High-bandwidth memory will see its average selling price climb 121% next year, according to a report by South Korea's Maeil Business Newspaper. The projection puts a hard number on what buyers of advanced memory already know: supply cannot keep up with demand.

The figure refers to the average selling price of HBM products across the market, not a single contract or a single supplier. A jump of that magnitude in one year is rare even by the standards of the memory industry, where pricing cycles swing hard in both directions. The driver, according to the report, is a straightforward supply shortage.

Why is HBM supply falling short?

HBM production competes directly with conventional DRAM for wafer capacity. Manufacturers must dedicate advanced fabrication lines to HBM stacks, and the packaging steps required — stacking multiple DRAM dies and bonding them into a single module — add bottlenecks that standard DRAM does not face. When demand for HBM surges, capacity cannot expand quickly, because both the wafers and the advanced packaging lines are constrained.

That imbalance is now showing up in pricing. A 121% rise in average selling price means buyers will pay more than double, on average, what they pay today for equivalent HBM throughput.

What does the price increase signal?

Sharp price movements in memory markets typically indicate that demand has outrun planned capacity. In this case, the shortage implies that customers committing to HBM volumes for next year are competing for a limited pool of supply, and suppliers are pricing contracts accordingly.

For memory manufacturers, a rising average selling price translates directly into higher revenue per wafer. The same fab capacity, directed at HBM rather than conventional DRAM, yields substantially more income when ASP more than doubles. That arithmetic gives producers a strong incentive to keep shifting output toward HBM — which in turn tightens supply of standard DRAM products elsewhere in the market.

For buyers, the consequence is budget pressure. System vendors that specify HBM in their products will face materially higher memory costs next year, and those costs will either compress margins or pass through to end prices.

Who is exposed to the increase?

Any designer of systems that require very high memory bandwidth falls within the blast radius of a 121% ASP increase. That includes processors for artificial intelligence workloads, high-performance computing platforms, and advanced networking and graphics products — the categories for which HBM exists in the first place.

The buying side of this market is concentrated. A small number of large system houses place the bulk of HBM orders, and their procurement volumes effectively set the demand curve. When several of them expand their HBM requirements at once, as the current shortage suggests, suppliers cannot rebalance capacity fast enough to hold prices flat.

What happens next?

Price signals of this size usually do two things. First, they pull additional capacity into the market as manufacturers expand HBM production lines and qualify more advanced packaging capacity. Second, they force buyers to lock in long-term supply agreements, since guaranteed allocation becomes more valuable than spot pricing when product is scarce.

Both responses take time. New fab capacity and expanded packaging lines operate on multi-year timelines, which means the supply gap that pushes average selling prices up 121% next year cannot close quickly. Buyers should expect tight conditions and elevated pricing to persist through the period covered by the projection.

The report from Maeil Business Newspaper, published in Seoul, gives the industry a concrete reference point for planning: memory costs for high-bandwidth applications will more than double next year under current supply conditions. Procurement teams, product managers, and financial analysts modeling semiconductor costs for the coming year now have a single number to build their assumptions around — 121%.