 controls, with China previously representing 30–35% of each company's annual sales. The financial impact has been partially offset by stronger demand from non-China geographies as the global semiconductor capacity expansion triggered by supply chain disruptions continues, but the structural reduction in China-accessible market is permanent under the current policy architecture. Taiwan Semiconductor Manufacturing Company — the pivotal company in any analysis of US-China technology conflict given its manufacturing of chips designed by both American and Chinese companies — is navigating an existential tension between its American customer relationships, its Taiwanese sovereign context, and its historical Chinese business that has now been substantially reduced by export control compliance requirements.
China's Domestic Technology Build-Out: Progress and Limits
China's response to the export control regime is the largest state-directed technology development programme in modern economic history, and its progress is simultaneously more significant than Western commentary typically acknowledges and more limited than Chinese state media claims. In AI software — the large language models, computer vision systems, and multimodal AI platforms that run on restricted hardware — Chinese companies have achieved genuine capability parity with Western leaders across most commercial applications. Baidu's Ernie, Alibaba's Qwen, Tencent's Hunyuan, and ByteDance's Doubao are competitive with GPT-4 class models on Chinese-language benchmarks and increasingly competitive on English-language tasks. In AI hardware — the chips required to train frontier models — the gap is more significant and harder to close: NVIDIA's Blackwell architecture achieves performance on transformer training workloads that China's domestic alternatives cannot match on the same task at comparable energy efficiency, and the manufacturing technology gap makes it unlikely that Chinese foundries will produce equivalent hardware within a five-year horizon under the current export control regime.
The Third-Country Dimension: Where the Controls Leak
The effectiveness of US export controls depends on the compliance of third countries — particularly in Southeast Asia, the Middle East, and Europe — whose companies can legally import restricted technology and from whose territory chips and equipment have historically been re-exported to China. The US Bureau of Industry and Security's enforcement actions against companies in Singapore, the UAE, Turkey, and Malaysia in 2023–2025 illustrate the scope of the third-country problem: dozens of companies have been added to entity lists for facilitating the export of controlled technology to China, and the enforcement resources dedicated to investigating and prosecuting these cases have expanded significantly. The July 2023 Disruptive Technology Strike Force — a joint DOJ-Commerce Department initiative — has brought criminal prosecutions against technology procurement networks operating through multiple intermediary jurisdictions. Despite these efforts, estimates from semiconductor industry analysts suggest that billions of dollars worth of restricted chips have reached Chinese customers through third-country channels, with Singapore and the UAE having been the most frequently used transit points before tightened enforcement began redirecting flows to less monitored jurisdictions.
The Long Game: What the Bifurcation Means for Technology Markets
The US-China tech war is creating a technology world that will increasingly operate on two parallel tracks — a Western-dominated ecosystem anchored in American chip design, American software standards, and the allied manufacturing capabilities of Taiwan, South Korea, Japan, and the Netherlands; and a Chinese domestic ecosystem that will be less capable at the frontier in hardware but increasingly self-sufficient in software, applications, and the technology stack required for Chinese domestic deployment. The commercial consequences are profound: global technology companies will need to maintain separate product lines, separate data governance structures, and in some cases separate engineering teams to serve these two markets simultaneously. The geopolitical consequences are more profound still: technology capability is converging with security capability in ways that make the bifurcation of the technology ecosystem equivalent to the bifurcation of the global security order. The US-China tech war has no obvious endpoint — each side's strategic objectives require capability independence that the other side's investment in domestic alternatives makes perpetually out of reach — and its management will define the geopolitical and commercial landscape of technology for the decade ahead.