ASML CEO: AI Chip Shortage Will Last Years

ASML Warns AI Chip Crunch Has No Quick Fix

ASML, the Dutch semiconductor equipment giant that holds a global monopoly on the most advanced chipmaking technology, says the world will face AI chip shortages for years to come. CEO Christophe Fouquet points to surging demand from hyperscalers and AI companies as the primary driver, even as the company ramps up production of its latest — and most expensive — extreme ultraviolet (EUV) lithography systems.

The warning comes as Microsoft, Meta, Amazon, and Google collectively plan to spend more than $600 billion on AI infrastructure this year alone, sending demand for ASML's equipment to unprecedented levels. With no competitor capable of producing EUV machines, ASML finds itself at the center of the global AI buildout — and at the heart of a supply bottleneck that shows no signs of easing.

Key Takeaways

• ASML's CEO says AI chip shortages will persist for multiple years, driven by massive infrastructure spending from Big Tech
• The company invests €4.5 billion ($5 billion) annually in R&D to stay ahead in lithography technology
• Each EUV machine costs between $200 million and $400 million+, depending on the generation
• ASML is Europe's most valuable company, with a market cap exceeding $530 billion
• The 4 largest U.S. tech companies plan $600B+ in AI spending this year, fueling equipment demand
• New-generation machines, while pricier upfront, reduce per-chip manufacturing costs through improved throughput and resolution

The Only Game in Town: ASML's EUV Monopoly

ASML's dominance in semiconductor manufacturing equipment is unlike anything else in the tech industry. The 42-year-old company, headquartered in Veldhoven, Netherlands, is the sole producer of EUV lithography systems — the machines that etch the microscopic circuit patterns onto silicon wafers used to make the world's most advanced chips.

Every time you interact with an AI chatbot, generate an image, or use a recommendation algorithm, you are relying on chips that were made using ASML equipment. The company's 44,000 employees and hundreds of suppliers work together to build machines roughly the size of a school bus, each taking months to assemble.

No other company — not in the United States, Japan, South Korea, or China — has managed to replicate this technology. This monopoly position has made ASML a linchpin of the entire global semiconductor supply chain, and by extension, the AI revolution itself.

Why the Shortage Won't End Soon

The math behind the chip shortage is straightforward but daunting. Demand for AI accelerators — primarily GPUs from NVIDIA, but also custom chips from Google (TPUs), Amazon (Trainium), and others — is growing exponentially. Yet the machines needed to manufacture these chips cannot be produced overnight.

Each EUV system involves an extraordinarily complex supply chain spanning hundreds of vendors. Assembly takes months, and qualifying a new machine at a customer's fabrication plant (or 'fab') can take additional months. Even if ASML were to double its output tomorrow, the lead times and installation complexities would prevent an immediate impact on chip supply.

Fouquet's assessment aligns with what chipmakers like TSMC and Samsung have been signaling: capacity expansion is underway, but it takes 2 to 3 years to bring a new fab online. The current wave of AI infrastructure spending is simply outpacing the industry's ability to build manufacturing capacity.

New Machines Cost More — But Deliver More Value

ASML's latest generation of lithography equipment — the High-NA EUV systems — carries price tags exceeding $400 million per unit, roughly double the cost of the previous generation. Even ASML's largest customers, including TSMC, Samsung, and Intel, have reportedly hesitated at these prices.

However, ASML argues that the economics ultimately favor adoption. The new machines offer several critical advantages:

• Higher resolution enables smaller, more densely packed transistors, increasing chip performance
• Improved throughput means more wafers processed per hour, reducing per-chip cost
• Fewer patterning steps required compared to using older-generation machines for equivalent results
• Better yield rates as the technology matures, reducing waste and defective chips
• Future-proofing for next-generation chip designs at 2nm and below

In other words, while the sticker price is eye-watering, the total cost of ownership can actually be lower when measured on a per-transistor or per-chip basis. For companies racing to build AI chips at scale, the calculus increasingly points toward investing in the latest equipment rather than relying on older, cheaper machines that require more complex multi-patterning techniques.

