Company history

SUMCO has no single founding. Between December 1958 and April 1960 three separate companies were set up to make high-purity silicon for semiconductors: Nichitsu Electronic Chemical, founded by Shin Nihon Chisso Hiryo; Nihon Denshi Kinzoku, founded by Mitsubishi Metal Mining and others in October 1959; and Komatsu Electronic Metals, founded jointly by Komatsu and the Ishizuka Research Institute in April 1960. In each case a non-ferrous metals or materials parent was extending its existing trade into a new one, and Japan met the shift from transistors to integrated circuits not with one national champion but with three parallel keiretsu ventures.

The Sumitomo line began wafer production at Osaka Titanium’s Amagasaki works in January 1962 and, in August 1973, set up Kyushu Electronic Metals jointly with Sumitomo Metal Industries. For three decades the Sumitomo, Mitsubishi and Komatsu lines walked the same market as separate businesses, each carrying its own plants, sales channels and development. While demand rose, that dispersion simply gave each of them room to grow. Once wafers moved to larger diameters and each new fab cost tens of billions of yen, the same structure read as duplicated investment and insufficient scale — and, from the customer’s side, as a fragmented supply base.

The reckoning came in the late 1990s, when a global glut in wafers coincided with the demand that the industry convert to 300mm. In October 1998 Sumitomo Metal Industries absorbed Sumitomo Sitix, pulling the Sumitomo silicon business back into the parent as a division; Mitsubishi Materials was weighing the same options through Mitsubishi Materials Silicon. With memory prices falling, neither group could justify carrying the next round of capital spending alone. Both, almost simultaneously, hauled their subsidiaries back in-house to make them usable as pieces in a merger.

In July 1999 Sumitomo Metal Industries, Mitsubishi Materials and Mitsubishi Materials Silicon jointly founded Silicon United Manufacturing — the vehicle that became SUMCO. The point was not tidiness. Converting to 300mm required tens of billions of yen per site, and where Shin-Etsu Chemical moved early and alone, the later-arriving Sumitomo and Mitsubishi camps chose to split the bill. That divergence — solo investment against pooled investment — set the rhythm of the two-player Japanese industry that followed.

Silicon United began 300mm production in October 2001. In February 2002 it took over Sumitomo Metal Industries’ Sitix division and merged with Mitsubishi Materials Silicon on the same day, renaming itself Mitsubishi Sumitomo Silicon. In August 2005 the name was unified as SUMCO, and that November — barely three years after the merger — the company listed on the First Section of the Tokyo Stock Exchange. The listing was not a victory lap: a public offering of $545.6M (¥60bn) repaid borrowings and moved the source of investment capital from two cash-constrained parents to the market.

The timing was fortunate. Riding the 300mm conversion, sales went from $2.0B (¥221bn) in FY05 to $4.0B (¥475bn) in FY07, with operating profit of $1.2B (¥140bn). In October 2006 SUMCO bought Komatsu Electronic Metals (later SUMCO TECHXIV), finally gathering all three original lines under one roof and leaving Japan with a two-firm industry alongside Shin-Etsu. But wafers are a capital-goods business: every expansion adds fixed cost, so scale pays while demand holds and depreciation bites the moment it does not. The boom hid that.

The 300mm build-out ran backwards as soon as the cycle turned. In FY09 SUMCO reported sales of $2.3B (¥218bn), an operating loss of $924.8M (¥87bn) and a net loss of $1.1B (¥100bn) — the worst result since the merger. Memory prices collapsed after the financial crisis, utilisation fell, and impairments piled on. Losses continued for three straight years: $746.3M (¥66bn) net in FY10 and $1.1B (¥84bn) in FY11. The forward investment sold to the market at listing as the company’s greatest strength had become, in a downturn, a machine for converting idle capacity into losses.

Leadership turned over with it. Shigematsu Kenjiro’s tenure ended with FY08; Taguchi Yoichi took over in FY09, and Hashimoto Mayuki from FY11. Under Taguchi the company closed Amagasaki in February 2011 — the plant where the Sumitomo line had first made wafers in 1962 — and Ikuno in July 2013, and withdrew from solar-grade wafers, cutting some 1,300 jobs. The restructuring was a re-selection of the plant network the three predecessors had contributed, judged strictly by whether a site served 300mm. Preserving the company meant cutting away its own origins.

Hashimoto Mayuki became president in FY11 and chairman and CEO from the end of 2015, holding the top job for more than a decade — the first stable command the merged company had. Recovery showed from 2012: operating profit returned to $165.4M (¥13bn) in FY12 and $241.9M (¥26bn) in FY14, and in March 2016 SUMCO shifted to a company with an audit and supervisory committee. Concentrating a decade of production reorganisation and customer renegotiation under one person, rather than parcelling it out to two-year presidencies, was itself the structural answer to the three loss-making years.

The signature of the Hashimoto years was price discipline through long-term contracts. In downturns he argued that SUMCO fell less than the market because memory was only about half of its mix against roughly 75% for the industry, and that contract prices held. The results followed the argument: operating profit of $770.8M (¥85bn) in FY18 and $834.3M (¥110bn) in FY22, the highest since the merger. The company that lost more than ¥100bn in FY09 was, a decade later, earning more than ¥100bn — on customer mix and pricing rules decided in the worst years.

Behind the peak sat capital spending of $2.2B (¥315bn) in FY23, up from $995.7M (¥131bn) — the largest single year in the company’s history, roughly double the FY07 build-out. Interest-bearing debt reached $1.6B (¥224bn) and the equity ratio fell from 59.8% to 53.3%. The case was AI: SUMCO’s FY23 briefing put wafer consumption at about 1.8 300mm wafers per AI server, some 3.4 times a general-purpose server, and forecast 26% CAGR in advanced-process wafer demand for data centres from 2023 to 2027. President Awa Toshihiro pointed to China, autonomous driving, EVs and robotics as the medium-term drivers, and in March 2023 SUMCO took a stake in Mitsubishi Materials’ new high-purity polysilicon company to secure feedstock. The wager is that AI, not the DRAM cycle, sets the next decade — and in February 2025 the company acted on it, deciding to end small-diameter wafer production at SUMCO TECHXIV’s Miyazaki works by the close of 2026 and booking an extraordinary loss of $38.8M (¥6bn), redirecting the descendants of the 1962 Sumitomo line to advanced 300mm logic wafers.