Is the semiconductor industry growing now?
In our semiconductor industry deck, you will find everything you need to understand the market
SUMMARY
Yes. The semiconductor industry is growing now, and the expansion is strong enough to show up across revenue, physical production, factory utilization and investment rather than in AI-chip earnings alone.
The headline growth rate is much more spectacular than the underlying increase in chip volumes. Memory prices have exploded, so semiconductor revenue is currently rising far faster than the amount of silicon physically moving through the industry.
AI is still the biggest accelerator. GPUs, custom AI accelerators, networking silicon, high-bandwidth memory and leading-edge foundry capacity are all expanding together, which makes the current cycle much broader inside AI infrastructure than a simple Nvidia story.
Consumer electronics are moving in the opposite direction. PC and smartphone shipments are falling, yet semiconductor revenue is surging, showing how a relatively small number of extremely expensive AI systems can now outweigh weakness across hundreds of millions of consumer devices.
The recovery is also spreading underneath AI. Automotive, industrial, analog and embedded semiconductor suppliers have returned to growth after the inventory correction that hurt those markets during the previous downturn.
Manufacturing data support the idea that this is a real production cycle. Silicon wafer shipments are increasing, advanced-node revenue is rising, mature-node fabs are highly utilized and semiconductor-equipment demand is reaching record levels.
Memory is both the strongest part of the boom and the obvious source of fragility. AI has structurally increased demand for high-value memory, but revenue growth above 300% cannot be separated from extraordinary pricing that will eventually normalize.
The industry's customers are behaving as though demand will remain strong. Hyperscalers are committing enormous amounts of capital to computing infrastructure while chipmakers and equipment suppliers are expanding capacity for 2027 and beyond.
That investment creates the next risk. Semiconductor shortages today and semiconductor oversupply later are not contradictory: the stronger this boom becomes, the more aggressively the industry builds the capacity that could eventually weaken pricing.
Our conclusion is that the semiconductor industry is in a genuine growth cycle, with AI providing most of the acceleration but no longer standing alone. The key question is now whether demand can absorb the capacity being added before memory pricing normalizes and the familiar semiconductor supply cycle catches up.
This market map, featured in our semiconductor industry deck, highlights top companies and startups in the semiconductor industry
Is semiconductor industry revenue really growing right now?
Yes. Global semiconductor revenue is growing extraordinarily fast right now.
The latest World Semiconductor Trade Statistics data put worldwide semiconductor sales at about $702 billion for the first half of 2026, up 102% from the same period a year earlier. For perspective, the industry generated $795.6 billion during all of 2025. In six months, chipmakers had already sold roughly 88% of the previous full year's total.
The most recent monthly data point is just as strong. The Semiconductor Industry Association reported $403.3 billion of global chip sales in the second quarter, 35.1% more than in the first quarter. June sales reached $134.5 billion, up 123.6% from a year earlier and 9.7% from the previous month.
This comes after an industry that had already recovered from its 2023 downturn. Global semiconductor sales went from roughly $527 billion in 2023 to $631 billion in 2024 and then $796 billion in 2025. The current jump sits on top of that recovery rather than simply reversing one bad year.
WSTS's latest data also show that memory revenue increased 305% year over year and logic increased 45%, while the other major semiconductor categories were positive as well. At the industry level, there is very little doubt left: semiconductor sales are booming today.
| Period | Global semiconductor sales | Change |
|---|---|---|
| 2023 | ~$527B | Downturn year |
| 2024 | ~$631B | Recovery |
| 2025 | $795.6B | +26.2% |
| H1 2026 | ~$702B | +102% YoY |
Are semiconductor companies selling more chips, or are prices just exploding?
Semiconductor volumes are rising too, but higher memory prices are making today's revenue growth look much bigger than the physical increase in chip production.
Micron gives us the cleanest example. In its latest reported quarter, DRAM revenue jumped 67% from the previous quarter. DRAM bit shipments increased only in the low-single-digit range, while average selling prices rose in the low-60% range. NAND revenue nearly doubled, yet bit shipments increased only in the mid-single digits because average prices climbed roughly 85%.
The year-over-year comparison is even more extreme. Micron's DRAM revenue increased 343%, while bit shipments grew only in the low-20% range. NAND revenue rose 361% with only low-double-digit bit growth. Pricing explains most of the gap.
