The semiconductor industry apparently looked at its previous records and decided they were merely suggestions.
Global semiconductor revenue climbed past $425 billion in the second quarter of 2026, according to Omdia, setting yet another quarterly high. More strikingly, revenue jumped 31.4% from the previous quarter, breaking the sequential growth record set just three months earlier.
That pushed semiconductor revenue for the first half of 2026 to approximately $752 billion.
AI remains the obvious protagonist in this story, but the numbers reveal something more interesting than simply “companies are buying lots of GPUs.” Memory prices are soaring, suppliers are rearranging production around AI hardware, and even chip categories outside memory are growing far faster than their usual seasonal patterns.
This is increasingly less of an AI accelerator boom and more of an AI infrastructure boom consuming large portions of the semiconductor supply chain.
Semiconductor growth is entering very unusual territory
Quarter-to-quarter semiconductor revenue normally does not move like this. Omdia has tracked the market quarterly since 2002. Across 97 quarters, only 10 have delivered sequential revenue growth above 10%.
Yet the industry has now produced four consecutive quarters of double-digit growth, beginning in the third quarter of 2025. The second quarter’s 31.4% increase also surpassed the previous record sequential gain of 29.2% in Q1 2026. Omdia expects Q3 to continue the double-digit streak.
For context, Omdia was already calling the start of 2026 historically unusual when first-quarter semiconductor revenue crossed $300 billion. At the time, memory revenue had surged by more than 80% sequentially.
Three months later, the industry accelerated again. Calling this a recovery cycle feels increasingly inadequate.
Maybe related: YMTC Explained: Can China Really Overtake Samsung in NAND?
Memory has become the industry’s giant revenue engine
The main reason semiconductor revenue is exploding is memory. DRAM, NAND flash, and NOR memory all reached their highest quarterly revenue levels on record during Q2, according to Omdia. Memory ICs collectively accounted for more than half of total semiconductor revenue during the quarter.
AI is reshaping where manufacturers put their production capacity.
High-bandwidth memory, or HBM, has become particularly important because modern AI accelerators require enormous amounts of very fast memory located close to the processor.
The problem is that producing HBM is substantially more complicated than manufacturing ordinary DRAM. It consumes manufacturing capacity while also requiring sophisticated packaging.
Omdia says HBM supply remains constrained, with SK hynix, Samsung, and Micron being the three suppliers able to manufacture the technology at meaningful scale. Advanced packaging capacity is tight as well, adding another bottleneck to the AI hardware supply chain.
The result is predictable. Demand goes up. Capacity remains constrained. Prices go up. Revenue goes through the roof. Semiconductor economics occasionally does an excellent impression of Econ 101.
Rising prices matter as much as rising demand
There is an important caveat hiding inside those enormous revenue numbers. A 31% revenue increase does not mean manufacturers suddenly shipped 31% more chips.
A substantial portion of the boom is being driven by higher average selling prices, particularly in memory.
Omdia has previously warned that the extraordinary semiconductor expansion in 2026 is being driven heavily by pricing rather than purely by rising shipment volumes. Tight DRAM and NAND supply has allowed manufacturers to charge significantly more for available capacity.
That makes this semiconductor cycle quite different from a conventional consumer electronics boom. The industry isn’t necessarily producing vastly more smartphones, laptops, and gadgets. Instead, data centers are demanding more expensive silicon, more memory per system, more sophisticated packaging, and more advanced processors.
Each server rack has effectively become a much larger semiconductor shopping cart.
This isn’t only a memory story
Memory is so dominant that it risks making everything else look ordinary. It isn’t. Strip DRAM and NAND out of the numbers and the broader semiconductor market still grew by more than 10% quarter over quarter in Q2.
Historically, non-memory semiconductor revenue grows only slightly above 3% during a typical second quarter, according to Omdia’s data going back to 2002.
So even without the spectacular memory cycle, semiconductor demand remains unusually strong. Microprocessors are a particularly good example. MPU revenue climbed 16% sequentially, compared with typical second-quarter growth of roughly 1%. That is a dramatic divergence from normal seasonality.
AI infrastructure is spreading demand across the chip industry
Why are processors outside GPUs suddenly performing so well? Because an AI data center requires far more than AI accelerators. Building modern AI infrastructure also means buying:
- server CPUs
- networking silicon
- memory controllers
- storage components
- power-management chips
- high-speed interconnects
- switches
- optical components
- advanced packaging
The accelerator may get the keynote presentation, but an NVIDIA GPU cannot run an entire data center by itself.
Omdia has previously described data-center servers as the dominant driver of semiconductor growth in 2026, with cloud giants such as Google, Microsoft, and Amazon continuing large infrastructure expansions. That explains why demand is increasingly spilling outward from GPUs and HBM into seemingly less glamorous categories.
AI is becoming an entire semiconductor ecosystem rather than a single hot product category.
The industry could cross $500 billion in a single quarter next
Apparently $425 billion is not enough. Omdia forecasts semiconductor revenue will exceed $500 billion in Q3 2026. If that happens, total industry revenue for the first nine months of the year would surpass $1.25 trillion.
That would be approximately 50% higher than semiconductor revenue for the entirety of 2025. Let that sink in for a moment. Nine months of 2026 could generate one-and-a-half times as much semiconductor revenue as all twelve months of the previous year.
Earlier this year, Omdia expected the semiconductor market simply to cross the $1 trillion annual threshold for the first time in 2026. Its forecasts have since been revised dramatically upward as AI infrastructure spending and memory pricing repeatedly exceeded expectations.
In January, Omdia projected roughly 31% annual semiconductor growth. By July, its forecast had jumped to 94.1%. Forecast models are having quite a year too.
The bigger question is how sustainable this becomes
Explosive semiconductor revenue is obviously good news for chip suppliers. But the underlying dynamics deserve some caution.
Much of the industry is currently being pulled forward by extraordinarily high AI infrastructure investment. Hyperscalers are spending hundreds of billions of dollars expanding data-center capacity, while memory and advanced-packaging supply remain constrained.
That creates two major questions. First, how quickly can manufacturers add enough capacity to normalize pricing? Omdia expects bottlenecks involving HBM, advanced packaging, and leading-edge manufacturing capacity to persist into at least 2027.
Second, will AI services ultimately generate enough economic value to justify this extraordinary wave of capital expenditure? For now, the industry clearly believes the answer is yes.
But semiconductor history is filled with cycles where shortages encourage enormous capacity expansion, only for the balance between supply and demand to eventually swing the other way.
Memory manufacturers in particular know that movie rather well.
AI has turned semiconductors into the industry’s hottest infrastructure market
The remarkable part of the current semiconductor boom isn’t simply that Nvidia and other AI-chip companies are selling expensive accelerators. AI is changing the economics of the broader chip industry.
Memory suppliers are redirecting capacity toward high-value products. Advanced packaging lines are filling up. CPUs and networking silicon are benefiting from data-center expansion. Tight supplies are pushing prices higher across multiple categories.
And despite memory accounting for more than half of industry revenue, the rest of the semiconductor market is still growing at more than triple its normal second-quarter pace.
That makes the $425 billion figure more than another impressive AI statistic. It suggests that the AI spending boom has spread deep enough into the supply chain that the entire semiconductor market is now operating under conditions rarely seen in more than two decades of data.
If Omdia’s Q3 forecast holds, the next milestone is already visible: the world’s semiconductor industry could soon become a half-trillion-dollar-per-quarter business. Apparently trillion-dollar annual markets are so early 2026.
