In June, Teikoku Databank, the Tokyo credit-research firm, could find 10,118 Japanese companies with operations in mainland China. Two years earlier it had counted 13,034. In 2012, the peak, there were 14,394.

The new count, published on 28 September, is the lowest since the firm began the survey in 2010. It is 22.4 per cent below the 2024 figure and 29.7 per cent below the peak. Since the 2024 survey, 1,221 Japanese companies have opened a subsidiary, factory or representative office in China, the fewest new entrants between any two surveys, though the firm cautions that the periods are not strictly comparable. A record 4,137 closed their operations or could no longer be traced.

The headline number suggests a retreat. The detail suggests something more specific: Japanese companies are keeping China as a place to make things while ceasing to treat it as the market they are built around.

The factory stays, the customer moves

The clearest evidence comes from a survey of about 450 companies already operating in China, which Teikoku Databank conducted in October 2025 and included in the report. Asked to name the single most important country for production, 43.7 per cent chose China. In 2019, 49.0 per cent did. China remains by far the first choice, and the loss is about five percentage points.

Asked to name the most important country for sales, 31.3 per cent chose China, against 47.7 per cent in 2019, a fall of more than 16 points. China still ranks first. The countries that gained were Thailand, whose share rose to 9.7 per cent from 4.8 per cent, India, up to 4.9 per cent from 1.4 per cent, and Indonesia, up to 3.1 per cent from 1.8 per cent. The United States ranked third, at 9.4 per cent.

For production, the alternatives are South-East Asian: Vietnam at 8.7 per cent, Thailand at 7.0 and Indonesia at 3.5.

The firm's reading is that China's appeal as a production base has weakened only slightly, while its appeal as a growth market has changed far more. China still has the factories, the supplier networks and the industrial clusters, it noted. What it has lost is the expectation that its buyers will carry a company's growth.

Who is leaving

The exits are concentrated among smaller firms and labour-intensive industries. The share of China-based Japanese companies with annual sales below ¥1 billion has been falling, while the share of mid-sized and large companies has risen. Companies with sales below ¥100 million now make up 4.8 per cent of the total, and those with ¥100 million to ¥1 billion 24.3 per cent.

By sector, manufacturers fell to 4,206 from 5,139 and wholesalers to 3,320 from 4,218. Services companies fell to 1,213 from 1,803, with the steepest losses in software and information services. Textile and clothing manufacturers fell to 218 from 310, and publishing, printing and furniture-making also shrank. General machinery makers remain the largest group of manufacturers, with 1,005 companies.

Geographically, Shanghai still has the most Japanese companies, 3,961, but that is down by 1,085, or 21.5 per cent, in two years, as companies merged local subsidiaries and trimmed regional headquarters, the firm said. Beijing's count of 555 is about half its 2022 level. Jiangsu province, home to manufacturing centres such as Suzhou and Wuxi, fell by only 7 per cent from 2024, to 1,517. An earlier move inland did not continue: counts fell in Shaanxi, Chongqing and Sichuan too.

Tariffs, and everything else

Teikoku Databank lists the reasons it believes are behind the decline: worse relations between Japan and China, the slump in China's property market, export restrictions on rare earths, which it describes as economic coercion, China's revised counter-espionage law, overcapacity and price competition from state-supported Chinese companies. Companies moving or spreading production to Vietnam, elsewhere in South-East Asia and back to Japan have also reduced China's importance as a cheap production base, especially for smaller firms.

American tariffs add to the pressure on companies that make goods in China and sell them in the United States. In the October 2025 survey, 47.1 per cent of the companies operating in China expected US tariff negotiations to reduce their profits for the 2025 financial year, against 33.4 per cent of all respondents. Among manufacturers in China it was 58.4 per cent. One company told the researchers that orders at its Chinese subsidiary had fallen because of the tariffs.

Yet moving production is hard. The firm found that many companies were continuing to produce in China while watching costs, because expanding in the United States, where labour is expensive, is difficult and switching to another country is not easy either. Its overall conclusion is that Japanese companies are lowering their dependence on China while keeping their operations there, rather than leaving outright, though it expects consolidations and closures to increase over the medium to long term.

Tariffs are reshaping the trade in other ways too. This publication reported on Saturday that the gap between China's and America's figures for the same trade has grown to almost $100 billion this year, with Vietnam's figures diverging the other way.

All figures are from Teikoku Databank's survey of Japanese companies with operations in the People's Republic of China, published in Japanese on 28 September 2026 and read in full, which counts companies in its own databases (COSMOS2, about 1.51 million companies, and its credit-report files) with subsidiaries, affiliates, investments or representative offices in mainland China, excluding Hong Kong, Macau and Taiwan, as of June 2026. The production- and sales-base preferences and the tariff findings come from Teikoku Databank's survey of companies conducted in October 2025, as reported in the same document. Translations and the comparisons of percentage-point changes are this publication's. Accurate to 5pm ET on 5 October 2026.

Topics worldjapanchinasupply chainstariffsmanufacturing

Technology Correspondent

Alison Acosta

Alison Acosta reports on artificial intelligence, enterprise software and the infrastructure behind the modern internet, with a focus on how technical decisions become business decisions.