But physical semiconductor demand is clearly rising as well. SEMI reported that worldwide silicon wafer shipments reached 3.573 billion square inches in the second quarter, up 7.4% year over year and 9.1% from the previous quarter. Silicon wafers sit near the beginning of semiconductor production, so this gives us a useful measure that is far less distorted by chip prices.
There are basically two booms happening at different speeds. More semiconductor material is moving through factories, while the value attached to some of those chips, especially memory, is rising much faster.
| Current measure | YoY growth |
|---|---|
| Global semiconductor revenue, H1 | +102% |
| Silicon wafer shipments, Q2 | +7.4% |
| Micron DRAM bit shipments | Low-20% range |
| Micron DRAM revenue | +343% |
If you want more recent data on this point, please see our latest semiconductor industry report.
As this chart shows, and as featured in our semiconductor industry deck, search interest in semiconductors has been rising steadily
Is AI responsible for most semiconductor growth today?
Yes. AI infrastructure is currently doing most of the heavy lifting behind the semiconductor boom.
Nvidia makes the scale easy to see. In its latest completed quarter, Nvidia generated $81.6 billion of revenue, 85% more than a year earlier. Data Center revenue alone reached $75.2 billion, up 92%, and represented roughly 92% of Nvidia's entire business. Within Data Center, networking revenue almost tripled.
AMD is moving in the same direction. Its data-center business has become one of its main growth engines as sales of AI accelerators and server CPUs rise. Broadcom is benefiting from another part of the same buildout through networking chips and custom AI accelerators designed for hyperscalers.
TSMC shows what this looks like inside the factories. High-performance computing represented 66% of its second-quarter revenue, compared with 22% from smartphones. Advanced nodes at 7 nanometers and below accounted for 77% of wafer revenue.
These companies sell different things, and that is precisely why the pattern is so convincing. GPUs are booming, custom accelerators are booming, networking silicon is booming, advanced foundry capacity is booming and high-bandwidth memory is booming at the same time. AI has become the semiconductor industry's dominant incremental source of growth.
Are PCs and smartphones growing with the semiconductor industry?
No. PC and smartphone shipments are currently shrinking even while semiconductor sales are surging.
IDC counted 68.2 million PCs shipped worldwide in the second quarter, down 4.9% from a year earlier. Global smartphone shipments fell even faster, dropping 6.7% to about 277.5 million units.
That divergence shows how much the semiconductor market has changed. Historically, hundreds of millions of PCs and more than a billion smartphones were obvious drivers of chip cycles. Today, a smaller number of extraordinarily expensive AI systems can generate enough semiconductor demand to overwhelm weakness in those huge consumer markets.
Memory prices are actually making the consumer side harder. IDC directly linked the recent PC and smartphone weakness partly to higher memory costs, which force manufacturers to raise prices, cut specifications or absorb lower margins.
So today's semiconductor boom does not need people to buy more laptops and phones. Data centers are spending enough on processors, networking and memory to more than compensate for weaker consumer-device volumes.
This chart, featured in our semiconductor industry deck, shows annual venture capital investment in semiconductor startups
Is semiconductor growth spreading beyond AI chips?
Yes. Semiconductor growth has lately spread into automotive, industrial and embedded chips, although those markets are growing nowhere near as fast as AI.
Texas Instruments is useful here because its business is dominated by analog and embedded semiconductors rather than AI GPUs. TI's latest quarterly revenue reached $5.46 billion, up 23% from a year earlier. Management said the improvement was broad, led by industrial, data-center and automotive customers.
NXP gives us a more detailed breakdown. Its quarterly revenue increased 19% to $3.50 billion, with every major end market growing. Automotive rose 12%, Industrial & IoT jumped 38%, and communications infrastructure increased 41%.
STMicroelectronics also reported a 26% year-over-year revenue increase in its latest quarter. These companies had struggled while automotive and industrial customers spent much of the previous downturn clearing excess inventories, so their return to growth is meaningful.
Calling the whole semiconductor expansion an “AI-chip bubble” therefore misses a fairly important part of the picture. AI explains the spectacular numbers, while more ordinary chip markets are recovering underneath it.
If you want more recent data on this point, please see our latest semiconductor industry report.
Are advanced semiconductor fabs still getting busier?
Yes. Advanced semiconductor manufacturing is still accelerating today, especially around AI processors and memory.
TSMC reported $40.2 billion of second-quarter revenue, then guided to roughly $44.6 billion to $45.8 billion for the following quarter. A company already operating at that scale is still expecting several billion dollars of sequential growth.
