Made in China 2025

On April 26, 2018, eleven days after the United States Department of Commerce had cut ZTE off from American technology, Xi Jinping put on a white smock and a hairnet and walked into a clean room in Wuhan. Two cameras followed him. Beside him, matching his stride, walked a wiry man named Zhao Weiguo, born in 1967 to parents who had been banished to Xinjiang during Mao’s anti-rightist campaign. Zhao now ran Tsinghua Unigroup, the conglomerate the Chinese state had picked to build the country’s first credible memory chip. By Zhao’s later account, Xi had wished him a happy birthday on the tour, and the kind word was meant not for him but for the entire Chinese semiconductor industry. By Xinhua’s account, what mattered was the line Xi gave the Yangtze Memory Technologies executives at the end. Semiconductors, Xi said, were as important to manufacturing as a heart was to a human body. No matter how big a body, if the heart was not strong, the body was not strong.

The metaphor was not new. Xi had used it before, and he would use it again at every stop on the trip. The previous summer Beijing had quietly downgraded the public phrase by which most of the world had come to know its industrial policy, the phrase Made in China 2025. The plan itself had not changed. It had simply lost the name on the door. What Xi was doing, in his white smock with Zhao at his shoulder, was reattaching the policy to a different word. The word was not industrial. It was anatomical. China had no heart of its own. It was time, the General Secretary was saying, to grow the organ.

The visit was theater, and like the best Chinese political theater, its target audience was inside the building. ZTE, the country’s second-largest maker of telecommunications equipment, had effectively stopped operating that week. A denial order from the Bureau of Industry and Security on April 15, 2018, had cut the company off from American chips, software, and components for seven years. ZTE could not buy from Qualcomm. It could not run Android on its phones. It could not build base stations for the 5G rollout Beijing was pouring tens of billions of yuan into accelerating. Within days the company announced a halt to “major operating activities.” A publicly traded firm with eighty thousand employees and seventeen billion dollars in revenue was being killed in slow motion by a piece of paper signed in Washington. The shock did not arrive in Beijing as a surprise. It arrived as confirmation of the case Xi had been making for half a decade about strategic vulnerability, demonstrated on national television by his counterparty.

To see what the white-smock photograph meant in Beijing, the camera has to pull back three years, to the spring of 2015, when the policy being rebranded was still a German idea translated into Chinese. The Made in China 2025 plan had been conceived at the Ministry of Industry and Information Technology under the elderly engineer Zhou Ji of the Chinese Academy of Engineering, drafted over two years by a working committee of around 150 academy fellows, vetted by four ministries, and signed by Premier Li Keqiang on May 8, 2015. The State Council published it on May 19. It was a ten-year plan, the first installment of a thirty-year arc aiming to make the People’s Republic a “manufacturing superpower” by the centennial of its founding in 2049. Its public model was Germany’s Industrie 4.0, the cyber-physical-systems framework Angela Merkel’s government had blessed in 2013. Miao Wei, the Industry Minister, told People’s Daily that the German and Chinese plans would arrive at their destinations in the same eight to ten years. He omitted that the Germans had only one industrial revolution to complete and the Chinese had three.

The ten priority sectors were familiar to anyone who had read the German document and noticed that the Chinese version had been bolted on top of an entirely different industrial baseline. Next-generation information technology came first, with semiconductors embedded inside it. Then high-end machine tools and robotics, aerospace, ocean engineering and high-tech ships, advanced rail, new-energy vehicles, power equipment, agricultural machinery, new materials, and biopharma. The architecture was Germanic. The numerical targets were not. A technical road map released by the Chinese Academy of Engineering in late 2015 set escalating self-sufficiency benchmarks for “core basic components and key basic materials”: forty percent domestic production by 2020, seventy percent by 2025. The 70 percent target did not technically apply to the entire semiconductor industry. It applied to the inputs Beijing classified as “core basic.” Washington and Brussels would cite it as if it applied to everything. The distinction was a footnote. The footnote did not survive translation.

Underneath the plan ran the money. In June 2014, almost a year before Made in China 2025 was promulgated, the State Council had quietly set up the National Integrated Circuit Industry Investment Fund. In Chinese finance circles it became known by the obvious shorthand: the Big Fund. Its first phase was incorporated on September 26, 2014, with registered capital of 138.7 billion yuan, roughly twenty-one billion dollars. Its chair was Ding Wenwu, a former MIIT electronic-information director with thirty years of work managing China’s responses to every previous chip-policy disappointment. Ding’s mandate was to deploy the capital in equity stakes rather than grants, the better to mimic venture-fund discipline. Sino IC Capital, the management company set up under China Development Bank, did the deal selection. The shareholders were a coalition of the Ministry of Finance, China Tobacco, China Mobile, Shanghai municipal funds, and the state-owned banks. By the end of Phase One, in 2019, the Big Fund had put thirty billion yuan into foundries, around twenty into design houses, nineteen into memory, ten into assembly and test, and a trickling two billion into equipment. The biggest single check went to Yangtze Memory in Wuhan; the second to SMIC’s Beijing fab; the third to Hua Hong’s Shanghai operations. Tsinghua Unigroup, Zhao’s vehicle, would in early 2017 sign a deal worth up to 150 billion yuan over five years, two-thirds from China Development Bank and the rest from the Big Fund. Around the central column of state money, a much larger column of leveraged private capital marched: by Ding’s own later estimate, every yuan of Big Fund equity drew in roughly five yuan of provincial, municipal, and private money. Phase One catalyzed something close to half a trillion yuan in fresh chip investment.

