Nvidia's Vera Rubin Enters Full Production: HBM4 Landscape Could Split Fortunes for Samsung and SK Hynix
Nvidia has begun full-scale production of its AI computing platform, Vera Rubin, which reduces server node assembly time to under five minutes. Samsung Electronics and SK Hynix supply HBM4 memory for Vera Rubin, but their market shares are diverging. Samsung's HBM4 revenue is expected to triple in Q3, while SK Hynix's share and forecasts have declined. SK Hynix's Q3 operating profit consensus is 76.964 trillion won, with an operating margin of about 78.4%.
How this was made
The 30-second read
Why it matters
The announcement creates a clear winner‑loser dynamic in the HBM4 market, with Samsung positioned to capture a larger share while SK Hynix faces headwinds from certification delays.
Market read
The news directly impacts memory‑chip suppliers and the broader AI‑hardware ecosystem, creating short‑term trading opportunities.
What to watch
Currency fluctuations and inventory levels in data‑center OEMs may temper the projected HBM4 revenue growth.
Background
Nvidia announced its Vera Rubin AI platform is now in full production, cutting server assembly time dramatically. The platform relies on HBM4 memory supplied by Samsung and SK Hynix, prompting a reassessment of their market shares.
Market effects
Accelerated AI‑chip demand may lift the broader memory‑chip sector, benefiting suppliers with faster HBM4 roll‑outs.
South Korean semiconductor stocks could diverge, with Samsung gaining relative strength over SK Hynix.
Nvidia's production shift reinforces AI‑driven demand, influencing global memory‑chip supply dynamics.
Counterpoint
If Nvidia's next‑gen accelerator faces further delays, HBM4 demand could stall, muting Samsung's upside.
Key entities
- CompanyNvidia
AI computing platform provider launching Vera Rubin.
- CompanySamsung Electronics
Supplier of HBM4 memory, gaining market share.
- CompanySK Hynix
Supplier of HBM4 memory, losing market share.
