€8.77 billion. That's what ASML pulled in during Q1 2026.
Wall Street expected €8.5B. ASML blew past that. Net income came in at €2.76B versus the €2.54B consensus. But the real headline is what happened next: ASML raised its full-year guidance to €36–40B, up from the previous €34–39B range. That's a €1B bump at the top end.
Why should you care about a chip equipment company's earnings? Because ASML is the only company on Earth that can build EUV (extreme ultraviolet) lithography machines. Every cutting-edge AI chip needs an ASML machine to exist. So ASML's numbers are essentially the thermometer for the entire AI semiconductor industry.
How ASML Became the Only Game in Town
ASML's monopoly took 30 years to build.
Spun out of Philips in 1984, the small Dutch startup made a bold bet on EUV technology in the late 1990s. While competitors Nikon and Canon focused on extending existing ArF (argon fluoride) lithography, ASML went all-in on a radically different approach: 13.5nm extreme ultraviolet light.
That bet took two decades to pay off. The first commercial EUV system shipped in 2017. TSMC first used EUV in its 7nm process in 2019. Today, every advanced node at 5nm, 3nm, and 2nm is impossible without EUV.
| Metric | Detail |
|---|---|
| Price per EUV system | Approximately €150–200M |
| Weight | About 150 tons |
| Component count | Around 100,000 parts |
| Tier-1 suppliers | Approximately 5,000 |
| Companies that can build EUV | ASML. That's it. |
This monopoly structure has turned ASML into the critical bottleneck of the AI age.
Inside the Q1 Numbers — Memory Flipped the Script
The most dramatic shift this quarter was in the memory-to-logic ratio.
Memory accounted for 51% of new tool sales in Q1, up from just 30% in Q4 2025. That's a 21 percentage point jump in a single quarter. The signal is unmistakable: Samsung and SK Hynix are ramping HBM (High Bandwidth Memory) production lines at unprecedented speed.
HBM is the high-speed memory essential for AI accelerators like GPUs and TPUs. The HBM3E chips in Nvidia's H200 and B200 GPUs offer more than 10x the bandwidth of standard DDR5. As AI models grow larger, they need more HBM, and making HBM requires more EUV machines. It's a virtuous cycle for ASML.
| Metric | Q4 2025 | Q1 2026 | Change |
|---|---|---|---|
| Total Revenue | Approximately €9.3B | €8.77B | -5.7% |
| Net Income | Approximately €2.7B | €2.76B | +2.2% |
| Gross Margin | 52.2% | 53.0% | +0.8pp |
| Memory share of sales | 30% | 51% | +21pp |
| Logic share of sales | 70% | 49% | -21pp |
The gross margin improvement to 53% is worth noting too. EUV systems carry higher margins than older DUV equipment, so as EUV shipments increase, profitability naturally improves.
Geographic Shifts — China's Share Collapses
The regional breakdown tells a geopolitical story.
South Korea surged to first place at 45% of total sales. Taiwan came in at 23%. China dropped to just 19%, down from 36% the prior quarter.
China's share getting cut nearly in half is a clear sign that U.S. export controls are biting. Washington has been restricting ASML's most advanced EUV exports to China since 2023, and expanded restrictions to include certain DUV (deep ultraviolet) systems in 2025. ASML CEO Christophe Fouquet flagged export control uncertainty as an ongoing risk.
Meanwhile, South Korea's 45% share reflects the simultaneous HBM line expansions at Samsung's Pyeongtaek campus and SK Hynix's Icheon and Cheongju facilities. Industry insiders are calling 2026 "year zero of the HBM war."
The Bigger Picture — AI Is Reshaping the Entire Chip Supply Chain
ASML's results aren't just about one company beating estimates.
They're evidence that AI is fundamentally restructuring the power dynamics of the semiconductor supply chain. For decades, the industry revolved around logic chips. Intel sat on the throne for a generation because PCs and servers ran on its CPUs.
But in the AI era, memory has become the critical bottleneck. No matter how fast your GPU is, it's useless without high-speed memory to feed it data. HBM is the key that unlocks that bottleneck, and only three companies in the world can make it: Samsung, SK Hynix, and Micron.
This might be the first time in semiconductor history that memory matters more than logic. This AI-driven inversion could persist for the next 5–10 years.
ASML's guidance raise is the confirmation signal. Their customers are accelerating expansion plans for 2026 and beyond.
What This Means for You
If you're building with AI, ASML's numbers carry practical implications.
First, GPU shortages aren't going away soon. Even if ASML ships machines at maximum capacity, it takes 12–18 months for a new fab line to start producing chips. Equipment ordered today won't become actual silicon until late 2027.
Second, AI infrastructure costs aren't coming down anytime soon. Rising equipment prices flow into chip costs, which flow into cloud GPU hourly rates. If you're budgeting for AI compute in 2026–2027, plan for flat or rising costs rather than the declines many hoped for.
Third, the geographic concentration of AI's hardware foundation is intensifying. South Korea, Taiwan, and the Netherlands form the critical triangle of AI chip manufacturing. Disruption to any node in this triangle would ripple through the entire AI stack.
References
- Chip giant ASML raises 2026 guidance as AI semiconductor demand stays strong
- ASML Raises Revenue Forecast Amid Strong Demand for AI Chip Manufacturing Equipment
- ASML lifts 2026 forecast as surging AI chip demand boosts new orders
- AI chip demand pushes ASML to target up to €40B in 2026 sales
Originally published on spoonai.me | Daily AI briefing at spoonai.me
Top comments (0)