The technology mix is moving toward the most advanced processes at the same time. TSMC's 3-nanometer nodes produced 30% of wafer revenue, 5 nanometers produced 33%, 7 nanometers contributed another 11%, and the newly ramping 2-nanometer process had already reached about 3%.
The equipment suppliers are seeing the same pressure upstream. ASML sold 86 new lithography systems in its latest quarter, up from 67 in the previous one. Management said customers were accelerating capacity plans and announced plans to raise both EUV and immersion-lithography production capacity for 2027.
That is difficult to dismiss as inventory noise. Customers are ordering the machines needed to manufacture future generations of chips while current leading-edge production is still growing.
This chart, featured in our semiconductor industry deck, looks at TSMC’s strategy in semiconductors
Are mature-node semiconductors recovering too?
Yes. Mature-node semiconductor factories are busy again, particularly in China, even though long-term overcapacity remains a real risk.
SMIC's recent results showed revenue growth of roughly one-third year over year and fab utilization above 90%. Hua Hong, another major Chinese foundry focused heavily on mature processes, has also been running its fabs at very high utilization.
Those factories produce the less glamorous chips that still sit inside cars, appliances, industrial systems, power electronics, displays and consumer devices. High utilization gives us another indication that the recovery extends beyond Nvidia-style AI accelerators.
China is also adding a huge amount of semiconductor capacity. SEMI expects the country to remain the world's largest market for fab capacity and one of the largest spenders on semiconductor manufacturing equipment.
For now, existing mature-node capacity is being absorbed surprisingly well. The worry comes later: if Chinese fabs keep expanding faster than demand for commodity chips, prices could eventually come under much more pressure.
Is the memory-chip boom actually sustainable?
AI has made the memory market structurally bigger, but today's extraordinary memory revenue growth will not continue at anything close to the same rate.
WSTS says memory revenue increased about 305% year over year in the first half. SK hynix reported quarterly revenue of KRW 79.3 trillion, up 257%, while operating profit increased 557%. Micron's latest quarterly revenue reached $41.46 billion, more than four times the level from a year earlier.
High-bandwidth memory is a major reason this cycle looks different. AI accelerators need large quantities of extremely fast memory placed close to the processor, and each new accelerator generation keeps pushing memory capacity and bandwidth higher. SK hynix has begun mass shipments of HBM4, while Micron says HBM4 is already shipping in high volume for its lead customer's platform.
At the same time, the pricing numbers are impossible to ignore. Micron's 343% DRAM revenue increase came with bit shipments growing only in the low-20% range. Memory customers are currently paying far more per bit.
That makes the current memory boom unusually powerful but also vulnerable to normalization. AI should keep increasing the amount and value of memory used in servers. The 200%-plus price increases are a very different story and can reverse once supply catches up.
If you want more recent data on this point, please see our latest semiconductor industry report.
This chart, featured in our semiconductor industry deck, shows annual funding in semiconductor startups
Are semiconductor companies still spending like this boom will last?
Yes. Chipmakers are currently spending as though semiconductor demand will remain strong for years, not quarters.
SEMI now expects semiconductor manufacturing-equipment sales to reach a record $165.9 billion in 2026, up 23.2%. Wafer-fab equipment should reach about $143.9 billion, while DRAM equipment spending is forecast to rise 39% and NAND equipment spending about 31%.
The revision at ASML is even more telling. Early in the year, ASML expected 2026 revenue of €34 billion to €39 billion. It first raised that range to €36 billion to €40 billion, then lifted it again to €43 billion to €45 billion after its second quarter.
ASML is also preparing roughly 30% more low-NA EUV capacity for 2027 and about 30% more immersion DUV capacity. Those machines can cost well into nine figures and require long production lead times. Customers do not reserve them casually.
SEMI separately expects 300mm fab-equipment spending to rise 18% this year to $133 billion and another 14% to $151 billion in 2027. Investment is therefore continuing after today's demand boom rather than peaking alongside it.
| Semiconductor manufacturing measure | Current outlook | Growth |
|---|---|---|
| Total semiconductor equipment | $165.9B | +23.2% |
| 300mm fab equipment | $133B | +18% |
| DRAM equipment | $38.8B | +39% |
| 300mm fab equipment in 2027 | $151B | +14% |
Are cloud companies actually buying enough AI chips to support this semiconductor boom?
Yes. The companies paying for AI infrastructure are still increasing cloud revenue and spending enormous amounts on computing capacity.
