French Expert: China’s Unconventional Strategy to Break U.S. Chip Sanctions: Not Pursuing the Most Advanced, but Perfecting Mature Technologies

Moreover, under U.S. sanctions, what matters most for Beijing is not producing “perfect” semiconductors, but manufacturing domestic chips that can be continuously improved over the coming years—this is the analysis published by Robin Rivaton, CEO of Stonal, in the economic column of the French weekly *L'Express*.

In his column article, Rivaton writes that computer enthusiasts love “lifting the hood” to see how things work. One such enthusiast made a revealing discovery. In late May, a user inspecting a Corsair memory module accidentally found it contained Chinese-made chips—not the usual Korean or American silicon. The memory module bore the label CXMT (Longsys Memory), which many had mistakenly assumed (incorrectly) served only the domestic Chinese market.

This incident precisely encapsulates the strategy explicitly stated in China’s "14th Five-Year Plan" introduced in March this year. The opening intent regarding integrated circuits can be roughly translated as: “Refine and perfect mature process nodes.” The outside world originally expected China to achieve some dramatic technological breakthrough to circumvent U.S. sanctions—but instead, China has taken the opposite approach: focusing on mature, even older processes, constantly refining, stacking, and pushing them to their limits.

The article continues: The most striking example comes from photolithography—the step of projecting circuit patterns onto silicon. Due to U.S. sanctions, China cannot obtain Dutch ASML’s extreme ultraviolet (EUV) lithography machines, which can pattern the finest features in a single exposure. In June 2022, Huawei filed a patent, made public by the end of 2025. Its solution? No longer relying on a single ultra-high-precision exposure; instead, using older equipment to repeatedly expose and overlay the same pattern up to six or seven times, thereby narrowing the lines to nearly advanced-chip levels. From principle to factory practice, this is replacing optical precision with brute force.

The patent indeed exists, but industrialization remains in its infancy. In reality, each additional exposure amplifies alignment errors and reduces yield—the proportion of final chips that are actually usable. SMIC’s yield on its most advanced processes still lags significantly behind Taiwan’s industry leader, TSMC. However, compared to Beijing’s obsession with self-reliance, performance loss, increased costs, and energy waste seem less significant. For Beijing, the goal is not to produce perfect chips, but to create a domestically made chip capable of continuous improvement year after year.

Building New Walls with Old Bricks

The article notes another strategic evidence from a presentation held on May 25 in Shanghai at an IEEE (Institute of Electrical and Electronics Engineers) conference. Huawei’s semiconductor lead proposed a mathematical theory: the metric for progress is no longer transistor size, but circuit response speed. The core idea involves stacking functional modules across multiple layers and then precisely bonding them together. Based on this, Huawei claims it can operate without EUV equipment. Analysts argue this is less a revolution and more a clever rebranding of existing 3D integration technology. Yet this perfectly reflects China’s approach: repurposing available “bricks” into an industrial pathway.

In the sensitive memory market, a similar trend is unfolding on a larger scale. Demand for high-bandwidth memory (HBM), essential for AI chips, has surged, almost draining resources from other memory segments. Samsung, SK Hynix, and Micron have shifted focus to the most profitable niches, leaving gaps in entry-level and mid-range markets. Longsys Memory has seized these gaps: starting with low-end products like Corsair modules, it is now advancing toward higher-end applications.

Early June saw Beijing urging domestic manufacturers to launch a “full mobilization” in the high-end memory sector. The logic hasn’t changed: accept a less efficient product initially, then gradually close the gap through massive output and production learning. Yet this dedication to mature processes tells only half the story. The five-year plan demanding “refinement of existing technologies” simultaneously funds exploration of all possible alternatives to silicon—such as 3D stacking, integration of electronics and photonics, and new materials like gallium oxide and diamond.

The article concludes by warning that China is not choosing between “catching up” and “surpassing.” It does both simultaneously: extracting maximum value from mature technologies while placing “options” on every potential alternative technology.

In practice, this strategy has already emerged in the nuclear energy sector—China, like Russia, is among the few countries actively exploring most fourth-generation nuclear technology pathways in parallel.

According to Rivaton, the West’s biggest misjudgment lies in believing the semiconductor war is solely about the fineness of process node dimensions.

After reading this economic column article from *L'Express*, we realize that earlier discussions about China’s chips often focused on whether it could break through 7nm, 5nm, 3nm and other advanced processes. Now, increasing analysis is turning to another question: Can China build a complete industrial ecosystem independent of external supply chains, capable of continuous iteration and upgrading?

Translation Note: Robin Rivaton is currently CEO of Stonal and a member of the Scientific Committee at France’s Fondation pour l’innovation politique.

Source: rfi

Original: toutiao.com/article/1867357527735308/

Disclaimer: The views expressed in this article are those of the author.