SKHY
New factory, same AI sugar rush
SK Hynix is putting more than $4 billion into an Indiana facility, and it’s not exactly whispering about why. The company says the site at Purdue Research Park will start volume production of next-gen HBM4E chips in the third quarter of 2029. In other words: this is a long runway bet that today’s AI chip frenzy still has plenty of miles left.
The shortage that won’t quit
CEO Kwak Noh-Jung also said the memory shortage will persist through 2030. That’s the kind of sentence that makes memory-chip bulls sit up straight, because when supply stays tight, pricing can stay friendlier — and that’s usually the part investors like best.
SK Hynix says the U.S. plant is meant to get closer to big customers like Nvidia, Microsoft, and Alphabet. Translation: if AI servers are the new gold rush, SK Hynix wants to be parked next to the shovel makers.
Why the market cares
The company already held 58% of the global HBM market by revenue in Q1 2026, according to Counterpoint Research, with Micron and Samsung far behind at 21% each. That’s an enviable lead, but also a reminder that this market is turning into a heavyweight boxing match, not a solo act.
- More HBM supply in the U.S. could ease customer concerns about sourcing and logistics.
- A persistent shortage suggests demand is still outrunning supply, which is usually good news for pricing power.
- The Indiana fab also shows how the AI buildout is becoming a geopolitical story, not just a tech one.
Big picture
SK Hynix isn’t just building a factory — it’s building a bridge into the center of the AI ecosystem. If the memory crunch really lasts into 2030, this looks less like a capex splurge and more like a seat reservation at the hottest table in semis.