Intel confirmed late last week that CPU prices are going up by 10 to 15 percent effective immediately, with major PC manufacturers and OEMs receiving revised pricing at the end of March. AMD is following with its own increases starting in April. Both companies saw their stock prices surge over 7% on the news, because on Wall Street, the ability to raise prices without losing customers is a signal of market power even when the reason you can raise prices is that nobody else has any supply either.
The supply numbers are ugly. CPU delivery lead times, which sat at a comfortable one to two weeks for most of 2025, have stretched to eight to twelve weeks on average, with some orders now quoting six months. The bottleneck is not a manufacturing defect or a logistics hiccup. It is a structural reallocation of global semiconductor capacity toward AI infrastructure. Nvidia, Broadcom, Google, and Amazon are consuming fabrication capacity at a rate that leaves less room for the processors that go into the laptops, desktops, and servers that regular businesses and consumers buy. When Big Tech committed $650 billion to AI capex this year, that money had to come out of a finite supply chain. Now everyone else is finding out where the cost lands.
The Iran war is making it worse. Oil above $100 per barrel raises energy costs at every stage of semiconductor manufacturing, from the fabs in Taiwan and Arizona to the packaging facilities in Southeast Asia. The Strait of Hormuz closure and Qatar’s helium supply disruption are adding pressure to a supply chain that was already running at capacity before the first strike. TSMC is accelerating its Fab 21 expansion in Arizona, but new fab capacity takes years to come online and solves nothing for the price hikes hitting OEMs right now.
The three-way squeeze
Three forces are converging on the same supply chain simultaneously. AI demand is pulling fabrication capacity toward high-margin accelerators and away from consumer CPUs. The Iran war is raising energy and logistics costs across every node in the chain. And the geographic concentration of advanced semiconductor manufacturing, still overwhelmingly dependent on TSMC in Taiwan, means there is no backup capacity to absorb the pressure. The $2.5 billion chip smuggling operation Laterstack covered last week shows how desperate the demand for semiconductor supply has become. When the legitimate supply chain cannot keep up, the gray market fills the gap.
Consumer electronics price increases of 10 to 15 percent on finished products are expected to begin rolling through in the second half of 2026. That number assumes the war does not escalate further, that TSMC’s operations are not disrupted, and that AI demand does not accelerate beyond current projections. All three assumptions are optimistic.
The counter argument
CPU price increases of 10 to 15 percent are significant but not catastrophic, and the semiconductor industry has weathered supply crunches before without permanent structural damage. The 2020-2022 chip shortage produced similar panic and prices eventually normalized as new capacity came online and demand rebalanced. Intel and AMD both have incentives to communicate supply constraints publicly because it justifies price increases and supports their stock prices. The “AI ate your chips” narrative may overstate the actual capacity reallocation, given that AI accelerators and consumer CPUs often use different fabrication processes and compete less directly for fab time than the headlines suggest.
The AI subsidy era is over. For a decade, the implicit bargain of the consumer technology market was that hardware got cheaper every year because the scale of production kept costs falling. That bargain depended on fabrication capacity growing faster than demand, and AI just broke that equation. Nvidia guided $78 billion in revenue and every dollar of that guidance represents fabrication capacity that is not making your next laptop cheaper. The war makes it worse, the geographic concentration makes it fragile, and the timeline for relief, measured in years of new fab construction, means this is not a quarter or two of higher prices. This is the new cost structure for computing, and the companies driving it are the same ones telling you AI will make everything more efficient.
What This Means for Everyday People
Your next laptop, desktop, or server will cost more, and the reason is not inflation in the traditional sense. It is that the companies building AI systems are consuming so much of the world’s chip manufacturing capacity that there is less left over for everything else. The 10 to 15 percent increase on CPUs will work its way into every finished product that contains a processor by the end of the year. If you were planning a hardware purchase, the price is not getting better from here.