Economy & Tech

Samsung Unveils HBM5 Memory Mock-Up at Computex 2026, Touting New Heat Tech

By K-Brief Editorial Desk /
Close-up of a stacked high-bandwidth memory chip module displayed on an exhibition stand
Editor’s Note for international readers

Why it matters. HBM is the memory that makes modern AI hardware possible, and Samsung is one of only a few firms in the world capable of making it — so its roadmap shapes the global AI supply chain.

Background. Samsung Electronics is South Korea's largest company and the world's biggest memory-chip maker, though it has trailed local rival SK hynix in recent HBM generations supplied to leading AI chipmakers. The 'DS division' refers to Device Solutions, Samsung's semiconductor arm, and 'technology gap' (chogyeokcha) is a phrase Korean firms use for a lead so wide that rivals cannot easily close it. Computex, held annually in Taipei, is Asia's premier computing trade show.

What to watch next. Watch for Samsung to announce an HBM5 mass-production timeline and to confirm which AI chipmakers will adopt it, key signals of whether it can regain ground on SK hynix.

Samsung Electronics unveiled a physical mock-up of its eighth-generation high-bandwidth memory, HBM5, at the Computex 2026 trade show in Taipei, Taiwan, on June 2. The reveal — presented by Song Jae-hyuk, president and chief technology officer of Samsung’s Device Solutions (DS) division — signals that development of the AI-focused memory is nearing completion and underscores the company’s drive to stay ahead in the race for advanced AI memory.

What Was Shown

A mock-up is a non-functional physical model of a product, typically displayed when a design is largely finalized but before mass production begins. By bringing HBM5 to one of Asia’s largest IT exhibitions, Samsung is publicly staking a claim to technological leadership in a market where memory chips have become a critical bottleneck for artificial intelligence systems.

High-bandwidth memory, or HBM, stacks multiple memory chips vertically to move data far faster than conventional memory. It has become essential to the AI boom because it feeds data to the powerful processors that train and run large AI models. Each new generation pushes performance higher, and HBM5 represents the eighth iteration in that lineage.

The Heat Problem — and Samsung’s Answer

The headline feature of the HBM5 mock-up is a thermal-management technology Samsung calls HPB (Heat Path Block). As AI memory grows more powerful, it also generates more heat, which can throttle performance and threaten stability. HPB is designed to spread and release that heat more efficiently, helping the chip run reliably at higher speeds.

Song said the company had already validated the heat-management approach in its seventh-generation product, HBM4E. “In HBM5, we added a heat-transfer path to lower thermal resistance and improve operational stability,” he explained. Lower thermal resistance means heat escapes the chip more easily — a key advantage as memory stacks become denser and hotter.

Why It Fits Samsung’s Strategy

For Samsung, the demonstration is as much about signaling intent as about the hardware itself. The company has framed its memory roadmap around the idea of a “technology gap” — staying far enough ahead of rivals that competitors struggle to catch up. Putting an advanced HBM5 model on a global stage, complete with a named heat-management innovation, is a way of telling customers and competitors alike that its next-generation product is on track.

The AI memory segment has become one of the most lucrative and closely watched corners of the semiconductor industry, with demand driven by data centers building out AI infrastructure. Heat dissipation is widely seen as one of the central engineering challenges for the next wave of HBM, making thermal features like HPB a meaningful point of differentiation.

Samsung did not announce a mass-production timeline for HBM5 at the event, and the mock-up represents a design milestone rather than a shipping product. Still, the reveal offers an early look at how the company intends to tackle the performance and heat hurdles that define the next stage of AI memory.