ASML Raises Full-Year Sales Outlook as AI Demand Collides With Supply

ASML Raises Full-Year Sales Outlook after reporting €11.7 billion in third-quarter sales, but its bigger signal concerns supply. The Dutch chip-equipment company says planned production for 2027 and 2028 will not fully satisfy customer demand.

The new forecast puts 2026 net sales between €44 billion and €45 billion. That raises the bottom of ASML’s previous €43 billion to €45 billion range while leaving the ceiling unchanged.

This is more than another quarterly guidance increase. ASML’s customers include TSMC, Samsung, Intel, SK Hynix, and Micron, which sit near the foundation of the AI computing market. Their appetite for ASML equipment reveals where chipmakers expect capacity shortages to persist.

The central tension has therefore moved beyond whether AI investment will continue. The harder question is whether ASML and its suppliers can add manufacturing capacity without creating a costly excess when the current expansion cycle slows.

ASML Raises Full-Year Sales Outlook After Record Quarterly Sales

The guidance increase is modest, but the operating figures behind it show a much larger shift in production and demand.

ASML announced its third-quarter results on October 14, 2026. The company reported €11.66 billion in net sales, up from €9.33 billion during the second quarter.

Net income reached €4.96 billion, compared with €2.92 billion in the prior quarter. Gross margin rose from 54% to 57.5%.

The company sold 113 new lithography systems and eight used systems during the quarter. It had sold 86 new systems and five used systems during the preceding quarter.

Lithography is the process that projects extremely small circuit patterns onto silicon wafers. These repeated patterns eventually become the transistors and connections inside a processor.

The most advanced version is extreme ultraviolet lithography, commonly called EUV. It uses very short-wavelength light to print features required for leading-edge chips.

ASML is the only commercial supplier of EUV lithography systems. That position makes its production schedule a critical constraint for advanced chip manufacturing.

The company’s quarterly results also showed €3.29 billion in installed-base management sales. That category covers services and upgrades for machines already operating in customer factories.

Those upgrade sales exceeded ASML’s expectations. They matter because customers can increase output without waiting for an entirely new lithography system or production line.

ASML now expects fourth-quarter sales between €14.2 billion and €15.2 billion. It forecasts a gross margin near 57% for that period and approximately 56% for the full year.

The revised ASML sales outlook follows two earlier increases during 2026. At the start of the year, the company expected annual revenue between €34 billion and €39 billion.

After the first quarter, it raised that range to roughly €36 billion through €40 billion. In July, the company increased its forecast again, this time to €43 billion through €45 billion.

The latest revision narrows that range to €44 billion through €45 billion. ASML has therefore raised the lower bound by €10 billion since its original annual forecast.

However, this latest change does not raise the maximum. Investors should read it as greater confidence in near-term execution, not an unlimited expansion of the current cycle.

ASML AI demand has already produced record quarterly sales and stronger margins. The new constraint is how quickly equipment can move through ASML’s specialized production network.

That distinction creates the article’s main conflict. The company is not struggling to find buyers, but it must decide how aggressively to build for demand several years away.

AI Chip Spending Is Extending ASML’s Order Horizon

Customer interest stretching into 2029 and 2030 suggests that leading chipmakers see AI infrastructure as a multiyear capacity problem.

Chief Executive Christophe Fouquet said customers continue to increase their short-term and medium-term equipment forecasts. He tied those revisions mainly to expectations about the scale and duration of AI investment.

Some customers are already considering orders for 2029 and 2030, according to Fouquet. That is an unusually distant horizon for an industry known for sharp cycles.

The important mechanism begins with spending on AI models and services. Those workloads require accelerators, advanced memory, networking chips, and substantial data-center infrastructure.

Chip designers such as Nvidia depend on manufacturers such as TSMC to produce their leading processors. Those manufacturers need lithography equipment before they can add advanced wafer capacity.

Memory suppliers also require additional equipment as high-bandwidth memory becomes more important to AI systems. That places Samsung, SK Hynix, and Micron in the same capital-spending chain.

