Company history

Sanken did not begin as a company. In October 1937, to mark the fiftieth anniversary of Toho Electric Power, its president Matsunaga Yasuzaemon endowed the Toho Industrial Research Institute; inside it, a semiconductor laboratory worked on selenium rectifiers and, during the war, delivered prototypes as a designated military research establishment. When the institute was wound up after the defeat, the laboratory’s head, Kotani Tetsuji, decided to take its engineers and its equipment with him. In September 1946 he incorporated Toho Sanken Denki in Shiki, Saitama, on capital of ¥150,000, keeping the institute’s name so that the continuity of the technology would be visible from outside. The founding declaration carried one clause that would outlast everything else: we shall devote ourselves to the semiconductor industry.

What the new company had inherited was research; what it lacked was volume production and a customer base. That imbalance decided its first strategic move. When germanium rectifiers appeared around 1952–53 and Sanken found itself behind rivals in germanium work, it judged that germanium was “no more than a stage on the way from selenium to silicon” and skipped the generation entirely, concentrating on silicon before anyone else. In 1958 it completed Japan’s first diffused silicon rectifying element of wholly domestic design. Chasing the incumbent material would have played to a weakness; getting to the next material first turned the laboratory’s only real asset into a product.

The company went over the counter in March 1961 and onto the second section of the Tokyo Stock Exchange that October, which for the first time gave it capital for volume production rather than prototypes. Its old connection to Matsunaga and Toho Electric Power survived the war as a commercial one, supplying the industrial power-supply customers on which the early rectifier business ran.

In June 1962 Toho Sanken Denki became Sanken Electric. The change was less cosmetic than it looked: it moved the company’s self-definition from successor-to-a-laboratory to semiconductor manufacturer. The technology followed the same line. That year Sanken was first to mass-produce silicon rectifying elements for automobiles — the entrance to the automotive semiconductor business that would define it half a century later — and in 1966 its silicon power transistor became the first device of its kind certified by Nippon Telegraph and Telephone Public Corporation.

Capacity was then built out prefecture by prefecture: the Kawagoe plant in 1963, Kashima Sanken as an integrated-circuit subsidiary in 1970, Ishikawa Sanken in 1978 from the merger of five affiliated firms, Yamagata in 1981, Fukushima in 1988. A five-year plan running to the twenty-fifth anniversary in 1971 fixed the terms of that growth in three numbers — ¥3.5bn of sales, 20% annual growth, a 5–6% margin held — a deliberate attempt to scale without leaving the semiconductor trade. In August 1970 the company reached the first section of the TSE, thirty-three years after the institute that produced it. In June 1973 Korea Sanken became its first overseas plant, early for a Japanese manufacturer, and aimed squarely at the labour cost of semiconductor assembly.

By the turn of the 1980s the shape that would carry Sanken for the next thirty years was set: semiconductor devices (transistors and diodes for consumer and industrial equipment), power supplies (industrial switching power units), and social systems (aviation obstruction lights, LED displays and other public-infrastructure products). Three different customer bases were meant to cancel out one another’s cycles — a portfolio logic that held until the group’s centre of gravity moved abroad.

Building automotive magnetic sensors, power ICs and a North American customer base from scratch would have taken a Japanese power-semiconductor maker far too long. So Sanken bought them. In 1990 it acquired the semiconductor division of Sprague Electric in Massachusetts and turned it into Allegro MicroSystems, taking on the accumulated technology and the design floor in a single transaction. Hong Kong (1988), Indonesia (1991), China and Thailand (1996), Tianjin (1996) and Taiwan (1998) filled in around it, and the group settled into a three-layer division of labour: design and development in North America, assembly across Asia, production engineering in Japan.

It worked. Consolidated sales rose from ¥140.0bn in FY01 to ¥203.8bn by FY06, and the year to March 2007 produced ¥11.3bn of ordinary profit as consumer and industrial demand and Allegro’s automotive earnings peaked together. Then the same network became the problem. The collapse of world semiconductor demand after September 2008 took FY09 sales down to ¥134.1bn and produced an ordinary loss of ¥6.0bn; FY09 and FY10 were consecutive loss years. The plants added during the expansion turned into fixed cost the moment the cycle broke.

The recovery confirmed how far the centre had shifted. By the year to March 2014 sales were back to ¥160.7bn with ¥10.3bn of ordinary profit, and semiconductor devices — Allegro’s automotive power ICs above all — were close to 80% of group sales, with power systems around 20% and social systems only a few per cent. Sanken’s consolidated result had become, in effect, a read on one American subsidiary’s exposure to the car market, which cut both ways: FY15 slipped back to ¥155.9bn of sales, ¥3.8bn of ordinary profit and ¥0.2bn of net income on currency swings and soft South-East Asian demand.

The medium-term plan SK-13, from 2018, put the portfolio itself on the table. FY18 showed why: ¥175.2bn of sales and ¥11.8bn of ordinary profit, but a net loss of ¥11.4bn once the impairments taken in restructuring — at Allegro and in power systems alike — were recognised. FY20 lost ¥5.6bn at the net line during the pandemic and FY21 lost ¥7.0bn on ¥156.8bn of sales and a ¥3.4bn ordinary loss. Dependence on a single subsidiary magnified every downturn as reliably as it had magnified the good years.

The dismantling ran on two tracks. In September 2020 Sanken listed Allegro on NASDAQ, a first step down in its holding, and signed a partnership with STMicroelectronics to co-develop silicon-carbide power devices. In May 2021 it absorbed Sanken Densetsu and sold the whole of it to GS Yuasa, exiting the social systems business it had entered in 1974 — an area whose links to the core had thinned over nearly fifty years, and the clearest statement of what “selection and concentration” meant here. Its own manufacturing was renewed in parallel: a development centre opened in September 2021, Niigata Sanken was founded in Kashiwazaki in 2023 as an automotive power-module plant, and Korea’s EK Co. was acquired in December 2023. FY23 came close to a record at ¥225.4bn of sales, ¥27.2bn of ordinary profit and ¥9.5bn of net income.

Then the parent cut the engine loose. Allegro’s NASDAQ listing had left the subsidiary worth four to five times its parent — a parent-child inversion that made Sanken’s own shares look permanently cheap — and in August 2024 the company sold down and reclassified Allegro from consolidated subsidiary to equity-method affiliate, followed by Polar Semiconductor in September. Roughly $904.3M (¥137bn) of proceeds went to the new four-year plan SK-26, to debt reduction and to shareholder returns. The arithmetic is stark: FY24 sales of ¥235.2bn with a ¥8.1bn net loss, then FY25 sales of ¥121.6bn — about half — with a ¥14.3bn ordinary loss and ¥50.9bn of net income created almost entirely by the disposal gain. Eighty-eight years after the institute, Sanken is once again a Japanese power-semiconductor maker on its own account, with profitability as SK-26’s first objective, xEV power modules as its growth driver, and a target of more than 50% of sales from the automotive market.