Samsung Electronics Accelerates Research and Development of Next-Generation High-Bandwidth Memory, First Batch of HBM4E to Be Produced in May


Samsung Electronics is fully advancing the R&D process of its next-generation high-bandwidth memory (HBM) products, striving to further consolidate its advantages in the high-end artificial intelligence memory market. According to reports, Samsung plans to produce the first batch of HBM4E samples that meet NVIDIA standards as early as May 2026. Insiders in the industry say Samsung has a clear and tight timeline. Its target is to have its foundry division successfully produce samples of the HBM4E core logic chips before mid-next month.
#Gate广场四月发帖挑战. $ETH
ETH3,97%
View Original
post-image
post-image
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • 10
  • 9
  • Share
Comment
Add a comment
Add a comment
GateUser-991fc58a
· 24m ago
The key still depends on whether we can pass NVIDIA's verification and supply schedule; no matter how tight the timeline is, it all depends on the yield rate.
View OriginalReply0
ApyDaydreamer
· 1h ago
NVIDIA standard = threshold; passing it is a passport.
View OriginalReply0
YieldYeti
· 2h ago
If HBM supply increases, the cost of AI computing power might really decrease a little bit.
View OriginalReply0
TransparentDomeCity
· 2h ago
Hopefully, no more "paper mass production"; actually delivering to major clients is the real win. Let's wait until next year to see the data.
View OriginalReply0
SeaSaltFlavorAirdrop
· 2h ago
I feel that HBM is the true hidden king of the AI arms race; whoever controls the production capacity is the more formidable.
View OriginalReply0
Yield慢炖锅
· 2h ago
Market level: With the release of HBM news, A-shares/Korean stocks in the semiconductor sector are about to experience another wave of sentiment.
View OriginalReply0
RetroRadioWaves
· 2h ago
#Gate Square By the way, just asking, on the ETH side, do you focus more on fundamentals or follow macro liquidity?
View OriginalReply0
YieldGardenKid
· 2h ago
Let the manufacturing department prototype the HBM core logic chip first; this step is very critical, and the packaging/staking coordination should not fail.
View OriginalReply0
FoldedYield
· 2h ago
Don't forget the bottlenecks like TSMC's CoWoS and advanced packaging capacity; having only strong memory isn't enough.
View OriginalReply0
Wax-SealedPrivateKey
· 2h ago
The AI memory track is going crazy.
View OriginalReply0
View More
  • Pin