ASML AI demand is therefore indirect but highly concentrated. The company does not sell AI software or data-center services, yet it supplies equipment needed far upstream.

A hypothetical AI assistant provides a useful example. More users create demand for additional inference capacity, which is the computing required to run a trained model.

Cloud providers respond by ordering more accelerators. Chip designers place larger foundry orders, while foundries invest in fabrication capacity and advanced packaging.

ASML receives demand signals near the beginning of that physical expansion. Its order conversations can reveal confidence before newly constructed capacity produces a single commercial chip.

This does not mean every equipment order maps directly to one AI deployment. Lithography systems serve smartphones, automotive electronics, memory, and other markets alongside AI processors.

However, ASML says AI expectations are now the primary driver behind customers’ higher forecasts. That claim is consistent with the long production plans described by major foundries and memory manufacturers.

The company’s July sales forecast already raised expectations to €43 billion through €45 billion. At that time, it outlined plans to expand selected equipment capacity.

ASML planned to increase output capacity for low numerical aperture EUV and immersion deep ultraviolet systems by 30% during 2027. It was also investigating another 30% increase for 2028.

Numerical aperture describes how effectively an optical system gathers light and resolves small features. A higher value allows lithography equipment to print finer patterns.

Low-NA EUV refers to the established generation of EUV machines used in current advanced manufacturing. High-NA EUV is a newer platform designed for smaller features and future process technologies.

Deep ultraviolet lithography, or DUV, uses longer-wavelength light than EUV. Chipmakers still depend on it for many layers, including less demanding steps in advanced production.

ASML now says those previously disclosed capacity plans will not meet expected demand in 2027 or 2028. The company is studying additional expansion beyond its earlier target.

The signal reaches far beyond ASML. TSMC, Samsung, and Intel must secure tools years before planned factories reach full production.

A delayed system can affect when a process node enters volume manufacturing. It can also limit how quickly a successful chip moves from initial production into broad availability.

For cloud companies and enterprise buyers, these constraints appear later as tight accelerator supply, longer deployment schedules, or restricted access to newer hardware.

The revised outlook therefore does not simply describe ASML’s financial year. It indicates that the physical buildout behind AI services remains active across several layers of the supply chain.

TSMC, Samsung, and Intel Face a Capacity Allocation Contest

ASML’s production limits turn equipment procurement into a strategic contest among chipmakers with different technologies and commercial pressures.

TSMC holds a central position because it manufactures advanced chips for customers across the technology industry. Demand from those customers determines how aggressively TSMC adds leading-edge capacity.

Samsung competes in foundry manufacturing while also producing memory. It must divide capital and management attention across several markets experiencing AI-related demand.

Intel is expanding its manufacturing capabilities while pursuing outside foundry customers. Its plans depend on timely access to advanced equipment as it works to improve process execution.

These companies are not competing to buy an interchangeable machine from several vendors. EUV availability ultimately depends on ASML’s own assembly capacity and a specialized supplier network.

An EUV system combines high-precision optics, light sources, vacuum equipment, software, and mechanical components. Increasing production requires coordinated investment across that network.

ASML can improve supply through several routes. It can assemble more systems, help suppliers expand, raise machine throughput, and upgrade equipment already installed inside factories.

Chief Financial Officer Roger Dassen said upgrades and higher-throughput machines would help address shortages. Those measures can expand wafer capacity before entirely new production facilities become available.

ASML also broke ground on a major Eindhoven expansion in September 2026, according to AI spending coverage. However, the project’s first phase is not expected to finish until 2029.

That timing explains why the current challenge cannot be solved with construction alone. Demand is already pressing against production plans for 2027 and 2028.

ASML must bridge that period using existing sites, supplier expansions, operational improvements, and upgrades. Each option carries execution risks and practical limits.

For customers, the contest is partly about allocation and partly about planning confidence. A foundry needs tools to arrive in the sequence required by its process roadmap.

