David Brooks
The Company Nobody Knows But Everyone Needs
My verdict is clear: if I had to bet on a single company to understand the future of artificial intelligence, it wouldn't be OpenAI, NVIDIA, or Google. It would be ASML.
This morning, the Dutch company reported its Q4 2025 results, and the numbers are extraordinary: 13.2 billion euros in orders, when analysts expected 6.32 billion. That's 109% above forecasts.
For context: that's like Apple announcing they sold double the expected iPhones. But instead of phones, ASML sells $150 million machines that are the only path to manufacturing advanced chips.
The stock reacted with a 7% surge in a single day. And analysts, already bullish, raised their price targets to as high as $1,642 per share.
But the truly interesting part isn't the numbers. It's what they reveal about the real state of the AI boom.
The Most Powerful Monopoly Nobody Talks About
ASML manufactures the extreme ultraviolet (EUV) lithography machines needed to produce semiconductors smaller than 7 nanometers. Without these machines, the iPhone's A18 chip doesn't exist. NVIDIA's H100 GPU doesn't exist. The processors powering ChatGPT, Claude, or Gemini don't exist.
Market Position
| Aspect | ASML | Competitors |
|---|---|---|
| General lithography share | 90% | Nikon + Canon: 10% |
| EUV share | 100% | No competition exists |
| Market cap | $527 billion | Largest tech company in Europe |
| Price per EUV machine | $150-200 million | N/A |
| Price High-NA EUV machine | $380 million | N/A |
I won't sugarcoat it: ASML has a more absolute monopoly than OPEC has over oil. OPEC's 40% share looks modest compared to ASML's 100% in EUV.
Why There's No Competition
Getting ASML to where it is today required:
- 30 years of continuous R&D
- $9 billion invested in development
- A global supply chain with 800 suppliers coordinated
- Accumulated know-how that's practically impossible to replicate
China has been trying to develop its own EUV technology for years. The most optimistic estimates say they could have a working prototype by 2028-2030. And that's assuming U.S. export restrictions don't tighten further.
Meanwhile, ASML sells every machine it can produce.
The Numbers That Matter: Q4 2025 in Detail
Financial Results
| Metric | Q4 2025 Result | vs Expectations |
|---|---|---|
| Orders (bookings) | 13.2B EUR | 6.32B EUR (+109%) |
| EUV orders | 7.4B EUR | All-time record |
| Net sales | 9.7B EUR | 9.6B EUR expected |
| Net profit | 2.84B EUR | In line |
| Gross margin | 52.2% | Stable |
| Total backlog | 38.8B EUR | All-time high |
Full Year 2025 Results
| Metric | 2025 | 2024 | Growth |
|---|---|---|---|
| Total sales | 32.7B EUR | 28.3B EUR | +15.5% |
| Net profit | 9.6B EUR | 7.6B EUR | +26.3% |
| Gross margin | 52.8% | ~51% | +1.8pp |
| EPS | 24.73 EUR | ~19.58 EUR | +26% |
The 2026 guidance also beat expectations: between 34 and 39 billion euros in sales, when analysts expected 35.1 billion.
What Analysts Are Saying
| Analyst | New Price Target | Rating |
|---|---|---|
| Bernstein | $1,642 | Buy |
| JPMorgan | $1,518 | Buy |
| Wells Fargo | $1,450 | Buy |
| Morgan Stanley | 1,400 EUR | European top pick |
| UBS | 1,400 EUR | Buy |
Consensus is nearly unanimous: 22 buy recommendations, 1 hold, 0 sells. UBS projects 23% revenue growth in 2026 and 14% in 2027.
The Chain That Feeds AI
To understand why ASML matters, you need to understand the AI value chain:
ASML (EUV machines)
↓
TSMC / Samsung / Intel (manufacture chips)
↓
NVIDIA / AMD (design GPUs)
↓
Microsoft / Google / Amazon / Meta (buy for data centers)
↓
ChatGPT, Claude, Gemini, etc. (generative AI)
If ASML can't produce enough EUV machines, TSMC can't manufacture enough chips. If TSMC can't manufacture enough chips, NVIDIA can't deliver enough GPUs. If NVIDIA can't deliver enough GPUs, Microsoft can't scale Azure AI.
ASML is literally the bottleneck for the entire AI industry.
ASML's Customers
| Customer | Primary Use | Dependency |
|---|---|---|
| TSMC | Manufactures chips for NVIDIA, Apple, AMD | Critical |
| Samsung | Memory + logic chips | Critical |
| Intel | Own chips + foundry services | Critical |
TSMC has its 2nm capacity completely booked for 2026. Samsung just invested $773 million in two High-NA EUV machines. Intel is going "all in" on High-NA to try to recover the technological leadership it lost years ago.
They all depend on ASML producing enough machines.
The AI Boom in Concrete Numbers
ASML's record orders aren't coincidence. They're the reflection of unprecedented investment in AI infrastructure.
Hyperscaler CapEx 2026
| Company | Projected 2026 CapEx | Primary Destination |
|---|---|---|
| Amazon | >$150 billion | AWS, AI data centers |
| Microsoft | $140+ billion | Azure, OpenAI infra |
| >$100 billion | Cloud, TPUs, Gemini | |
| Meta | >$80 billion | Llama, data centers |
| Top 5 Total | ~$602 billion | +36% vs 2025 |
75% of that CapEx is destined for AI infrastructure. That's approximately $450 billion translating into servers, GPUs, data centers, and ultimately, demand for chips manufactured with ASML machines.
