China Semiconductor Self-Sufficiency and the 70% Wafer Question

• China’s semiconductor self-sufficiency ambitions are tested by new silicon wafer target.
• Jensen Huang says Nvidia’s China market share has hit zero.
• Beijing moves to fill the gap.

China’s semiconductor self-sufficiency drive has a new headline number. According to an exclusive report by Nikkei Asia published on May 5, China is targeting more than 70% of silicon wafers used by its chip-makers to be sourced domestically by the end of this year – one of the most aggressive localisation milestones the country has set for its chip supply chain to date.

The target, described by Nikkei as an internal government directive rather than a formal public announcement, centres on silicon wafers – the foundational substrate on which virtually every chip is built. It is worth noting upfront: this is a leaked or internal goal, not an officially declared government policy.

Additionally, China has a habit of announcing ambitious self-sufficiency targets, but has a history of falling short.

A 70% target with precedent

Made in China 2025, launched a decade ago, set the same headline figure: 70% semiconductor self-sufficiency as a goal for 2025. That target was missed by a significant margin. The broader semiconductor self-sufficiency rate, which covers the full stack from design to fabrication to materials, was estimated at around 50% by TrendForce heading into last year, well behind the original ambition.

But the 2026 wafer target is narrower and more specific, and that specificity is arguably why it is more credible than its predecessor. Silicon wafers are one area where China has made measurable, documented progress. The global market is dominated by five players – Japan’s Shin-Etsu Chemical and SUMCO, Taiwan’s GlobalWafers, Germany’s Siltronic, and South Korea’s SK Siltron – who between them held over 80% of 12-inch wafer capacity as recently as 2020.

China’s share of that market sat at roughly 3% the same year. By 2025, according to Bernstein analysis cited by CrASIAthat share had climbed to approximately 28%. By 2026, it could reach 32%. The company leading this expansion is Xi’an Eswin Material Technology, the wafer-making arm of Beijing ESWIN Technology Group.

According to TechNewsEswin is simultaneously building out facilities in Xi’an and Wuhan, targeting 1.2 million 12-inch wafer starts per month in combined capacity this year alone, enough to serve roughly 40% of China’s domestic 12-inch wafer demand and push Eswin’s own global market share above 10%.

Its customer list already includes SMIC, Hua Hong, CXMT, and YMTC on the domestic side, with UMC and GlobalFoundries purchasing its products for mature-node work internationally. Samsung and SK Hynix are currently conducting product validation.

None of that happened quietly. Eswin is selling 12-inch wafers at prices that distributors describe as dramatically below global production costs, some below US$40 per wafer, against a market norm of US$60 – 80 and a production cost floor, for most global leaders, above US$50. SEMI’s Clark Tseng put it plainly in an assessment last year: “It is only a matter of time before Chinese wafer suppliers catch up with global rivals, and they have many Chinese customers to hone their capabilities.”

What Jensen Huang said five days ago changes the context

The wafer target does not exist in isolation. It sits inside a much larger, rapidly shifting picture, one that Nvidia CEO Jensen Huang described with unusual bluntness last Wednesday during an interview on the Special Competitive Studies Project’s Memos to the President programme.

“Nvidia had, you know, called it 90-some odd per cent of the world’s market share,” Huang said. “Today, in China, we have now dropped to zero.”

The collapse unfolded in stages. US export controls, beginning in October 2022, progressively restricted what American chip companies could sell into China. Nvidia engineered its H20 chip specifically to comply with those thresholds and continue serving Chinese hyperscalers. Then, in April 2025, the Trump administration required a licence even for H20 exports. Nvidia took a US$4.5 billion charge on H20 inventory it could no longer move.

A US export licence to ship the more capable H200 chip to China was granted in February 2026, but Nvidia has generated no revenue from it. Commerce Secretary Howard Lutnick confirmed the chips have not been delivered. Beijing has been directing domestic companies to buy locally, and the H200 approval has effectively stalled at the border.

Huang’s view on the policy outcome was direct: “Conceding an entire market the size of China probably doesn’t make a lot of strategic sense. I think that has already largely backfired.” He pushed back on the notion that cutting off hardware access caps China’s AI capability. “They have cheaper energy. They have incredible talent. The number of AI researchers in China is quite extraordinary – it’s one of their national treasures.”

Nvidia’s own annual report framed the commercial stakes clearly: “As of the end of fiscal year 2026, we were effectively foreclosed from competing in China’s data centre computing market, and our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems that can challenge us globally.”

The last clause is pivotal. Exclusion from the market has redirected investment into domestic alternatives. Huawei’s AI chip revenue is projected to surge 60% year-on-year to US$12 billion in 2026, and Cambricon posted record profits of 1.04 billion yuan in the first half of 2025 on revenue that surged over 4,000% year-on-year. Goldman Sachs expects Cambricon’s AI chip shipments to grow from 143,000 units in 2025 to 2.1 million by 2030.

Where self-sufficiency wins

The honest accounting of China’s semiconductor self-sufficiency requires separating what is working from what remains structurally out of reach. On materials and mature-node equipment, progress is real. Three Chinese chip equipment manufacturers ranked among the world’s top 20 by sales volume for the first time in 2025, according to Japanese research firm Global Net.

ACM Research, AMEC, and Naura Technology Group are generating revenue and scale. Domestic photoresist production, long considered a near-impossible gap to close, has recorded commercial shipments.

On advanced logic fabrication and lithography, the picture is considerably more constrained. SMIC continues to rely on ASML deep ultraviolet systems for its 7nm output, and the Dutch government’s EUV export ban remains in place. A prototype domestic EUV machine is reportedly in testing, with functional chips targeted by 2028 and a 2030 timeline considered more realistic by outside analysts.

China’s new semiconductor plan, reported by Electronics Weekly in March, has set an 80% self-sufficiency target and a 2030 goal of building and validating a 7nm production line using entirely domestic equipment.

The wafer target, then, is the part of China’s semiconductor self-sufficiency where the numbers are most defensible. Wafers are not advanced logic. Achieving 70% domestic use of silicon wafer substrate is a different, and considerably more achievable, milestone than 70% self-sufficiency in the semiconductor stack. But it is a genuine milestone.

The wafer target is the part of China’s semiconductor self-sufficiency where the numbers are most defensible. Achieving 70% domestic use of silicon wafer substrate is a genuine milestone, but a narrower one than the headline implies.

Want to learn more about Cloud Computing from industry leaders? Check out Cyber Security & Cloud Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.