Receiving one machine does not create a functioning production line. A factory also needs deposition, etching, inspection, metrology, cleaning, and packaging equipment.

This broader requirement benefits other equipment suppliers, including Applied Materials, Lam Research, KLA, and Tokyo Electron. Yet none can substitute for ASML’s EUV systems.

The ASML sales outlook consequently acts as an industry barometer. Rising sales imply broader investment, but supply limits can still determine which manufacturers scale first.

TSMC enters this contest with strong demand from leading chip designers. Samsung combines foundry ambitions with large memory commitments, while Intel is financing a manufacturing turnaround.

Each company also faces different customer expectations. TSMC must preserve execution across many external clients, while Samsung and Intel must balance internal products with foundry commitments.

The pressure extends to smaller manufacturers and new projects. Established customers with validated roadmaps, trained teams, and existing infrastructure can present a lower execution risk.

New entrants may have ambitious plans but less evidence that equipment will translate into stable volume production. ASML must assess those opportunities while serving long-standing customers.

Dassen said the company was still evaluating the scale of the opportunity associated with Elon Musk’s proposed TeraFab project. That caution illustrates the difference between stated interest and firm demand.

The primary contest is not ASML against another lithography supplier. It is customer demand against ASML’s ability to expand a uniquely complex production system.

That framing also explains why increasing capacity is not an easy victory. Moving too slowly risks delaying customer fabs, but moving too quickly raises exposure to a future correction.

What the Higher ASML Sales Outlook Does Not Prove

Strong orders validate present investment plans, but they do not settle whether AI infrastructure spending will produce durable economic returns.

ASML Raises Full-Year Sales Outlook at a moment when investors are debating the sustainability of AI capital spending. The results answer part of that debate, but not all of it.

They show that chipmakers are planning additional capacity and discussing commitments several years ahead. They do not prove that every planned data center will achieve its expected utilization or revenue.

Semiconductor manufacturing remains cyclical. Customers can adjust delivery schedules when inventories rise, demand weakens, or financing conditions change.

ASML itself lists order cancellations, shipment delays, customer demand, supply-chain capacity, and export controls among its business risks. These are not theoretical qualifications.

The current expansion also compresses several forecasts into one decision. ASML and its suppliers must invest before the final demand for 2029 equipment becomes fully visible.

If AI services continue scaling, insufficient equipment capacity can constrain chip output. Customers may then accept longer waits or prioritize their most profitable products.

If spending slows, recently expanded capacity could become underused. That would weaken orders across the semiconductor-equipment market even if AI adoption continues.

This difference matters because AI adoption and infrastructure spending do not move in perfect alignment. Software usage can rise while improvements in chip efficiency reduce required hardware growth.

Companies can also improve model efficiency, extend the useful life of accelerators, or shift workloads to specialized chips. Each development changes the relationship between AI usage and equipment demand.

Market expectations create another risk. ASML shares had risen substantially during 2026 before the report, according to Reuters-linked market coverage.

A stronger business outlook does not guarantee a proportionate stock-market gain. Investors can recognize strong demand while questioning whether that strength is already reflected in expectations.

Jefferies analysts described the increased 2027 and 2028 demand outlook as broadly expected. They also warned that valuation multiples across the sector could contract.

That view does not dispute ASML’s operational strength. It argues that a company can deliver good results while its shares face limited upside from an already optimistic starting point.

Export policy adds another uncertainty. ASML cannot ship its most advanced EUV systems to China, and additional restrictions could affect parts of its DUV business.

The company opposes proposed U.S. measures that would further restrict DUV sales and servicing in China. Dassen argued that broader controls could accelerate Chinese efforts to develop alternatives.

Policy decisions can affect ASML’s geographic sales mix, service revenue, and production allocation. They can also reshape competition among semiconductor manufacturers.

Meanwhile, High-NA EUV adoption remains an important technical variable. The systems promise finer patterning, but customers must justify their higher complexity through better manufacturing economics.