Big Tech's $600 Billion AI Bet Fuels Equipment Demand

The scale of investment pouring into AI infrastructure in 2025 is staggering. The combined capital expenditure plans of just 4 companies — Microsoft, Meta, Amazon, and Google — exceed $600 billion this year. Much of that money flows toward data centers packed with AI accelerators, which in turn drives demand upstream to chipmakers and, ultimately, to ASML.

Microsoft alone has signaled plans to spend more than $80 billion on AI-capable data centers. Meta has committed roughly $65 billion. Amazon and Google are on similar trajectories. These figures represent a dramatic escalation from just 2 years ago, when combined spending was a fraction of current levels.

This demand cascade creates a multiplier effect throughout the supply chain:

• Hyperscalers order chips from foundries like TSMC
• TSMC orders lithography equipment from ASML
• ASML orders components from its network of specialized suppliers
• Suppliers must scale their own capacity, creating further bottlenecks

Each link in this chain has its own lead times and capacity constraints, making it nearly impossible to quickly resolve shortages at the end-user level.

ASML Becomes Europe's Most Valuable Company

ASML's strategic importance has not gone unnoticed by investors. The company's market capitalization now exceeds $530 billion, making it the most valuable publicly traded company in Europe. To put that in perspective, ASML is worth more than most European banks, automakers, and energy companies combined.

The company's stock has benefited from a confluence of factors: its monopoly position in EUV technology, the AI-driven surge in chip demand, and the geopolitical importance of semiconductor supply chains. Governments in the U.S., Europe, and Asia have all identified semiconductor manufacturing as a national security priority, further reinforcing ASML's strategic value.

ASML's annual R&D spending of €4.5 billion (approximately $5 billion) is one of the highest in the global tech industry relative to its size. This relentless investment in next-generation lithography helps ensure that no competitor can realistically catch up. Building an EUV system from scratch would require decades of research, billions of dollars, and access to highly specialized components that ASML's supply chain has spent years developing.

What This Means for the AI Industry

For companies building AI products and services, ASML's warning about prolonged chip shortages has significant practical implications. Access to cutting-edge AI hardware — particularly the latest NVIDIA GPUs and custom accelerators — will remain constrained and expensive for the foreseeable future.

Developers and businesses should consider several strategic responses. First, cloud-based AI services from major providers may offer more reliable access to compute than purchasing hardware directly. Second, model optimization techniques like quantization, distillation, and pruning can reduce the amount of compute required for inference. Third, alternative architectures — including smaller, more efficient models — may prove more practical than chasing the largest possible model size.

Startups in particular face a challenging environment. The cost of training frontier AI models continues to rise, and the hardware needed to do so is increasingly allocated to the largest, best-funded players. This dynamic could accelerate consolidation in the AI industry, favoring companies with deep pockets and existing relationships with chip suppliers.

Looking Ahead: A Multi-Year Capacity Buildout

The semiconductor industry is in the early stages of what will likely be a multi-year capacity expansion cycle. New fabs are under construction in the United States (led by TSMC's Arizona facility and Intel's Ohio plants), Europe (Intel's Germany project and TSMC's Dresden fab), and across Asia.

But building a fab is just the beginning. Equipping it with ASML's lithography systems, qualifying the manufacturing process, and ramping to full production volume takes additional years. Industry analysts estimate that meaningful relief from the current chip shortage is unlikely before 2027 or 2028 at the earliest.

ASML, for its part, is working to increase its own production capacity, hiring aggressively and expanding its facilities in the Netherlands and elsewhere. The company has signaled that it expects revenue growth to continue well into the second half of the decade, driven by sustained demand for both its current EUV systems and the newer High-NA variants.

For the global AI ecosystem, the message is clear: the bottleneck at the lithography level is real, it is structural, and it will shape the pace of AI progress for years to come. Every AI breakthrough — from GPT-scale language models to autonomous driving systems — ultimately depends on the availability of advanced chips, and those chips depend on ASML's machines. In the AI supply chain, ASML is not just a link — it is the foundation.