ASML, the Dutch company that builds nearly every machine capable of printing the world’s most advanced chips, says it cannot make enough of them. On October 14, 2026, ASML Chief Executive Officer Christophe Fouquet told investors that planned supply for 2027 and 2028 will not fully meet customer demand for chipmaking tools, a gap driven almost entirely by the artificial intelligence buildout now consuming the semiconductor industry’s attention and capital. The disclosure, confirmed by Reuters and CNBC following ASML’s third-quarter earnings report, marks one of the clearest signals yet that the AI infrastructure race has outrun even the equipment suppliers who are supposed to enable it.
ASML has long operated as a chokepoint in global chip production. Its extreme ultraviolet (EUV) lithography systems are the only machines on Earth capable of etching the nanometer-scale patterns needed for cutting-edge logic and memory chips, the kind that power Nvidia GPUs, custom AI accelerators, and the high-bandwidth memory stacks that feed them. When ASML says demand is outstripping its own expansion plans, every company racing to build AI data centers, from Microsoft, Meta, and Google to Nvidia’s foundry partners, is effectively being told to expect a longer wait.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What ASML Actually Said on October 14, 2026
According to reporting from CNBC and Bloomberg, ASML said its planned supply increases for 2027 and 2028 will not be enough to satisfy customer orders for chipmaking tools. The company had previously told investors it intended to raise capacity by 30% in 2027 and by another 30% in 2028, a two-year expansion plan that was supposed to give customers breathing room after years of tight supply. Instead, Fouquet said ASML is now investigating ways to push 2028 capacity beyond that already-aggressive 30% target, according to Bloomberg and Business Standard.
Fouquet described the order book in blunt terms. “A lot of order intake for 2028. But also some strong interest to already place orders for 2029 and sometimes even 2030,” he said, according to Reuters’ account of the earnings call. That timeline is striking: customers are not just asking ASML to hit its near-term targets, they are trying to reserve machines that will not ship for three or four years. In a separate comment carried by Business Standard, Fouquet put it even more plainly: “Customers are asking us every single day to get more tools.”
ASML’s own capital-markets guidance, repeated by Fouquet and cited by Reuters, framed the shift as structural rather than seasonal. “Our customers continue to increase their short- and medium-term equipment demand forecasts with us, driven primarily by their views on the size and duration of AI-related investments,” Fouquet said, according to CNBC’s report on the earnings call. Separately, he told the same call, “A lot of our discussions with our customers now are shifting in fact beyond 2027,” a detail reported by KSL in its coverage of ASML’s fourth-quarter sales forecast.
The company’s reporting also referenced third-party analysis. JPMorgan analysts flagged that ASML could be examining production of more than 110 EUV systems in 2028, a figure that has circulated in market commentary but which ASML itself has not confirmed as a formal target. That distinction matters: ASML has confirmed the demand gap and the 30%-plus-30% capacity framework, but the specific unit counts beyond that remain analyst estimates rather than company commitments.
Why This Is an AI Story, Not Just a Chip Story
It would be easy to file this under semiconductor-industry plumbing and move on. That would miss the point. ASML’s lithography tools sit at the top of a supply chain that ends, several steps later, in the GPUs and accelerators training and running the large language models reshaping software, search, and enterprise IT. When ASML cannot build machines fast enough, the bottleneck ripples forward: foundries like TSMC and Samsung cannot expand wafer capacity on schedule, chip designers cannot get allocation for new accelerator designs, and cloud providers cannot stand up the data centers they have already budgeted for.
This is the same dynamic that has shown up elsewhere in the hardware stack this year. Memory makers have been squeezed so hard by AI-driven demand for high-bandwidth memory that smartphone shipments fell to a 14-year low partly because DRAM and NAND capacity got redirected toward data-center customers willing to pay more. The same memory crunch helped drag PC shipments down sharply in the third quarter, as chronicled in reporting on how DRAM price surges hit the broader electronics market. ASML’s warning extends that pattern one layer further upstream: the machines needed to make the chips are now as scarce as the chips themselves.
