Nvidia Certifies SK Hynix, Samsung, Micron for HBM4 Production
All three dominant memory chipmakers have received approval to mass-produce the latest high-bandwidth memory for AI accelerators.
Nvidia has given the green light to all three major memory manufacturers to supply HBM4 chips for its AI accelerators, CEO Jensen Huang announced this week.
SK Hynix, Samsung Electronics, and Micron Technology have each completed Nvidia's certification process and can now begin mass production of the fourth-generation high-bandwidth memory. The approval marks a significant milestone in the intensely competitive market for advanced memory components that power AI computing.
The HBM4 qualification race
The three companies collectively dominate the global market for storage semiconductors used in computing systems. Each has invested heavily in developing HBM4 technology and competed aggressively for Nvidia's certification, which effectively serves as a gateway to the lucrative AI accelerator supply chain.
High-bandwidth memory differs from conventional DRAM by stacking multiple memory dies vertically and connecting them with through-silicon vias, enabling dramatically faster data transfer rates. These speeds are essential for AI training and inference workloads, where moving data between processors and memory often becomes the primary bottleneck.
HBM4 represents the latest evolution of this technology, offering improvements in bandwidth, capacity, and power efficiency over the HBM3 and HBM3E generations currently in production.
Why it matters
Nvidia's certification decisions directly shape the economics of AI infrastructure. By qualifying all three major suppliers rather than selecting a narrower group, Nvidia may be positioning itself to secure more favorable pricing and avoid supply constraints as demand for AI accelerators continues to surge. For the memory manufacturers, HBM4 certification opens access to one of the semiconductor industry's fastest-growing revenue streams, though it also ensures they'll face continued competition from each other for Nvidia's orders.
Supply chain implications
The simultaneous certification of three suppliers suggests Nvidia anticipates robust demand for its next-generation AI products. Diversifying the supply base reduces risk from production issues at any single manufacturer and provides leverage in pricing negotiations.
For SK Hynix, Samsung, and Micron, the certification validates their technical capabilities but guarantees nothing about order volumes. Nvidia will likely allocate production based on factors including manufacturing yield, delivery timelines, and pricing as it ramps its next product generation.
Bloomberg first reported the certification announcement from Huang on June 5, 2026.
This is an original analysis by the Omega editorial team. Source reporting: AI Watch.
Want systems like this working for your business?
Book a Call