This was not, as the plan’s foreign critics often suggested, a single decree. It was an architecture. Above the Big Fund sat the State Council. Beside it sat provincial industrial guidance funds in Beijing, Shanghai, Hubei, Anhui, Jiangsu, Sichuan, Fujian, and Guangdong, each with its own local mandate and its own wager on which Chinese chip company would be its ticket up the political ladder. Below sat the recipient firms: SMIC, the foundry founded in 2000 that was still the country’s most credible producer of advanced logic; YMTC, the memory house spun out of Tsinghua Unigroup and the Hubei provincial government in July 2016; ChangXin Memory in Hefei, working on DRAM under former Inotera executive Charles Kau; Fujian Jinhua, the provincial DRAM venture that would later teach the world how a single Justice Department indictment could collapse a state project; and HiSilicon, Huawei’s in-house design unit, which by 2018 had become the most capable chip designer in mainland China. Around them spun a grant economy of tax breaks, accelerated depreciation, subsidized land, cheap utilities, and below-market loans from the Big Four state banks. By the OECD’s later count, Chinese chipmakers received more government support per dollar of revenue than firms in any other major economy.

For most of the policy’s first eighteen months, the world barely noticed. The plan was discussed politely at chambers of commerce, analyzed in MERICS papers in Berlin. Joerg Wuttke, the long-serving president of the EU Chamber of Commerce in China, gave warning interviews. American firms filed comments at MIIT meetings. Then, in late 2016 and through 2017, the mood snapped. A string of attempted Chinese acquisitions of Western chip assets, including Aixtron in Germany, Lattice Semiconductor in the United States, and the unsuccessful Tsinghua bid for Micron, ran into political vetoes that surprised almost no one inside the deals and almost everyone in Beijing. Brussels began drafting an investment-screening regulation. The Trump administration ordered a Section 301 investigation in August 2017. By the time the Section 301 report landed in March 2018, it had named Made in China 2025 116 times. Robert Lighthizer, the trade representative who had built his career on Japan in the 1980s and now had a larger competitor to manage, made the policy the spine of the American case. The April 2018 ZTE order was, technically, an unrelated enforcement action against a recidivist sanctions violator. The technicality did not register in Beijing. The April denial order and the March Section 301 report were treated as the same instrument, fired from the same gun.

The Chinese response to the backlash had two layers. The first was rhetorical and visible. The second was structural and not.

The rhetorical layer began in late spring 2018. Xinhua, which had used the phrase “Made in China 2025” more than 140 times in the first half of the year, abruptly stopped after early June. Internal propaganda directives instructed officials to retire the slogan in speeches and op-eds. By the first quarter of 2019, references in People’s Daily had fallen 83 percent against the prior year. The plan disappeared from Premier Li’s annual government work report. The white papers continued, under euphemism: high-quality development, dual circulation, scientific and technological self-reliance. Foreign reporters who asked Chinese officials about Made in China 2025 were told it had never been the central plan they imagined. Some were told, with a straight face, that it had been a draft document only. The performance fooled no one, but it accomplished its two domestic tasks. It gave Lighthizer fewer paragraphs to quote in his next round of tariffs. And it allowed Beijing to remove the brand without removing the policy.

Underneath the rhetorical retreat, the structural response was the opposite of a retreat. It went into Xi’s own mouth. On April 19, 2016, two years before the ZTE shock, Xi had told a cybersecurity work conference that depending on imported core technologies was like building one’s house on someone else’s wall: no matter how big and beautiful the house, it would not stand in a storm. The wall-and-storm metaphor had not yet caught on. On April 20, 2018, five days after the ZTE order, he upgraded the formulation. Core technologies, Xi said, were “important instruments of the state,” guo zhi zhongqi, 国之重器, a phrase from the Han Feizi whose register was closer to “vital state weapons” than to anything in conventional industrial-policy English. To rely on imported core technologies, he said, was to plant your vegetables in someone else’s garden. The metaphor stack now had three layers: a heart that did not belong to the body, a roof that sat on a borrowed wall, and a vegetable patch on land that could be foreclosed. He was building a vocabulary for an emergency.

A month later, on May 28, 2018, Xi addressed the joint conference of the Chinese Academy of Sciences and the Chinese Academy of Engineering in the Great Hall of the People. The text was published years afterward in Qiushi, the party’s theoretical journal, and circulated through research institutes for cadre study. In it, Xi did something he had not done explicitly before. He invoked the Mao-era Two Bombs, One Satellite project, the liang dan yi xing program of the 1960s under which Chinese scientists, cut off from Soviet help, had built an atomic bomb in 1964, a hydrogen bomb in 1967, and the Dongfanghong I satellite in 1970. The Two Bombs scientists were the founding myth of self-reliant Chinese science: men like Qian Xuesen, the Caltech professor expelled from the United States in 1955, and Deng Jiaxian, the bomb designer who refused to leave a hospital bed when radiation poisoning made him unable to work. The semiconductor industry, by implication, was being asked to do what the bomb-builders had done: produce, under embargo, what the country could not buy.

