Industry analyst Claus Aasholm on why traditional supply-demand logic no longer explains what buyers are experiencing.
The global semiconductor market is moving into a phase where traditional supply-demand logic no longer explains what buyers are experiencing on the ground. According to Claus Aasholm, Founder of Semiconductor Business Intelligence and one of the industry's most followed independent analysts, the market hasn't broken—it has entered a fundamentally different equilibrium.
Aasholm's work is used by semiconductor suppliers, OEMs, and institutional investors to understand how pricing power shifts, how capital is allocated across memory and foundry nodes, and where structural supply constraints will emerge before they become visible in channel data.
In a recent conversation with Andrew Czuczwa, Market Research Manager at Fusion Worldwide, Aasholm outlined two dynamics reshaping the semiconductor landscape: the growing mismatch between analog and digital foundry capacity, and the reality that AI demand is driving a durable infrastructure build-out rather than a speculative spike.
Czuczwa brings buyer-side visibility from supporting thousands of active sourcing teams navigating these shifts in real time. When the analyst perspective and ground-level procurement behavior align, the signal becomes clear.
The Foundry Capacity Paradox: Plenty of Supply Where It's Not Needed
Aasholm describes a widening split between analog/mature nodes and memory/high-performance compute nodes—a divergence that undermines any narrative about "the semiconductor industry" expanding or contracting as a unified whole.
While analog and mixed-signal manufacturers expanded aggressively earlier in the decade, utilization collapsed during the downturn and has not recovered to historical norms.
"Some analog companies more than doubled their capacity earlier this decade. But utilization dropped hard during the downturn. Renesas was down near 33% and has only climbed back to around 50%. That's still very low."
— Claus Aasholm, Founder, Semiconductor Business Intelligence
This excess capacity stands in stark contrast to memory production, where demand for high-bandwidth memory (HBM) continues to escalate faster than capacity can adjust.
"Inside DRAM, there's only one priority today, and that's HBM. Each new generation requires more wafer capacity to make the same amount of bits. And there's only one place to take that capacity from — DDR5 and DDR4."
— Claus Aasholm, Founder, Semiconductor Business Intelligence
The paradox is structural: we have spare capacity in the wrong part of the industry.
This means buyers cannot rely on broad "capacity expanding" narratives to inform sourcing decisions. The relevant question has become: capacity for what products, on which process nodes, controlled by which manufacturers?
The AI Shift: A Structural Build-Out, Not a Bubble
Much of the public debate around AI investment focuses on valuation, hype cycles, and speculative demand. Aasholm's assessment is direct: this is not a repeat of the dot-com era.
"People compare this to the dot-com bubble, but back then there were no revenues. This time, the demand is real, the production is real, and the profits are real. We're in the early innings."
— Claus Aasholm, Founder, Semiconductor Business Intelligence
According to Aasholm, this cycle is grounded in deployed systems, measurable workloads, and revenue-generating infrastructure—not speculative positioning ahead of uncertain demand.
The consequence is that memory demand no longer responds to price pressure the way it once did. In traditional DRAM cycles, rising prices eventually cooled PC and handset upgrades as buyers deferred purchases. Today, memory demand is set by hyperscaler infrastructure roadmaps, not consumer elasticity.
This fundamentally shifts the basis of procurement from price prediction to strategic positioning.
What This Means for Supply Chain Strategy
Czuczwa's perspective from the open market distribution channel reinforces the structural nature of these shifts. Operating across multiple geographies and end markets, Fusion Worldwide functions as an early-warning system for supply-side changes—occupying the space between manufacturer allocation decisions and buyer procurement needs.
This positioning provides visibility into how sourcing behavior shifts before those changes show up in published market data. When qualification timelines accelerate, when multi-sourcing strategies move from contingency plans to operational requirements, or when design teams begin factoring component availability into architecture decisions earlier in development cycles, these signals emerge first in the distribution channel.
The implications for OEMs, contract manufacturers, and sourcing teams are becoming operational realities:
Price elasticity has weakened. Waiting for historical price "resets" may mean missing allocation entirely. The assumption that time always favors the buyer no longer holds in segments where supply is structurally constrained.
Memory strategy now follows compute strategy. Procurement must align with AI system rollout timelines, not market cycles. Component planning increasingly mirrors infrastructure deployment schedules rather than responding to spot pricing.
Mature-node availability is not a fallback for constrained memory supply. The bottleneck is architectural, not operational. Excess analog capacity does not solve tightness in advanced memory nodes, they serve different functions in different systems.
Vendor roadmaps now determine competitive advantage in supply assurance. Organizations that understand manufacturer priorities and product mix shifts can position themselves ahead of allocation decisions rather than reacting to them.
A Market That Has Entered a New Equilibrium
The market is not misbehaving. According to Aasholm, it has simply entered a new equilibrium, one where AI infrastructure demand anchors the memory market differently than PC and smartphone upgrades once did, and where capacity abundance in mature nodes coexists with persistent tightness in advanced memory.
For procurement teams and supply chain strategists, the challenge is recognizing when the rules governing market behavior have changed, and adjusting sourcing approach accordingly before those changes become constraints.
The organizations adapting their procurement strategy now, based on structural shifts rather than historical cycle patterns, are positioning themselves for sustained component access in a market that no longer corrects the way it used to.
Access Ongoing Market Intelligence
Understanding these shifts requires ongoing visibility into how supply-side decisions translate into component availability. Fusion Worldwide's monthly Greensheet provides procurement teams with analysis across memory, analog, and digital markets—the kind of intelligence that helps sourcing teams recognize structural changes before they become allocation constraints.
Create a free account to access monthly Greensheet insights.
For a deeper technical perspective, revisit our full interview transcript with Claus Aasholm and Andrew Czuczwa, explore Claus’s ongoing research on LinkedIn and Semiconductor Business Intelligence, and see how these insights connect to the dynamics outlined in our first article with Claus, AI Sets the Price: Why DRAM Shortages Are Rewriting Memory Market Economics.
FAQ
Why is analog capacity underutilized while memory supply is still tight?
Because earlier-decade capital expansion happened in analog and mixed-signal nodes, not in advanced memory. AI workloads are pulling wafer capacity toward HBM, leaving DDR4/DDR5 structurally constrained even as analog sits half-utilized.
What makes today’s AI-driven demand different from past semiconductor cycles?
Traditional DRAM cycles cooled when prices rose and consumers delayed upgrades. Today’s demand is driven by hyperscaler infrastructure roadmaps, which do not slow down when prices increase. This removes the “self-correcting” mechanism buyers relied on in the past.
Is AI demand actually a bubble?
Claus Aasholm’s view is no. The demand is backed by deployed systems, real workloads, and measurable revenue — not speculative hype. The imbalance stems from where capacity exists and where it does not.
How should procurement teams adapt to this new equilibrium?
Waiting for historical price resets is risky. Buyers need earlier alignment with manufacturer roadmaps, parallel sourcing strategies, and closer monitoring of how wafer capacity is being redistributed between memory and compute.
