![\"ASML](\"https:\/\/www.europesays.com\/netherlands\/wp-content\/uploads\/2026\/04\/c844c1858fbb88aa8ab7ab608ec60026.jpeg\")
ASML Earnings Show AI\u2019s Real Engine Is Still Hardware – Moby <\/p>\n ASML Earnings Show AI\u2019s Real Engine Is Still Hardware – Moby <\/p>\n
AI still looks unstoppable, but ASML\u2019s latest numbers suggest the story is shifting. This is no longer just about who builds the smartest model. It\u2019s about who owns the machines, the bottlenecks, and the infrastructure powering the entire system.<\/p>\n
ASML delivered a solid first-quarter update and, crucially, raised its full-year 2026 sales outlook. The Dutch semiconductor equipment giant reported revenue of around \u20ac8.8 billion (about $10.4 billion) and net income of roughly \u20ac2.8 billion, while lifting its annual sales guidance to a range of \u20ac36 billion to \u20ac40 billion.<\/p>\n
That upgrade reflects one simple reality: demand for advanced chips is still running ahead of supply. Customers, from foundries to hyperscalers, are accelerating capacity expansion plans, betting that AI demand will continue to scale well beyond current expectations.<\/p>\n
ASML sits at the center of that buildout. It is the only company in the world that manufactures extreme ultraviolet (EUV) lithography machines, the highly specialized tools required to produce leading-edge semiconductors. If chips are the brains of AI, ASML builds the printing press.<\/p>\n
But the market reaction was notably restrained. Despite stronger full-year guidance, the company\u2019s near-term outlook for the second quarter came in slightly below expectations, and the stock had already rallied heavily into the release.<\/p>\n
At the same time, developments elsewhere in the AI ecosystem reinforced the broader trend. Meta expanded its long-term partnership with Broadcom to develop custom AI chips and scale infrastructure over the coming years. The agreement includes large-scale deployment of compute capacity and next-generation chip design, extending the buildout well into the end of the decade.<\/p>\n
Taken together, the message is clear: AI spending is not slowing. It is evolving, from chip demand alone into full-stack infrastructure investment.<\/p>\n
For the past two years, the AI trade has been dominated by a simple narrative: more models, more compute, more chips. That narrative is still intact, but it\u2019s no longer sufficient.<\/p>\n
ASML\u2019s results highlight the next phase: AI as industrial infrastructure.<\/p>\n
This matters because infrastructure cycles behave differently from software cycles. They are slower, more capital-intensive, and far more dependent on physical constraints. You cannot scale lithography machines, power grids, or data center networks overnight. The system moves at the pace of its weakest link.<\/p>\n
And right now, there are a lot of weak links.<\/p>\n
Start with manufacturing. ASML is ramping production of its EUV systems, but even its own capacity expansion, targeting dozens of units per year, underscores how finite supply remains. Each machine is extraordinarily complex, expensive, and time-consuming to build. That creates natural bottlenecks.<\/p>\n
Story Continues <\/p>\n
Then there is the broader supply chain. Advanced packaging, memory, networking, and power infrastructure all need to scale in parallel. If one segment lags, the entire AI buildout slows. Investors used to focus on GPUs alone. Now the constraint set is much wider.<\/p>\n
This is why the Meta\u2013Broadcom partnership matters. It signals that hyperscalers are no longer relying on a single supplier or architecture. They are building entire ecosystems: custom silicon, cloud capacity, networking fabrics, and hybrid deployment strategies. AI is becoming less about a hero product and more about system integration.<\/p>\n
That shift has two implications.<\/p>\n
First, the value pool is expanding. It is no longer just chip designers capturing upside; it is also equipment makers, networking firms, and infrastructure providers. The trade is broadening.<\/p>\n
Second, the risk profile is changing.<\/p>\n
In the early phase of the AI boom, the main risk was whether demand would materialize. Today, demand is not the question. The risks are more nuanced.<\/p>\n
One is valuation. Many of the key players in the AI supply chain have already seen significant share price appreciation. Strong earnings may not move stocks meaningfully if expectations are already elevated. ASML\u2019s muted reaction is a case in point.<\/p>\n
Another is policy. Export controls remain a live issue, particularly around China. Restrictions on advanced semiconductor equipment can distort demand patterns, with customers pulling orders forward or shifting investment geographically. That can create short-term strength that does not fully reflect underlying trends.<\/p>\n
Then there is the question of returns.<\/p>\n
AI infrastructure spending is enormous and still accelerating. But capital intensity cuts both ways. If too much capacity is built too quickly, pricing power can erode. The industry could find itself in a familiar position: high investment, strong demand, but uneven profitability.<\/p>\n
In other words, the narrative is maturing. The easy phase, \u201cAI is coming\u201d, is over. The harder phase, \u201cwho actually makes money from it\u201d, is just beginning.<\/p>\n
The next chapter of the AI trade will be defined less by announcements and more by execution.<\/p>\n