SK hynix has announced it is considering investment proposals from its major tech customers to help fund new production lines.
This move is happening because the AI industry is booming, creating an insatiable appetite for specialized chips called High-Bandwidth Memory (HBM). HBM is essential for training and running powerful AI models, and SK hynix is a world leader in producing it. The company has stated that demand is so high it simply cannot make enough, creating a critical bottleneck for the entire AI ecosystem.
So, why are these funding deals emerging now? There are three main reasons. First is the EUV scarcity. To make the next generation of HBM, you need extremely complex and expensive machines that use Extreme Ultraviolet (EUV) lithography. Only one company in the world, ASML, makes them, and their order book for 2026 is reportedly full. By co-investing, tech giants can help SK hynix secure these scarce tools.
Second, the scale of investment is immense. SK hynix has committed to spending tens of billions of dollars on new factories and equipment. Customer financing provides a direct injection of cash, reducing the financial risk for SK hynix and allowing it to build faster without diluting its stock value through public offerings.
Finally, the entire industry is changing its habits. Short-term deals are being replaced by multi-year contracts, often with large prepayments. Companies like Microsoft and Google are reportedly negotiating 3-to-5-year deals to lock in their supply. This proposal is a logical extension of that trend, moving from just prepaying for chips to co-funding the factories that make them.
This new arrangement creates a powerful partnership. Big Tech companies get a guaranteed supply of the "gold" of the AI era—HBM chips. In return, SK hynix gets the capital and demand certainty it needs to undertake its ambitious expansion plans. While SK hynix might offer slightly lower prices in these deals, the stability gained is a significant advantage in the notoriously cyclical semiconductor market. It’s a strategic shift toward sharing both the risks and the rewards of the AI revolution.
- HBM (High-Bandwidth Memory): A type of high-performance memory crucial for AI accelerators like GPUs, allowing for faster data processing.
- EUV (Extreme Ultraviolet) Lithography: An advanced chip-making technology that uses extremely short wavelength light to etch circuits, necessary for producing the most powerful and efficient semiconductors.
- Capex (Capital Expenditure): Funds used by a company to acquire, upgrade, and maintain physical assets such as property, plants, buildings, technology, or equipment.