A customer’s interest in a future machine does not guarantee rapid deployment. Fabs require process development, supporting equipment, trained staff, and stable yields.

Yield is the share of manufactured chips that meet quality requirements. Poor yield can make an advanced process uneconomic even when its lithography equipment works as intended.

ASML’s strong installed-base revenue offers one counterweight. Existing factories can buy productivity upgrades, which may expand output with less risk than entirely new production lines.

Still, upgrades cannot replace every new system. Demand for advanced logic and memory ultimately requires additional physical capacity when existing factories reach their practical limits.

The skeptical reading should therefore remain precise. ASML’s report supports the claim that chipmakers expect sustained demand, but it cannot verify their end customers’ future returns.

The company also has not issued a complete 2027 sales forecast with this report. Investors know current supply plans are insufficient, yet they lack a finalized revenue range for next year.

That missing forecast leaves room for timing uncertainty. Demand can be strong while quarterly revenue shifts because of shipment schedules, installation milestones, or customer readiness.

ASML AI demand is real in the company’s current order conversations and financial performance. Its durability still depends on decisions made across the entire computing market.

Three Signals Will Test the ASML AI Demand Story

The next phase will be judged by customer spending, ASML’s expansion decisions, and the conversion of distant interest into executable orders.

The first signal is the capital-spending guidance from TSMC, Samsung, Intel, and major memory producers. Their budgets will show whether equipment demand remains broad.

TSMC deserves particular attention because it manufactures advanced processors for many AI chip designers. A sustained commitment to leading-edge capacity would reinforce ASML’s outlook.

Memory spending matters for a different reason. AI accelerators depend on high-bandwidth memory, so SK Hynix, Samsung, and Micron must expand alongside logic production.

If several manufacturers maintain or raise capital budgets, ASML’s supply warning gains credibility. If plans diverge sharply, demand may be concentrated in fewer customers than the headline suggests.

The second signal is ASML’s final capacity plan for 2028. Management is investigating expansion beyond the previously discussed 30% increase.

That decision will reveal how much confidence ASML places in customer forecasts for 2029 and 2030. It will also show whether suppliers can support a faster ramp.

A confirmed expansion would strengthen the view that demand exceeds a temporary quarterly surge. Repeated delays would suggest operational constraints are harder to solve.

The company’s approach to upgrades will be equally important. More throughput from installed machines could ease shortages without requiring every incremental wafer to depend on a new system.

The third signal is the quality of ASML’s order conversion. Expressions of interest are useful, but firm bookings and scheduled shipments carry greater financial meaning.

Readers should compare future bookings with revenue, customer readiness, and delivery capacity. A growing backlog is constructive only when ASML can convert it without repeated postponements.

Orders for High-NA EUV also deserve scrutiny. Broader adoption would indicate that customers are preparing for process technologies beyond today’s production generation.

Slow adoption would not erase demand for established EUV systems. It would, however, alter ASML’s product mix and the timing of future revenue.

The company’s next full-year guidance should connect these signals. Investors need more than a statement that demand exceeds supply.

They need to understand expected system output, upgrade contributions, gross margins, supplier readiness, and the timing of customer installations.

For developers and AI product teams, these manufacturing details can feel distant. Yet they influence the availability and operating cost of the computing resources their products consume.

Enterprise buyers face a similar connection. Longer equipment lead times can flow through chip production, server delivery, cloud capacity, and deployment schedules.

That does not mean teams should plan around every lithography order. It means AI infrastructure should not be treated as an unlimited resource with predictable availability.

ASML Raises Full-Year Sales Outlook because its customers expect the AI buildout to continue. The stronger conclusion is that those customers are committing to a longer hardware cycle.

The remaining question concerns execution. Can ASML, its suppliers, and chipmakers turn multiyear interest into capacity without outrunning profitable demand?

Watch the next customer budgets, ASML’s 2028 capacity decision, and firm bookings for later years. Together, those signals will show whether today’s shortage becomes durable growth or tomorrow’s excess.