Samsung's Chairman Says the Company Can't Make AI Chips Fast Enough

The AI chip shortage is real, and Samsung’s chairman just admitted it out loud. Lee Jae-yong said Monday that the company’s current fabrication capacity simply can’t keep up with surging demand for AI-related semiconductors. His solution is a new advanced packaging factory in Gwangju, a city in southwestern Korea that could become Samsung’s next major chip investment hub.

There’s more to the plan than just Gwangju. Samsung is spreading its bets across the country — robots in Gumi, biopharmaceuticals in Incheon, batteries in Ulsan, and semiconductor substrates in Busan. It’s a multi-front investment strategy that reveals just how aggressively the company is trying to reposition itself for the AI era.

The pressure is especially acute in the HBM (high-bandwidth memory) market, where Samsung is chasing SK Hynix’s dominant position. Samsung’s customers already include NVIDIA, AMD, and Google — three of the biggest names in AI infrastructure. Each of them needs more HBM memory than Samsung can currently supply.

In May, Samsung started shipping samples of its latest 12-layer HBM4E memory to customers. That’s a clear signal the company is racing to close the gap in next-generation AI memory products, even as it tries to build out the physical capacity to manufacture them at scale.

The question now is whether the Gwangju factory — and the parallel investments across Korea — will arrive quickly enough to satisfy the insatiable hunger of the AI industry.