Christophe Fouquet, ASML's CEO, explained it in the earnings call:
"In recent months, many of our customers have shared a notably more positive assessment of the medium-term market situation, primarily based on more robust expectations around the sustainability of AI-related demand."
Translation: ASML's customers believe the AI boom is real and sustainable. And they're putting their money where their mouth is.
The Risks Nobody Mentions
It's not all optimism. ASML faces significant challenges.
China Exposure
| Year | % Revenue from China |
|---|---|
| 2024 | 41% |
| 2025 | 33% |
| 2026 (expected) | ~20% |
U.S. export restrictions are forcing ASML to drastically reduce its sales to China. Going from 41% to 20% in two years is a brutal adjustment that requires other markets to compensate.
Workforce Cuts
Amid record results, ASML announced the layoff of 1,700 employees (3.8% of workforce), part of a larger plan to eliminate 3,000 management positions. The stated goal: "reduce bureaucracy" and hire more engineers.
Sounds familiar to Big Tech's justifications for their own layoffs.
Elevated Valuation
After rising 30% so far in January and 7% today, ASML trades at valuations that assume perfect execution. Any stumble could trigger a significant correction.
Sector Cyclicality
Semiconductors are notoriously cyclical. Visibility beyond 2026 is limited, and AI demand could moderate if companies don't see returns on their investments.
What This Means for the Future of AI
After years covering the tech sector, here's what ASML's numbers tell me:
1. The AI Boom Is Real, Not Hype
When ASML's customers (TSMC, Intel, Samsung) double their orders, it's not speculation. It's real demand for production capacity for AI chips. Hyperscalers are investing $602 billion because they believe AI will generate returns.
2. The Bottleneck Is Hardware, Not Models
Everyone talks about GPT-5, Gemini 3, Claude 4. But the real limitation isn't the ability to develop larger models. It's the physical capacity to manufacture enough chips to train and run them. ASML is the gatekeeper of that capacity.
3. The Concentration of Power Is Concerning
A single Dutch company controls 100% of the EUV machine market. Three companies (TSMC, Samsung, Intel) manufacture virtually all advanced chips in the world. A single company (NVIDIA) dominates 80%+ of the GPU market for AI.
This concentration creates systemic vulnerabilities. An earthquake in Taiwan, an escalation of tensions with China, or simply a production problem at ASML could paralyze AI development globally.
4. Investment Doesn't Guarantee Success
The $602 billion hyperscalers will invest in 2026 is a bet, not a guarantee. If generative AI doesn't generate sufficient returns in the next 2-3 years, we could see a brutal correction in both AI companies and the entire supply chain, including ASML.
My Final Take
If you ask me directly: ASML is probably the "safest" investment to bet on AI without depending on which company wins the model race.
It doesn't matter if the future belongs to OpenAI, Anthropic, Google, or Meta. They all need chips. All those chips need ASML machines.
But "safe" is relative. ASML trades at valuations that assume the AI boom will continue indefinitely. If demand moderates, or if China restrictions impact more than expected, the stock could suffer.
What's not debatable: today's results confirm that money is flowing toward AI infrastructure at an unprecedented rate. And all that money, eventually, passes through a factory in Veldhoven, Netherlands.
ASML is the silent monopoly that controls the future of artificial intelligence. And after today, it's not so silent anymore.
Data in this article comes from ASML's official Q4 2025 report, Bloomberg, CNBC, and analysis from UBS, JPMorgan, and Morgan Stanley.
Frequently Asked Questions
What does ASML do and why is it important for AI?
ASML manufactures the extreme ultraviolet (EUV) lithography machines needed to produce advanced semiconductors. Without these machines, it's impossible to manufacture chips smaller than 7 nanometers, including NVIDIA's GPUs that power ChatGPT, Claude, and other AI systems. ASML has 100% of the global EUV market.
Why did ASML's orders exceed expectations?
The 13.2 billion euro orders (109% above expectations) reflect massive hyperscaler investment in AI infrastructure from companies like Amazon, Microsoft, and Google. These companies will invest over $600 billion in 2026, and a significant portion translates into demand for chips manufactured with ASML machines.
What is High-NA EUV and why does it cost $380 million?
High-NA EUV is ASML's next generation of lithography machines, with a larger numerical aperture that enables manufacturing even smaller and denser chips. The $380 million price reflects the extreme complexity of the technology: each machine has over 100,000 components from 800 different suppliers.
Who are ASML's main customers?
The three main customers are TSMC (Taiwan), Samsung (South Korea), and Intel (USA). These three companies manufacture virtually all advanced chips in the world, including Apple's processors, NVIDIA's GPUs, and Google's AI chips.
Is ASML a good investment?
ASML has a unique monopolistic position and directly benefits from the AI boom. However, the stock trades at elevated valuations after rising 37% in January, and faces risks including China export restrictions (which will reduce revenue from that market from 41% to 20%) and the inherent cyclicality of the semiconductor sector.
