Get weekly AI video rankings, evaluation updates, and industry news delivered to your inbox.
Omdia's third upward revision in months reflects a supply chain under strain
Three months ago, Omdia forecast 2026 semiconductor revenue would grow 62.7% year over year. That figure is now 94.1%. The revision, Omdia's third upward adjustment in recent months, was published July 30 and reflects AI demand for DRAM and NAND memory that has repeatedly outrun the firm's own projections.
A 31-percentage-point swing in a single quarter suggests the industry is outpacing the analytical frameworks used to track it.
Memory is the center of gravity
Omdia projects memory ICs will account for more than half of total semiconductor revenue in 2026. That concentration reflects how AI infrastructure spending translates into chip demand. Large AI models and the data centers that run them require enormous volumes of DRAM and NAND, and that appetite has overwhelmed existing production capacity.
High Bandwidth Memory sits at the sharpest point of that constraint. HBM production is technically demanding, and only a small number of suppliers can manufacture it at scale. Advanced packaging capacity at TSMC is fully booked, and leading-edge fabrication nodes are similarly committed. Omdia expects these bottlenecks to persist through at least 2027.
The cost flows downstream
The supply crunch is not contained to AI buyers. Manufacturers of consumer electronics and smartphones compete for the same memory chips and packaging capacity. Omdia's data on smartphone shipments, published alongside the semiconductor forecast, showed global handset volumes fell 6% in the second quarter of 2026, a period when supply-side pressures were already forcing changes to procurement decisions across the industry.
Rising memory costs driven by AI demand are part of that pressure. When HBM and advanced packaging capacity are locked up by hyperscalers, the remaining supply for conventional applications tightens and prices rise. Consumer electronics markets are absorbing cost increases generated by AI infrastructure investment.
What the forecast does not resolve
Omdia's numbers describe demand. They do not adjudicate whether that demand is durable. The forecast rests substantially on continued AI infrastructure buildout, and the pace of that buildout depends on factors that sit outside semiconductor supply chains, including model efficiency gains, enterprise adoption curves, and regulatory shifts.
The firm's track record of upward revisions across a single year suggests the demand signal has been consistently underestimated. Whether the 94.1% figure itself gets revised again will be visible when Omdia publishes its next quarterly semiconductor update.
Get weekly AI video rankings, evaluation updates, and industry news delivered to your inbox.