Friday, September 11, 2026
AI 인프라 · 뉴스 & 분석
정책리포트
정책 · 리포트

수출 통제로 베이징의 진전을 늦추려 했던 것과 달리, 미국의 반도체 대중국 수출은 130% 급증했다.

중국 내 AI 컴퓨팅 인프라 구축을 가속화하고 더 엄격한 규제 감독을 초래할 수 있는 공백이나 무역 동향의 변화를 강조한다.
업계 전문지Slicast · September 9, 2026 · 미국 · 출처: Martin Cid Magazine
중요도 75

China’s semiconductor exports surged 129.8% year-over-year—the sharpest annual increase since the current U.S. export-control framework took effect in 2022. The surge accompanied a broader trade surplus of $119.1 billion, up from $112.5 billion in July and on track for a full-year record. Published by China’s General Administration of Customs, the data spans the entire range of exported chips: logic circuits, power management semiconductors, memory, and automotive-grade components integrated into virtually every major Chinese electric vehicle.

This scale directly challenges a foundational premise of U.S. trade and technology policy. Controls introduced by the Biden administration in 2022, expanded in 2023, and further extended under the current administration aimed to restrict China’s access to equipment required for manufacturing advanced chips—specifically those below the 16-nanometer threshold needed for AI training accelerators and next-generation processors. The underlying assumption was straightforward: without frontier hardware, China’s tech sector would stall. August’s figures test that premise. They do not suggest Beijing has solved the advanced-node challenge—at least not at scale—but rather demonstrate that the legacy tier, largely bypassed by the controls, has evolved into a formidable production ecosystem.

China’s automobile exports also climbed 43% year-over-year in August, heavily driven by semiconductor-intensive electric vehicle drivetrains. Shipments to the United States rose 34.4% to $42.5 billion for the month, widening the bilateral trade imbalance to approximately $29 billion. Southeast Asia absorbed a 30.2% increase in Chinese shipments, while Latin America saw a 17.5% rise. Both regions represent manufacturing and consumer markets increasingly reliant on Chinese-supplied components.

The year-over-year comparison requires context. August 2025 represented a weak period for Chinese chip output due to supply-chain constraints, making the 2026 baseline appear unusually strong in retrospect. Additionally, the customs data does not differentiate between chips manufactured entirely within China and components assembled in bonded processing zones using foreign-sourced wafers. While Chi Lo of BNP Paribas described China as “very competitive in its tech goods exports,” analysts specializing in advanced semiconductors note that leading-edge production—directly targeted by U.S. controls—remains bottlenecked by equipment restrictions. The structural gap in the control regime lies squarely in the legacy tier: below the frontier, yet above basic commodity chips.

The economic impact is felt most acutely outside advanced-node fabrication plants. Workers and investors in the industrial and automotive sectors—where Texas Instruments, Infineon Technologies, and STMicroelectronics compete primarily on cost and delivery reliability—are bearing the brunt. Meanwhile, TSMC’s Arizona facility and Intel’s Ohio plant are scaling advanced-node capacity, largely insulated from the same pricing pressures. Yet European automakers sourcing components globally are already facing direct competition from Chinese-designed chips in Southeast Asian showrooms. This dynamic represents a consequence that aggregate trade statistics only partially capture.

The next key data release comes in early October, when China publishes its September trade figures. Concurrently, the U.S. Commerce Department is expected to complete its next review of the Entity List—which dictates which Chinese firms qualify for advanced chipmaking equipment—before year-end. August’s semiconductor export data will carry substantial weight in that assessment.

원문 보기