The headline number everyone is quoting is the big one: global semiconductor revenue topped $425 billion in the second quarter of 2026, a record. That figure is real, and it is worth a moment. But it is not the important number in Omdia's report. The important number is the one buried two paragraphs down, and it is the kind of number that changes who wins and who loses in the industry.
Memory chips, the DRAM and NAND flash that every computer, phone, and AI accelerator depends on, now account for more than half of all global semiconductor revenue. For the first time in the market's tracked history, a single component class has crossed the 50% line. That is not a headline. It is a structural break in how the industry is organized, and it is about to show up in the price of the devices that buy the memory.
The facts at list price
Omdia, the technology research firm that has tracked the semiconductor market since the first quarter of 2002, put total 2Q26 revenue above $425 billion, up a record 31.4% from the quarter before, according to its new quarterly research. The first half of 2026 came to $752 billion. The 31.4% quarter over quarter beat the previous sequential growth record of 29.2%, set just one quarter earlier in 1Q26.
The growth is historically rare. Since Omdia began tracking the market, only 10 out of 97 quarters have recorded sequential growth above 10%. The last four quarters, starting in the third quarter of 2025, have all grown by double digits, and Omdia expects the third quarter of 2026 to keep that streak alive.
The engine is memory. Memory ICs alone made up more than half of all semiconductor revenue in 2Q26. DRAM, NAND, and NOR each posted their highest sequential growth for a second quarter since Omdia began tracking, and all three hit their highest recorded quarterly revenue in the period. Omdia is explicit about the mechanism: AI demand is reshaping how memory suppliers allocate production, which tightens the supply and demand balance, which pushes up average selling prices and revenue.
Stripping memory out, the rest of the market still grew more than 10% quarter over quarter, well above the roughly 3% that is typical for a second quarter based on Omdia's data from 2002 to 2025. Microprocessor revenue grew 16% quarter over quarter, against a seasonal norm of about 1%. So this is not purely a memory story. But memory is the engine, and the rest of the market is being dragged along by it.
Working the numbers
The headline is a revenue figure, so let's turn it into something concrete. If memory is more than half of a $425 billion quarter, memory chips did more than $212 billion in revenue in a single quarter. That is more than the entire revenue of most of the companies in the S&P 500, and it is one component class.
Working backwards from the $752 billion first half and the 31.4% growth rate, the first quarter of 2026 was roughly $325 billion and the second quarter roughly $427 billion. (That is my own calculation from Omdia's two stated figures, not a number Omdia printed.) The gap between the two quarters, about $100 billion in one quarter, is the size of the memory swing.
Omdia forecasts revenue will top $500 billion in the third quarter of 2026, which would put the first three quarters of the year above $1.25 trillion. It says that figure would be 50% higher than the entire 2025 total. Working backwards from that, 2025's full year was roughly $830 billion. So the first three quarters of 2026 are on pace to exceed the whole of last year by a wide margin. That is the scale of what is happening.
The most telling comparison in the report is the one Omdia makes about seasonality. A normal second quarter for non memory semiconductors grows about 3%. This one grew more than 10%. A normal second quarter for microprocessors grows about 1%. This one grew 16%. Every category is running at several times its seasonal pace, which means the whole market is being lifted, not just the memory corner.
The cost the pitch leaves out
Here is where the record revenue number stops being a celebration and starts being a warning. Omdia is explicit that the memory growth is driven by higher average selling prices, not by a corresponding jump in the number of chips sold. When a category's revenue rises because the price rose, that is a cost push, not a demand boom. And the cost lands on whoever buys the chips.
The downstream data in the same Omdia research shows exactly where it is landing. In the United States, average PC sell in prices crossed $1,000 for the first time, up 12% year over year, while PC shipments barely moved, up just 1% to 18.8 million units, according to coverage of the report. Omdia attributes the small shipment uptick to retailers pulling inventory forward ahead of expected price increases from memory and storage supply constraints. That is not demand. That is a one time pull forward.
And the forecast is the tell. Despite the price jump, Omdia still expects the full year 2026 US PC market to decline 10.7%. Prices are up 12%, volume is forecast down 10%. That is the signature of a market where the input cost is squeezing the buyer, not a market where buyers are lining up. The same force that pushed memory revenue to a record is about to shrink the device market that consumes the memory.
The pattern repeats elsewhere in the report. Global TV shipments grew 3.6% year over year in the quarter, but Omdia flags that consumer inflation and tightening memory supply are raising cost pressures across the TV industry. The extended reality headwear market fell 38.7% year over year in the first half, its fourth consecutive annual decline. The devices that buy memory are not growing. The memory is.
Our read
The 50% threshold is the real story, and it deserves a clear position. When a single component class crosses half of an entire industry's revenue, the industry's fortunes become coupled to that one class. Every downstream device maker, from PC and TV to phones and consoles, is now a price taker on memory. The memory makers set the price, and everyone else absorbs it.
Our position is that this is a cost push supercycle, and the winners and losers are already sorted. The winners are the memory suppliers and the equipment makers that supply them. The losers are the device original equipment manufacturers who cannot pass the price through to a consumer who is already feeling inflation. The record $425 billion quarter will make chip stocks look strong, and they should. But the more forward looking signal is the PC data: price up, volume forecast down. That is the shape of a market where the input cost is doing the work, not the demand.
Two caveats keep us from calling this a breakout. First, Omdia is an analyst firm, not the chip makers, so these are estimates and forecasts, not reported results. The $500 billion third quarter is a projection, not a number that has happened yet. Second, a revenue record driven by price rather than volume is a fragile kind of record. If the price that is inflating memory revenue is the same price that is shrinking the PC and TV market, then the two numbers are two sides of the same coin, and the coin is about to land on the consumer electronics side.
The number to watch is not the next quarterly chip revenue figure. It is the full year 2026 US PC forecast of minus 10.7%. That is the canary. If memory keeps taking a bigger share of chip revenue while the devices that buy it keep shrinking, then the record revenue is the top of the cycle for everyone downstream, even as it is a banner year for the memory makers.
See also: The 2026 Memory Shortage: HBM4's Hidden Cost Curve