Reports citing Samsung and TSMC timelines said the two foundries plan to begin mass-producing chips with high-NA EUV tools, ASML’s newest and most advanced lithography class, in 2028 and 2030 respectively. Those dates sit squarely inside the window ASML now says it cannot fully supply. If Samsung and TSMC are depending on high-NA EUV machines to hit those production milestones, any slippage in ASML’s rollout could cascade into delayed node transitions at exactly the moment AI chip designers are counting on smaller, denser transistors to keep performance scaling.
The Capacity Math: 30% Plus 30% Still Isn’t Enough
ASML’s original plan was already ambitious. A 30% capacity increase in a single year is a major industrial undertaking for a company that builds machines with hundreds of thousands of individually precise components, sourced from a supplier base spread across Germany, the United States, and Japan. Committing to two consecutive 30% jumps, in 2027 and then 2028, implied ASML believed it was getting ahead of demand. Fouquet’s comments on October 14 suggest the opposite: demand grew faster than the plan, and ASML is now examining how to exceed its own expansion target for 2028, per Bloomberg’s reporting.
| Metric | Previously Communicated Plan | October 2026 Update |
|---|---|---|
| 2027 capacity increase | 30% above prior baseline | Unchanged target, demand already outpacing it |
| 2028 capacity increase | 30% above 2027 level | ASML examining ways to exceed the 30% target |
| Order horizon | Primarily 2026-2027 | Strong order intake for 2028; interest for 2029 and 2030 |
| Stated demand driver | General logic and memory demand | AI-related investment size and duration, per Fouquet |
| Unconfirmed analyst estimate | Not previously public | JPMorgan flags possible 110+ EUV systems in 2028 (not an ASML commitment) |
The practical effect is that ASML’s backlog, already one of the longest lead times in the technology industry, is extending further into the future with each earnings cycle. Customers placing orders today for 2028 delivery are, in effect, making four-year bets on how large the AI buildout will still be when the machines finally arrive. That some customers are now asking about 2029 and 2030 slots suggests at least part of the industry believes the AI capacity crunch is not a temporary spike but a multi-year structural shift, a view that aligns with how aggressively hyperscalers have been raising their own capital spending.
Historical Context: From Chip Shortage to Tool Shortage
The 2020-2022 chip shortage was mostly a story about finished semiconductors: cars sitting unfinished on lots because they lacked a $2 microcontroller, graphics cards selling for triple MSRP because fabs could not keep up with demand. That shortage eventually eased as foundries added capacity and demand normalized. What ASML is now describing is a different and arguably more durable constraint, because it sits one step further back in the supply chain. You cannot build a new fab, retool an existing one for a smaller process node, or expand EUV-based production capacity without ASML’s machines, and ASML cannot build those machines any faster than its own supplier network allows.
This is also a reversal from the mood around ASML just two years ago, when concerns centered on Chinese export restrictions limiting ASML’s addressable market and on fears that AI capital spending might cool off. Instead, the opposite has happened: AI capital spending accelerated, and ASML’s problem shifted from finding enough customers to finding enough manufacturing capacity to serve the ones it already has. The tightening of export controls that cut into Nvidia’s China market share has not slowed the broader AI equipment race outside China. If anything, it has pushed more of the world’s AI chip demand toward the US, European, and allied-Asian supply chains that rely most directly on ASML’s tools.
Market Reaction and ASML’s Raised Outlook
ASML paired the demand warning with a stronger near-term outlook. Investing.com reported that ASML shares climbed after the company raised its 2026 outlook alongside its capacity expansion plans, and Bloomberg reported that ASML raised its full-year sales outlook as the AI boom drives demand. The combination, a near-term beat alongside a warning about multi-year supply constraints, is an unusual signal for a company this deep in the industrial supply chain. It tells investors that the current quarter looks strong, but it also tells chipmakers and cloud operators not to expect relief from tool scarcity anytime soon.
