SK's Japan Chip Play Signals a New Phase in Korea-Japan Rivalry

From Diplomacy to Deals

Choi Tae-won, chairman of SK Group, did not mince words during a September 2 interview with Japan’s Asahi Shimbun. He said SK is considering building a new memory semiconductor factory in Japan and floated the possibility of joint production with Keyoxia, the Japanese NAND flash manufacturer in which SK Hynix already holds a stake. But the most consequential line may have been the threat that accompanied the offer: if Keyoxia could not form a cooperative relationship, SK would terminate its investment there entirely.

This is not the language of a company seeking a symbolic partnership. It is the language of a boardroom strategist weighing leverage against exit. And it signals that Korea-Japan semiconductor cooperation — long stymied by historical grievances and competing industrial policies — has crossed a threshold from diplomatic goodwill into transactional alliance.

The Ownership Web

To understand the stakes, you need to trace the ownership chain. Keyoxia was once Toshiba’s semiconductor division. Last month, its largest shareholder changed: Toshiba sold its stake to BCPE Pangea Cayman 2, a special purpose company run by an American investment fund. SK Hynix holds convertible bonds issued by that very vehicle. If SK exercises its conversion rights in full, it would effectively control the SPC’s voting power — giving it a indirect path to influence, if not command, Keyoxia’s strategic direction.

This structure is unusual. It means SK Hynix sits not as a direct equity holder in Keyoxia, but as a creditor with the option to become one. Choi’s public comments suggest SK is now prepared to move from potential controller to active partner — or walk away if Keyoxia’s leadership declines. Either outcome gives SK more influence than it had before, and it forces Keyoxia into a corner it may not have anticipated.

Keyoxia responded calmly, noting that SK Hynix is already a supplier of its DRAM chips and that the two firms cooperate on MRAM (magnetic RAM) research. But the subtext was clear: Keyoxia does not want to lose SK as a customer or partner. In a market where demand is tightening and supply is fragile, every new relationship matters.

The Demand Shock Behind the Move

The driver behind SK’s pivot is the AI boom. Choi stated plainly that datacenter memory semiconductor demand is facing a 20 to 30 percent shortfall. That figure is not a projection — it is an observation from inside one of the world’s largest chip companies. AI training workloads consume massive amounts of high-bandwidth memory, and the supply chain simply has not caught up with the speed of demand growth.

This is the structural gap that China’s semiconductor push has been trying, and largely failing, to fill. China has invested billions in memory chip manufacturing, but its progress has been hamstrung by equipment export controls and a lack of access to cutting-edge process technology. The result: a widening supply-demand gap that benefits whoever can expand capacity fastest — and who can secure supply chain alliances that bypass Chinese dependence.

SK’s Japan move is designed to do exactly that. Japan already hosts a dense cluster of semiconductor equipment and materials manufacturers — companies like Tokyo Electron,SCREEN, and Shin-Etsu Chemical. Choi cited Japan’s supply chain ecosystem as a decisive factor in his decision to explore a local fab. In practical terms, this means SK could reduce lead times, lower logistics risk, and insulate its memory production from both Chinese disruption and Korean domestic capacity constraints.

Who Wins, Who Loses

Samsung, of course, is the company most likely to lose here. SK’s move into Japan directly contests Samsung’s decades-long dominance in memory chip manufacturing and its strategic foothold in the Japanese market. Samsung has historically treated Japan as a supplier base, not a production site. SK is treating it as a frontier.

Keyoxia is the other potential loser — or at least, the party most exposed to renegotiated terms. The convertible bond arrangement gives SK a lever it did not have a year ago. If SK decides to press that lever, Keyoxia faces either capitulation or the loss of its most significant customer.

On the winning side: SK itself, which gains geographic diversification, a stronger negotiating position with Keyoxia, and a foothold in the world’s second-largest semiconductor market. China loses again — its containment strategy gains another concrete expression, and SK’s Japan alliance adds a new pillar to the Western-led supply chain architecture that aims to reduce dependency on Chinese semiconductors.

What Happens Next

Several outcomes are plausible. SK may announce a specific Japanese factory site within the next fiscal year. It may convert its convertible bonds and consolidate control over Keyoxia’s strategic direction. Or it may walk away — though given the scale of the memory chip shortage, walking away seems economically irrational.

The more likely scenario is a hybrid: SK invests in a Japanese facility while deepening its operational ties with Keyoxia, turning the convertible bond stake into real influence. Keyoxia, for its part, will need to decide whether to accommodate SK’s demands or seek alternative partners — a decision that could reshape the global NAND flash market.

Japan’s government is also watching closely. Seoul has been eager to deepen economic ties with Tokyo, particularly around semiconductors, as a counterweight to Chinese influence. Choi’s announcement gives the Japanese side a tangible opportunity to attract foreign investment into its already-strong equipment and materials sector. Whether Tokyo welcomes SK as a partner or views it as a competitor to Samsung’s established presence remains an open question.

The broader implication is that the Korea-Japan semiconductor axis is moving from statement to substance. The diplomatic committees, trade summits, and joint press releases of recent years are being replaced by factory plans, convertible bond conversions, and operational partnerships. The era of Korea-Japan chip cooperation as talking points is ending. The era of Korea-Japan chip cooperation as deal-making has begun.