The ASML Customer Co-Investment Program is the strangest financing deal I have watched in this industry: the biggest chipmakers on earth are buying chunks of their own most important supplier. Intel led it this summer, fresh off the three-year FinFET lead its 22nm tri-gate process just opened up, with TSMC and Samsung following, each taking an equity stake in ASML and piling research money on top, all to make sure the machines they will need in five years actually get built. The logic underneath is not subtle. EUV and 450mm now cost more than ASML can bankroll on its own, and the customers would rather buy a piece of the risk than wake up one morning with no next machine to order.
Intel is putting in the most: a stake of around 15 percent and the largest slice of the research funding, with TSMC and Samsung in for smaller pieces. Together the three are buying close to a quarter of ASML and pledging well over a billion euros toward research, aimed at the two hardest problems on the roadmap, EUV and the jump to larger 450mm wafers. No board control, no exclusivity, nobody taking over ASML. What the customers are really doing is pre-paying for a future they cannot build themselves, and that alone resets how this industry thinks about who owns whom.
Years of cash have gone into EUV with nothing to sell at the end of it, which makes it the most expensive development program chipmaking equipment has ever seen. I wrote three years ago that the light source was losing, and it still has not won. ASML could keep funding the work alone, but doing that while also chasing 450mm would stretch the balance sheet thin and slow both at once. The co-investment money lets it push on every front without betting the company on any single one. For a firm that nearly folded in its early years, spreading the risk across its own customers is a shrewd piece of financial engineering.
ASML is buying Cymer, the San Diego company that has been supplying its EUV light sources, and that matters more than all the equity headlines. The source is the exact bottleneck that has kept the whole program stuck, so owning it lets ASML stop negotiating across a supplier boundary and run source and scanner as one engineering problem. It is the most important call the company has made since it committed to immersion in the 193nm days. The light source is where EUV lives or dies, and ASML has decided it cannot afford to leave that in anyone else’s hands.
More chips per pass and a lower cost per chip on paper: that is the 450mm pitch, the second half of what these checks are funding, a jump from today’s 300mm wafers to ones with more than twice the area. The catch is that almost every tool in the fab has to be redesigned, not just the scanner, and the whole supply chain has to move together or nobody moves at all. I am skeptical it lands anywhere near as fast as its backers hope. The economics only work if the entire industry commits, and the only players with the appetite to fund it are the same three that just signed these checks. Coordinated jumps like this stall the second one big participant blinks.
Strip out the percentages and what these three actually bought is insurance: protection against ASML stumbling, and a louder voice on a roadmap they cannot live without. Control was never on offer, and ASML made sure of that. The strangeness of it stays with me. An equipment supplier so critical that its biggest customers will fund its research and hold its stock just to keep it breathing is not how this business is supposed to work. It is a measure of how narrow the road to the next node has gotten. Veldhoven sells the only machine that matters, and the most powerful companies in the industry just paid for the privilege of standing in line for it.