Microsoft's latest results are probably the cleanest example. Azure revenue grew 43%, while Microsoft's total commercial remaining performance obligation reached $678 billion, up 84%. The company has said it expects roughly $190 billion of capital expenditure during calendar 2026 and still expects computing capacity to remain constrained through the year.
Microsoft is spending far more on infrastructure, and demand has continued rising faster than the capacity coming online.
The same pattern is visible across the large cloud platforms. Amazon keeps expanding AWS infrastructure, while Alphabet has sharply increased capital spending as Google Cloud and AI workloads grow. Meta is spending tens of billions of dollars on computing infrastructure despite operating a very different business model from the cloud vendors.
We still do not know whether every dollar of this AI spending will earn a good return. Semiconductor suppliers do, however, have something much stronger than speculative orders from small startups: their biggest customers include some of the most profitable companies in the world, and those customers are still fighting for more compute.
This chart, featured in our semiconductor industry deck, compares the main business model options for fabless semiconductor companies
Is semiconductor supply still tight today?
Yes. Semiconductor supply is still tight where the industry is growing fastest, especially in advanced memory, leading-edge manufacturing and AI infrastructure.
Memory pricing gives us the clearest evidence. DRAM and NAND prices have risen dramatically because suppliers cannot increase bits fast enough to match current demand. High-bandwidth memory adds another bottleneck because it requires advanced DRAM processes, sophisticated packaging and tight integration with accelerators.
Leading-edge manufacturing is under similar pressure. ASML has repeatedly raised its sales outlook while its customers accelerate future lithography orders. TSMC is also expanding advanced capacity while its high-performance-computing business keeps growing.
The shortage is increasingly visible outside the data center too. IDC says higher memory prices are now hurting PC and smartphone demand. AI infrastructure is consuming enough semiconductor resources to affect what manufacturers can economically put into consumer devices.
The industry no longer has the broad “everything is unavailable” shortage seen after the pandemic. Today's bottlenecks are much more concentrated, but they sit exactly where semiconductor value is rising fastest.
If you want more recent data on this point, please see our latest semiconductor industry report.
Could today's semiconductor boom create another chip glut?
Yes. The semiconductor industry's biggest medium-term risk is now overbuilding.
Semiconductor cycles have repeated the same basic mechanism for decades. High demand pushes utilization and prices up, manufacturers expand aggressively, the new fabs arrive with a delay, and supply eventually catches demand.
The scale of today's investment makes that risk impossible to ignore. SEMI expects global installed semiconductor capacity to increase roughly 5% in both 2026 and 2027. Equipment suppliers are enjoying record demand. Memory manufacturers are adding capacity, TSMC is expanding advanced nodes, and China continues building mature-node fabs.
Memory looks especially exposed because current prices are so unusual. A market where DRAM revenue can rise more than 300% while bit shipments rise only around 20% leaves plenty of room for revenue to fall once pricing normalizes, even if physical AI demand continues increasing.
AI could absorb far more capacity than previous computing cycles did. Current hyperscaler spending supports that case. But every extra year of extraordinary demand also encourages another year of extraordinary supply investment.
For the semiconductor industry, today's shortage and tomorrow's oversupply are perfectly compatible outcomes.
This chart, featured in our semiconductor industry deck, shows the revenue mix across customer segments in the semiconductor industry
Will the semiconductor industry keep growing into 2027?
Probably. Semiconductor growth currently has enough momentum to continue into 2027, although another year like this one would be extremely difficult to repeat.
WSTS's latest exercise using actual second-quarter results points to a semiconductor market of roughly $1.66 trillion for 2026 and about $2.1 trillion for 2027. WSTS stresses that these figures are mechanical updates to its spring assumptions rather than a completely new forecast, so the exact numbers deserve some caution.
The direction has stronger support than the precise market size. Semiconductor factories are adding capacity, equipment spending is forecast to rise again, cloud companies still say compute is constrained, silicon wafer shipments are increasing, and industrial and automotive chips have finally returned to growth.
The weak point is pricing. Semiconductor revenue cannot keep benefiting indefinitely from memory ASP increases measured in the hundreds of percent. If memory prices flatten or fall, headline market growth could slow dramatically while the underlying amount of computing installed keeps rising.
Another positive year therefore looks more likely than an abrupt industry contraction. What should fade is the current triple-digit growth rate.
So, is the semiconductor industry growing now?
