SK Hynix seeks $28bn in US listing as AI chip shortage bites
South Korean chipmaker SK Hynix is launching a $28 billion US share sale, a move that highlights the severe global memory shortage pushing up hardware prices for European consumers.
SK Hynix is planning to sell nearly 17.8 million shares in a US initial public offering. The South Korean memory chipmaker expects to raise around $28 billion based on its closing share price in Seoul last Friday. The company will offer American depositary receipts, with each certificate representing a tenth of a common share, pricing on Thursday and trading on Friday.
The listing capitalises on an unprecedented surge in demand for memory chips driven by artificial intelligence infrastructure. For European businesses and consumers, this demand has immediate economic consequences. Apple executives said the shortage is forcing it to raise prices on Mac computers and iPads, a cost pressure that will be felt directly by buyers across the continent.
Systems running AI require vast amounts of memory to function effectively. As hyperscalers like Amazon, Microsoft, Google and Oracle race to construct AI data centres, demand has outpaced supply for high-bandwidth memory, DRAM and NAND chips. The resulting bottleneck has been dubbed "RAMageddon" by industry observers.
SK Hynix, a rival to Samsung and US-based Micron, reported first-quarter revenues up nearly 200% compared to the same period last year. Its stock has climbed roughly 260% so far this year. Investors hunting for the next major AI winner are treating memory manufacturers as the closest available equivalents to Nvidia.
Micron provides a clear benchmark for the market's appetite for these stocks. The US firm has surged nearly 700% over the past year, reaching a valuation of more than $1 trillion on the back of record AI-driven memory demand.
However, this investment cycle carries substantial risks for the global tech supply chain. SK Hynix and Samsung have collectively vowed to spend over $550 billion on new manufacturing capacity to resolve the current bottleneck. Because these massive semiconductor fabrication plants take years to construct, there is a real danger that specific AI memory requirements will shift before they come online. Such a mismatch could eventually flood the market with excess supply and trigger a severe price crash, ultimately reshaping the cost of enterprise computing hardware across Europe.
