San Jose’s cavernous convention hall hummed with geek fervor as SK Group Chairman Chey Tae-won settled into Jensen Huang’s keynote, scribbling notes on Vera Rubin specs like a kid eyeing candy.
SK Group Chairman Chey Tae-won at GTC 2026. There it is, the headline-grabber. First time he’s graced NVIDIA’s GPU lovefest, and it’s no accident. SK hynix wants the world to know they’re not just memory nerds—they’re innovation partners in the AI gold rush. Or so the press release screams.
But here’s the thing. Chey didn’t fly across the Pacific to clap politely. He toured booths. Schmoozed Huang. Signed faux GPUs with hearts. It’s a flex, pure and simple.
Why Bother with a Chairman’s Joyride?
Look, execs jet-set all the time. What makes this different? Timing. Fresh off a “chimaek” beer-and-chicken summit with Huang last month—yes, really—they’re doubling down on HBM ties. High-bandwidth memory, the lifeblood of AI beasts like Blackwell and Rubin. SK hynix supplies it. NVIDIA gobbles it. Win-win? Sure. But SK’s playing catch-up to Samsung in volume, so this schmooze fest reeks of desperation wrapped in diplomacy.
Chey tracked Huang’s spiel on the “AI stack”—energy to apps, chips in the middle. Vera Rubin details dropped: mass production this year, Feynman lurking next. Huang’s roadmap screams demand for faster, denser memory. Chey listened. Nodded. Then hit the SK booth.
Together, they ogled HBM4 mockups for Rubin, SOCAMM2 for power-sipping servers, HBM3E on Grace Blackwell live. Huang scrawls “JENSEN ♡ SK HYNIX” on a Rubin demo. Photo op gold. But beneath the grins? Hardball talks on HBM4 adoption.
Short version: It’s working. Partnerships like this fueled SK’s AI pivot. Remember the memory crunch last year? NVIDIA begged for HBM. SK delivered. Now they’re locking in future supply.
“By engaging directly with key players in the global AI ecosystem, Chairman Chey is reinforcing SK hynix’s AI leadership,” an SK hynix spokesperson said. “His activities at GTC are expected to further solidify the company’s position in next-generation AI memory, including HBM4.”
Spokesperson spin, naturally. But facts back it. Long before AI hype exploded, SK and NVIDIA cooked up HBM for high-perf GPUs. Now HBM4 hits Rubin’s sweet spot.
Does Chey Tae-won’s GTC Stunt Actually Move the Needle?
Punchy question, right? Skeptics—and I’m one—might yawn. Another CEO photo dump. Yet dig deeper. Chey’s not just spectating. He’s scouting: Foxconn’s power gear for AI factories, NVIDIA’s Blackwell/Rubin demos in robo-cars and drones. It’s ecosystem mapping. SK wants in everywhere—memory, storage, even eSSDs for data centers.
And the dry humor? Huang’s heart doodle. Adorable. Or corporate cute-overload? Take your pick. Either way, it signals trust. SK’s HBM3E powers GB300 today. HBM4 tomorrow. They’re not waiting for invites.
But let’s call BS on the fairy tale. SK hynix trails Samsung in HBM bits shipped. Yields? Rumors say shaky. This GTC trip? PR polish on supply chain scrambling. Bold prediction: By 2027, SK flips the script, grabbing 50% HBM market share. Why? NVIDIA’s Rubin famine looms, and SK’s fabs are humming louder than rivals’.
Historical parallel nobody mentions: 2012 smartphone boom. SK crushed on LPDDR, rode iPhone/Android waves to billions. AI’s the new smartphone. Chey’s replaying that script, but with GPUs instead of galaxys.
He wandered booths like a chip whisperer. Foxconn’s thermal wizardry for AI infernos. NVIDIA’s agentic AI dreams. It’s all connected—your Vera Rubin needs SK’s HBM to not melt.
Critique time. SK paints Chey as ecosystem kingpin. Really? They’re suppliers, not stack overlords. Nice try, though. That “innovation partner” tag? Marketing fluff to mask vendor status.
Why Does HBM4 Matter More Than the Hype?
Because without it, your ChatGPT dreams crash. HBM4 packs density GPUs crave—1.2TB/sec bandwidth whispers say. Rubin Ultra? Feynman? They’ll starve sans SK’s juice. Chey knows. Huang knows. We’re all along for the bandwidth ride.
Live demo: HBM3E + GB300. Screaming fast. Mockups tease more. Chey grilled techs. Huang probed. Q&A gold turned signing schtick.
Wider lens: SK eyes on-device AI too—LPDDR5X, cHBM tweaks. Not just hyperscalers. Your phone’s next gen model? SK’s fingerprints.
Skeptic’s take: Overhyped? Partly. AI factories guzzle power like 747s. Memory’s key, but cooling and silicon shortages loom larger. Still, SK’s positioned. Chey’s trip cements it.
One wild aside—chimaek to GTC in a month. That’s not casual. That’s deal-sealing velocity.
Wrapping the booth beat: Strong interest, per SK. Investors nodded. Stock twitched up. Reality check passed.
🧬 Related Insights
- Read more: NVIDIA’s GTC Gambit: Local AI Agents on RTX PCs and DGX Spark Cut the Cloud Cord
- Read more: Memory Chip Shortage: AI’s Feast Means Your Famine for Years
Frequently Asked Questions
What did Chey Tae-won do at GTC 2026? Chey attended Jensen Huang’s keynote, toured the SK hynix booth with him, checked HBM4 and other AI memory demos, and visited partner booths like Foxconn and NVIDIA.
Is SK hynix leading HBM for NVIDIA’s Rubin? Yes, HBM4 from SK hynix is set for NVIDIA’s Rubin platform, building on years of partnership.
Why is Chey Tae-won personally at GTC? To strengthen SK hynix’s role as an AI memory powerhouse and expand global ties beyond just NVIDIA.
