You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
Future Horizons’ July 2026 semiconductor market update describes an extraordinary but highly concentrated industry boom. May 2026 global semiconductor sales rose 118.4% year on year and 18.1% month on month, marking the thirty-third consecutive month of annual growth. However, the report argues that headline growth is misleading because it is driven mainly by higher average selling prices rather than broad unit-demand expansion. Integrated circuits led the surge, especially memory, up 320.9%, and logic, up 49.2%, largely because of demand from AI hyperscalers for GPUs, processors, and memory.
The authors caution that traditional sectors—including discretes, analog, microcontrollers, optoelectronics, and sensors—remain weak amid a subdued global economy. Although AI may eventually transform productivity, reduce costs, and accelerate product development, such structural changes usually take longer than expected and often evolve differently from their earliest forms. The report also stresses that every previous semiconductor downturn ultimately resulted from excess supply, caused either by overinvestment or weaker demand. As new DRAM capacity enters production, oversupply could reverse memory price growth and trigger a downstream correction. Therefore, despite industry consensus forecasts for two more years of uninterrupted expansion, Future Horizons views the current ASP-led, AI-dependent upswing as historically unusual, potentially unsustainable, and vulnerable to correction.
Unit sales aren't increasing because the memory vendors are selling everything they manufacture, but unit demand would increase if supply increased. It seems obvious prices wouldn't go down as much as unit sales would go up, so revenues would still go a lot higher, IMO.
Malcolm & his team have been preaching the ASP-led market expansion for quite a while now. And to be honest I tend to agree.
The industrial and automotive demand is still low and resales will remain under high pressure from additional supply coming online in China.
The markets are in for a tough ride as soon as the short-term data center demand cools down.
Unit sales aren't increasing because the memory vendors are selling everything they manufacture, but unit demand would increase if supply increased. It seems obvious prices wouldn't go down as much as unit sales would go up, so revenues would still go a lot higher, IMO.
When ASPs rise too aggressively, system companies will give them some responses.
They will redesign the architectures, extend the product lifecycles, find some alternative suppliers, or move to lower-cost solutions.
This is especially true in automotive, industrial, and consumer markets. Because their BOM cost pressure is extremely high.