That dynamic mirrors what has played out across the AI hardware stack throughout 2026. Capital spending commitments from the largest cloud platforms have kept climbing even as supply constraints bite, a pattern visible in how Microsoft, Meta, and Google pushed combined AI capital expenditure past $725 billion this year. ASML’s order book for 2028 through 2030 suggests those hyperscalers, and the chipmakers that supply them, are not treating current capex levels as a peak but as a floor.
Competitive Landscape: Who Depends on ASML and How Much
ASML’s monopoly-like position in EUV lithography means essentially every advanced chipmaker is a customer, which makes the demand imbalance a shared problem rather than a competitive wedge for any single company. TSMC, Samsung, and Intel all depend on ASML for the tools needed to produce chips at the process nodes AI accelerators require. Reports citing Samsung and TSMC’s roadmaps said both companies plan to begin mass production using high-NA EUV tools, Samsung in 2028 and TSMC in 2030, placing both foundries directly inside the window ASML says it may not be able to fully supply.
| Company | Role in the AI Chip Supply Chain | Exposure to ASML’s Capacity Gap |
|---|---|---|
| ASML | Sole supplier of EUV lithography systems | Direct: cannot expand output fast enough for its own order book |
| TSMC | Leading foundry for AI accelerators (Nvidia, AMD, Apple designs) | High: plans high-NA EUV mass production from 2030, per reports |
| Samsung | Foundry and memory producer | High: plans high-NA EUV mass production from 2028, per reports |
| Nvidia | Designs AI GPUs manufactured on ASML-dependent nodes | Indirect: wafer allocation tied to foundry capacity growth |
| Hyperscalers (Microsoft, Meta, Google, Amazon) | Buyers of finished AI accelerators and custom silicon | Indirect: chip availability shapes data center build-out pace |
What makes this moment different from prior supply cycles is that the demand signal is coming directly from the equipment layer rather than from chipmakers themselves. In past shortages, foundries like TSMC would announce capacity expansions and ASML would respond to incoming orders. Now ASML is the one publicly describing a demand book that extends years past its planned output, which suggests the constraint is less about any single foundry’s ambition and more about an industry-wide scramble for every unit of lithography capacity that exists or can be built.
What This Means for AI Model Development Timelines
For the AI labs and cloud platforms racing to train and serve ever-larger models, ASML’s warning is a reminder that compute availability is not just a function of GPU design or data center construction, it is bounded by how many advanced chips can physically be manufactured each year. If foundries cannot expand wafer output on the schedule they planned because lithography tools are delayed, the downstream effect is fewer GPUs, fewer accelerators, and tighter allocation for the companies trying to scale AI infrastructure. That dynamic has already shaped competitive positioning in 2026, as seen in how tightly Nvidia, AMD, Google, and Meta have had to coordinate on data center hardware specifications to make the most of available silicon.
It also raises the stakes for any chipmaker trying to break into the AI accelerator market without an established position in the ASML order queue. A company trying to scale a new AI chip design now effectively needs not just design wins and software support, but a reserved slot in a manufacturing pipeline that is already booked years out. That is one reason the AI hardware conversation in 2026 has shifted so heavily toward long-term supply agreements and equity stakes rather than simple purchase orders.
Industry Data on Developer and Enterprise AI Adoption
The demand ASML describes does not exist in a vacuum, it is being pulled by real adoption of AI tools across the software industry. According to Stack Overflow’s 2025 Developer Survey, 84% of developers reported that they are using or planning to use AI tools in their development process, and 51% of professional developers reported using AI tools daily. The same survey found that 31% of respondents reported currently using AI agents, with another 17% planning to adopt them, and 69% of developers who had used AI agents at work agreed that the tools increased their productivity. Each of those agent workloads ultimately runs on inference infrastructure built from the same chips ASML’s tools help manufacture, which is part of why equipment demand has kept climbing even as individual AI products come and go.
Predictions: Where This Goes Next
- ASML will likely announce a formal upward revision to its 2028 capacity target within the next two to three earnings cycles, given Fouquet’s comments about actively investigating ways to exceed the existing 30% plan.