The analogy was emotionally accurate and operationally misleading. The Two Bombs scientists had built three discrete devices for the state’s exclusive use, on a budget the state would pay any price to fund, against an engineering frontier already decades old by the time China crossed it. Semiconductors were a continuous, civilian, commodity industry whose specifications doubled every two years and whose viability depended on selling at scale into global markets. The Soviet electronics industry had failed not because Soviet physicists could not design transistors but because Soviet planning could not deliver them at volume and at quality. The Two Bombs analogy, taken literally, prescribed exactly the institutional reflex that had killed the Soviet effort. Xi was not asking for that. What he was asking for, though it would not be clear to outside observers until the Phase Two and Phase Three Big Fund commitments, was a hybrid: the political mobilization of liang dan yi xing with the equity-fund discipline of a venture portfolio overlaid on top. It was a stack that had never been tried at this scale, and it would have been politically unthinkable under any other post-Mao leader. Xi could attempt it because, by 2018, he had collected the personal authority to insist.

The personalization mattered. Made in China 2025, in 2015, had been Premier Li Keqiang’s plan, with Miao Wei’s signature on the technical road map and Zhou Ji’s committee delivering the substance. Xi had endorsed it at the Central Economic Work Conferences and embedded it inside the Five-Year Plans, but he had not been its public face. Beginning in 2018, that changed. The press readout of every chip-related Politburo Standing Committee meeting led with Xi’s remarks rather than Li’s. Xi began visiting fabs personally, asking factory directors, in front of cameras, whether their wafers were being produced by domestic firms. He pulled the topic into the language he reserved for items he treated as personal: the Belt and Road Initiative, the anti-corruption campaign, common prosperity, national rejuvenation. The semiconductor program migrated, in his rhetoric, from one priority among ten into the master example of the entire Made in China 2025 logic. By the time the State Council approved Big Fund Phase Two on October 22, 2019, at 204.15 billion yuan, the policy had become Xi’s in a way it had never been Li’s.

That personalization had two consequences. The first was that the program acquired the tempo of a leader-led campaign rather than the cadence of an industrial policy. The investment was bigger, faster, and less patient than equivalent Korean or Taiwanese state efforts a generation earlier. The second was that, like every leader-led campaign in the Xi era, it acquired an enforcement risk. In July 2022, Ding Wenwu was taken away by the Central Commission for Discipline Inspection. Within weeks, Sino IC Capital executives, the head of strategy at Tsinghua Unigroup, and a handful of provincial-fund directors were detained as well. Zhao Weiguo, the man who had escorted Xi through the Yangtze Memory clean room in 2018, would be arrested later that year, tried in 2023, and convicted in May 2025 on corruption charges, receiving a suspended death sentence. The political subtext was harder to read than the embezzlement charges. Some observers in Hong Kong and Tokyo argued that the purges were a routine consequence of any campaign that pushed this much money into a sector this fast. Others argued they were the predictable accompaniment of a leader’s loss of patience with results: at the most charitable reading, China had moved from roughly fifteen percent self-sufficiency in the early 2010s to twenty-five or thirty by the early 2020s, well short of the seventy-percent target the foreign press had treated as a manifesto. The two readings were not exclusive. Both could be true. Both probably were.

What was not in dispute, by the time Big Fund Phase Three was approved at 344 billion yuan in May 2024, was that the project had outgrown the slogan that birthed it. Made in China 2025, the brand, would not survive 2025 itself; the 70-percent target would, in the 14th and 15th Five-Year Plans, quietly disappear. The underlying program would continue under different names with substantially more money. Xi had seen to that. He had also accepted, in a way the Premier who had signed the document in 2015 had not been asked to, that the program would not be a German-style supply-side modernization but a war-economy pursuit of industrial sovereignty. The vocabulary kept drifting. By 2021 the operative phrase was “self-reliance and self-improvement in science and technology.” By 2024 it was “new productive forces.” Industrie 4.0 had become a German export marketing term. Made in China 2025, with its name removed and its budget tripled, had become the Chinese state’s master industrial project of the early twenty-first century.

In the quiet years before the cameras came on at Yangtze Memory in 2018, the engineers inside Wuhan’s clean room had run their first 32-layer 3D NAND wafers. They were not yet competitive on cost or yield with Samsung or SK Hynix in Korea or Micron in Idaho. They were the first credible Chinese-made memory chips since the Mao-era abandoned attempts. Xi, in his white smock, told them their hearts had to grow. They went back to work. Within five years YMTC would be on a national entity list, denied access to American tools, embargoed from the equipment that made the next process node possible, and asked to keep going anyway. The plan that had sent them there had been written under one premier and rebranded under another. By then it belonged, in the way these things come to belong, to the man who had walked through their fab in April 2018 and stayed in office long enough to make sure that whatever came next would happen on his watch and to his account.