ASML Gave You a Two-Year Window; Don't Just Focus on Lithography Machines

When the lithography giant raises sales forecasts for the second time in a year, what's your first reaction? Bottom-fishing semiconductors? Longing AI? Or rushing to find a project with a chip concept to issue a coin?

My first reaction was: How long can this capital expenditure frenzy in AI chips last, and which piece of the pie can we entrepreneurs actually eat?

ASML's financial report is clear: AI chip demand is driving customers to frantically expand capacity, and its EUV lithography orders are booked until 2027. In the short term, this is a highly certain signal: Giants like Nvidia, TSMC, and Samsung are still pouring money into building fabs, and their hunger for computing power is far from satisfied. This means that for the next two years, the supply of high-performance GPUs will remain tight, and cloud inference costs won't drop quickly. For Web3+AI entrepreneurs, this is actually good news—if your project requires massive AI computing power, prices staying high in the short term forces you to optimize algorithms and reduce dependencies; if you're building a decentralized computing market, this window period is golden time for grabbing users and building networks.

But the question is, have downstream applications really kept up? I've seen too many projects get excited and rush into the "AI+Blockchain" track upon seeing computing demand grow, stacking nodes, mining, and airdrops, only to find users simply don't buy in. The core of Tokenomics design isn't how to distribute coins, but whether the value you create can run through a commercial closed loop. ASML's performance is good because its customers—chip manufacturers—have clear revenue expectations. What about your project? Who are you selling computing power to? Who is willing to pay? If you're just maintaining activity through token incentives, it's a castle in the air.

In the long run, I care more about two trends. First, the cyclicality of the semiconductor industry won't disappear because of AI. Current capacity expansion is immense. Once the pace of AI scenario implementation slows down (e.g., large model iterations hit a bottleneck, or killer apps take too long to appear), oversupply will bite back. Second, ASML's technological monopoly itself is a risk. Geopolitics, export controls, single-source supply chains—if any link fails, it causes tremors across the entire industry chain. The value of decentralized computing truly shines when centralized supply chains are fragile. So in the long term, I believe diversified computing networks (whether storage network upgrades like Filecoin or GPU rental platforms like Render) will gradually complement rather than replace centralized clouds.

Action advice for readers is simple: Don't follow the capital narrative. Think clearly about whose problem your product solves first. If you're building an AI+Web3 project, find paying customers first, then talk about token economics. Otherwise, when ASML's orders hit the ceiling, your project might not even pass the seed round.

In the short term, I'll focus on platforms that effectively convert idle GPU resources into usable computing power, especially vertical solutions connecting small and medium developers. In the long term, I'll bet on decentralized applications that solve "AI data privacy and ownership" issues, because that is the true irreplaceability of Web3 compared to centralized clouds. As for issuing coins, consider adjusting Tokenomics direction only after the product lands—start the business first, then raise funds. Don't put the cart before the horse.

Original Link: https://www.cnbc.com/2026/07/15/asml-2q-earnings-ai-chips-orders.html