- Expect foundries including TSMC and Samsung to publicly discuss contingency plans for high-NA EUV rollout timing, since both are depending on ASML’s newest tool generation for mass production starting in 2028 and 2030.
- Order deposits and long-term supply agreements for 2029 and 2030 delivery slots will become more common and more publicly disclosed, as chipmakers try to lock in allocation years ahead of need.
- Pressure will grow on ASML’s component suppliers in Germany, the US, and Japan, since the company’s ability to exceed its 30% capacity targets depends on expanding its own supplier base, not just its own factory floor.
- Expect continued volatility in AI hardware stocks tied to perceived lithography bottlenecks, with any ASML guidance update capable of moving sentiment across chip designers, foundries, and hyperscaler suppliers simultaneously.
Why ASML’s Statement Carries So Much Weight
Few companies in the technology industry have a vantage point as singular as ASML’s. Because its EUV machines are not substitutable, every advanced chipmaker’s expansion plan is, in effect, routed through ASML’s production schedule. When Fouquet says customers are asking every single day for more tools, he is describing demand from the same companies that supply chips to Nvidia, AMD, Apple, and every major cloud provider. That makes ASML’s commentary one of the most reliable early indicators of where the AI hardware buildout is actually headed, ahead of the announcements from chip designers and cloud operators who ultimately depend on the equipment ASML ships.
It is also a reminder that the AI boom’s physical constraints are becoming more visible than its software ones. Training a bigger model is, in many respects, a solved engineering problem. Building the chips to run it at scale, on a predictable schedule, is not. ASML’s acknowledgment that its own aggressive capacity plan will not be enough through 2028 suggests that constraint is not easing, and that the companies best positioned through the rest of this decade may be the ones that secured their place in ASML’s order book earliest.
Frequently Asked Questions
What exactly did ASML announce on October 14, 2026?
ASML said its planned supply increases for 2027 and 2028, a 30% capacity increase in each year, will not fully meet customer demand for chipmaking tools, driven primarily by AI-related investment, according to Reuters and CNBC.
Who is Christophe Fouquet?
Christophe Fouquet is the Chief Executive Officer of ASML. His comments on the company’s third-quarter earnings call form the basis of this report.
Is ASML increasing its capacity targets?
ASML said it is investigating ways to increase 2028 capacity beyond the previously communicated 30% increase, according to Bloomberg and Business Standard. A formal revised target has not yet been confirmed.
Are customers really ordering tools for 2029 and 2030?
Fouquet said some customers have expressed interest in placing orders for 2029 and sometimes 2030, according to Reuters’ coverage of the earnings call, though strong confirmed order intake was described specifically for 2028.
Is the 110 EUV systems figure for 2028 confirmed by ASML?
No. That figure was attributed to JPMorgan analysts, not confirmed directly by ASML as a company target, so it should be treated as an external estimate rather than an official commitment.
How does this affect Nvidia and other AI chip designers?
Nvidia and other designers do not buy ASML tools directly, but they depend on foundries like TSMC and Samsung that do. A lithography equipment shortfall at the foundry level can limit how many advanced chips those foundries can produce, indirectly constraining AI accelerator supply.
Did ASML’s stock react to this news?
Yes. Investing.com reported that ASML shares climbed after the company raised its 2026 outlook alongside its capacity expansion plans, even as it flagged the multi-year demand gap.
What is high-NA EUV, and why does it matter here?
High-NA EUV is ASML’s newest, most advanced lithography tool generation. Reports citing Samsung and TSMC said the two foundries plan to begin mass-producing chips with high-NA EUV tools in 2028 and 2030 respectively, timelines that sit inside the capacity gap ASML described.
Related Coverage
- Nadella's 6 AI Safeguards Demand Emergency Brake [2026]
- Anthropic's 15-Box AI Scare Faces Skeptic Backlash [2026]
- Nvidia Backs Doctolib, Genesis AI in 4-Startup Bet [2026]
- KT Debuts AI Network at 800Gbps, 8x Past Max [2026]
- FBI Names OpenAI Suspect Alfozan, Hired July 2025 [2026]
