IN Brief:
- Semiconductor revenue reached $319bn in Q1 2026 after a 27% sequential increase.
- DRAM and NAND revenue nearly doubled in one quarter, with memory taking more than 40% of market value.
- AI infrastructure demand is widening the gap between memory-linked growth and slower industrial component segments.
Omdia has recorded global semiconductor revenue of $319bn for the first quarter of 2026, with the market moving through one of its sharpest growth phases as memory pricing and AI infrastructure demand pull chip sales beyond normal cyclical patterns.
The first-quarter figure represents a 27% sequential increase from the final quarter of 2025, the highest quarter-on-quarter growth rate recorded since the first quarter of 2002. Omdia’s tracking also shows three consecutive quarters of double-digit growth, putting the first half of 2026 on a path that could exceed $700bn if momentum continues through the second quarter.
Memory is carrying the expansion. DRAM and NAND revenue nearly doubled in a single quarter, with combined memory products accounting for more than 40% of semiconductor revenue. That share is well above the long-term pattern, where memory has typically represented around one-fifth of total chip revenue, and it reflects the extent to which AI accelerators and data-centre infrastructure are reshaping demand.
NAND revenue alone approached $48bn, while average selling prices rose sharply as supply tightened around AI and server requirements. High-bandwidth memory, enterprise SSDs, and data-centre storage are now tied more closely to compute roadmaps, making memory availability a strategic constraint rather than a commodity purchasing line.
Beneath the headline growth, the market is becoming increasingly uneven. AI accelerators, high-bandwidth memory, server processors, networking devices, and storage-linked components are drawing investment and pricing power, while several non-memory segments are moving more slowly. Omdia’s figures show non-memory semiconductor revenue rising only slightly quarter on quarter, with microcontrollers, discretes, and optical components showing limited or negative movement.
That split is already visible across the production chain. Rising semiconductor equipment billings linked to AI demand show where capital is flowing, while first-quarter updates from suppliers such as Nexperia underline the uneven operating conditions still facing parts of the component market. A record revenue quarter does not automatically translate into broad relief across every device family.
The divergence complicates planning for industrial, automotive, medical, and embedded platforms, where lifecycle expectations differ sharply from data-centre buying cycles. Long-life systems still need predictable availability across controllers, power devices, analogue components, passives, sensors, and memory, yet investment attention is being pulled towards the highest-growth AI-linked segments. Procurement teams may face a market that looks buoyant in aggregate while remaining tight or sluggish in specific categories.
Memory’s stronger role also changes the design economics of compute systems. HBM is no longer an accessory to the processor; it is a defining part of accelerator performance. Its availability influences board architecture, package selection, thermal design, substrate demand, and system cost, while NAND pricing affects storage-heavy data-centre deployments and edge infrastructure.
Electronics manufacturing is already absorbing those pressures, with AI server-board demand increasing the need for advanced inspection, high-density assembly, and more capable production equipment. The recent rise in AI server board demand at ASMPT points to the same pattern further down the chain, where memory-led growth translates into board-level complexity and manufacturing investment.
The semiconductor market is expanding rapidly, but the gains are concentrated around AI compute, memory, and data-centre infrastructure. That concentration gives the industry its current momentum while leaving a more complicated picture for the broader electronics base, where the next cycle may be shaped less by total chip revenue than by which components receive priority, capacity, and capital.
