AMD Helios rack enters production with Samsung HBM4
Samsung Electronics is supplying high-bandwidth memory for AMD's Helios AI server rack as the system entered production to compete with Nvidia in the data center market, according to a Korea Herald report cited by ETCIO. AMD Chief Executive Officer Lisa Su announced the milestone on Thursday during the company's Advancing AI conference in Seoul. Su said shipments were scheduled to start late in the third quarter before expanding into the fourth quarter.
Memory capacity and rack architecture
Helios is built as a complete cabinet rather than individual chips, integrating 72 Instinct MI455X accelerators with central processors, networking gear, and cooling systems. The configuration supports up to 432 gigabytes of HBM4 per accelerator, totaling approximately 31 terabytes across the rack cabinet. Samsung's mass-produced HBM4 holds 36 gigabytes within a 12-layer stack, amounting to roughly twelve stacks per accelerator. By comparison, Nvidia's competing Rubin processor uses eight stacks, meaning each shipped AMD rack carries a higher volume of Samsung memory than a comparable Nvidia setup.
Performance claims and underlying technology
Su said Helios offers 50 percent more memory capacity and 6 percent higher bandwidth than Nvidia's Vera Rubin platform. ETCIO reported that the figures reflect AMD's internal specifications rather than third-party benchmark evaluations. During a Samsung-hosted session at the Seoul conference, senior engineer James An said HBM4 doubles the number of input-output connections to 2,000 from 1,000 in the previous generation, and that Samsung produced the logic layer beneath the memory stack on its own 4-nanometer foundry process.
Supplier agreement and customer roadmap
Samsung gained primary HBM4 supplier status for the MI455X under a contract signed in March at its Pyeongtaek chip facility, which Su visited during the agreement. Primary supplier status does not imply exclusivity, and neither firm disclosed pricing or volume allocation. Samsung initially began commercial HBM4 shipments in February and also supplies server memory modules for the sixth-generation EPYC "Venice" processors used in Helios. AMD announced hardware commitments from OpenAI, Oracle, and Anthropic, though actual deployments operate on staggered timelines extending through 2027.
Local governments weigh new data center restrictions
Local U.S. jurisdictions continued to scrutinize data center projects on July 24. GovTech reported that Oakley, Calif., leaders hinted at a permanent data center ban. WRVO reported that Auburn lawmakers were examining a data center moratorium despite a statewide one-year hold, and WABE reported that Clayton officials were holding further talks with residents over a 400-acre campus proposal in Lovejoy.
Share this article