Yes. The semiconductor industry is clearly growing now, and the evidence is much stronger than a few blockbuster Nvidia quarters.
Global chip sales have more than doubled year over year in the latest half-year data. Physical silicon wafer shipments are also increasing. AI processors, networking chips, advanced foundry capacity and memory are booming, while automotive, industrial and embedded semiconductors have moved back into growth.
The qualification is important because the headline numbers exaggerate how fast the underlying market is expanding. Memory prices are doing an enormous amount of work, and PC and smartphone shipments are currently falling. AI infrastructure has become large enough to push the whole industry's revenue sharply higher despite that weakness.
Still, we would need to ignore too much evidence to call this a narrow or artificial recovery. Advanced fabs are expanding, mature fabs are highly utilized, silicon shipments are up, analog chipmakers are recovering and semiconductor-equipment demand has reached record levels.
Our judgment is straightforward: the semiconductor industry is in a genuine growth cycle today, with AI acting as the accelerator rather than the only thing keeping the market alive.
The main question has already moved on. What we need to watch now is whether AI demand can absorb the huge amount of semiconductor capacity being built before pricing normalizes and the next supply cycle catches up.
If you want more recent data on this point, please see our latest semiconductor industry report.
This chart, featured in our semiconductor industry deck, shows how advanced foundry node manufacturing technology has evolved over time
OUR METHODOLOGY
We approached the question “Is the semiconductor industry growing now?” as a multi-dimensional one. Semiconductor growth can look very different depending on whether we measure industry revenue, physical output, pricing, end-market demand, manufacturing activity or investment, so we assessed those dimensions separately rather than letting one headline number determine the answer.
We prioritized recent operating evidence: monthly and quarterly industry data, company results, shipment volumes, utilization, capacity plans and capital spending. Company figures were used to examine specific parts of the market, while industry-wide datasets helped establish whether those patterns were visible beyond individual companies.
One important distinction is between revenue growth and physical semiconductor growth. Memory pricing is unusually strong, so we compared revenue figures with bit shipments and silicon-wafer shipments to avoid treating every dollar of additional semiconductor sales as an equivalent increase in chip volumes.
We also separated AI-linked demand from traditional semiconductor markets. Nvidia, AMD, Broadcom, TSMC and memory producers help show the scale of the AI infrastructure cycle, while Texas Instruments, NXP, STMicroelectronics and mature-node foundries provide a useful check on whether growth is spreading into automotive, industrial, analog and embedded chips.
Manufacturing activity and investment were treated as separate evidence. Current utilization, wafer shipments and advanced-node revenue tell us what is happening in factories today; equipment orders, fab spending and capacity plans help show how strongly semiconductor companies expect that demand to persist.
For forward-looking conclusions, we gave more weight to observable commitments such as capacity additions, equipment orders, hyperscaler infrastructure spending and company guidance than to standalone long-range market forecasts. Forecasts were used mainly to test whether those operating signals pointed in the same general direction.
Key sources used for the industry-level analysis include World Semiconductor Trade Statistics on the global semiconductor market in H1 and Q2 2026, WSTS on finalized 2025 semiconductor sales, the Semiconductor Industry Association on Q2 and June 2026 global sales, and SEMI on worldwide silicon-wafer shipments.
For pricing, AI demand and manufacturing, we relied particularly on Micron's latest results, Micron's 10-Q for DRAM and NAND shipment and pricing data, Nvidia's latest quarterly results, AMD's Q2 2026 results, Broadcom's Q2 2026 results, and TSMC's Q2 2026 results and manufacturing mix.
To test the breadth and durability of the cycle, we also used IDC's Q2 2026 PC shipment data, IDC's Q2 2026 smartphone data, Texas Instruments' Q2 2026 results, NXP's Q2 2026 results, SK hynix's Q2 2026 memory results, and ASML's Q2 2026 results and capacity plans.
Finally, the investment and end-customer sections draw on SEMI's semiconductor-equipment forecast, SEMI's 300mm fab-equipment outlook, Microsoft's FY2026 Q4 results, Microsoft's FY2026 Q3 earnings commentary, and Meta's Q2 2026 results.
In our semiconductor industry deck, we identify pain points entrepreneurs should prioritize
Related blog posts
- How strong is fundraising in the semiconductor industry right now?
- How large is the semiconductor industry really?
- How is the funding in the semiconductor industry now?
- The evolution of funding activity in the semiconductor industry
