Global Chip Sales Just Smashed Every Record With a $403 Billion Quarter

$403.3 billion in a single quarter. That is not a forecast or a target. That is what the global semiconductor industry actually sold in Q2 2026.

The Semiconductor Industry Association reported quarterly sales figures that would have seemed unrealistic two years ago. Revenue climbed 35.1% from Q1, and the year-over-year growth number for June alone was 123.6%. For an industry measured in the trillions of transistors, that kind of growth rate is almost surreal.

The companies driving the numbers

AMD reported $11.5 billion in Q2 revenue, up 50% year-over-year, with data centre demand more than doubling. The AMD Ryzen 9 9950X continues to be one of the most popular enthusiast processors, but the real money is in server chips. AMD's EPYC and Instinct lines are feeding the same AI infrastructure buildout that is reshaping the entire industry.

Astera Labs hit $392.4 million in revenue, up 104% year-over-year. They make the connectivity chips that link GPUs together in AI data centres. When Nvidia sells a rack of H100s, Astera Labs sells the glue that makes them work as a single system. That is a business that did not exist at this scale three years ago.

SMIC, China's largest chipmaker, is raising prices. When foundries raise prices, it means demand is outstripping capacity. That is happening globally, not just in AI.

Why these numbers are not normal

A 123% year-over-year jump in June is extraordinary for an industry this large. Semiconductors are a $1.6 trillion annual market. Growth rates like this are more common in startups, not in global industrial supply chains.

The driver is AI infrastructure. Every major cloud provider is building out GPU clusters as fast as they can source the chips. Samsung committed $648 billion to AI chip infrastructure earlier this year. Tesla and SpaceX are building a terafab in Texas that will consume chips at an industrial scale. The optical networking layer connecting these data centres is scaling up alongside them.

This is not just a software story anymore. AI is physically reshaping manufacturing, logistics, and capital allocation across the semiconductor supply chain.

The capacity crunch is real

The entire industry is running near capacity, and new fabs take three to five years to build. TSMC, Samsung, and Intel are all investing tens of billions in new fabrication plants, but those will not produce chips until 2028 or 2029 at the earliest. In the meantime, the companies that control existing capacity have pricing power over every industry that depends on chips.

Which is, increasingly, all of them. Automotive, medical devices, telecommunications, consumer electronics, and defence all compete for the same foundry slots as AI training clusters. When an AI company buys up wafer capacity for the next two years, that is capacity that is not available for car manufacturers or smartphone makers.

The semiconductor industry has been cyclical for decades, booming and busting on roughly four-year cycles. Whether AI demand has broken that cycle or is simply making the current boom taller before a steeper fall is the question nobody can answer yet.

What the numbers do tell us: Q2 2026 is the biggest quarter the chip industry has ever had. And Q3 is on track to be bigger.

Some links in this article are affiliate links. We may earn a small commission at no extra cost to you.