ASML intends to increase the production speed of EUV machines by 30% due to the demand for AI surpassing its current manufacturing capacity.

      TL;DR: ASML is reducing the EUV build cycle from 22 weeks to 15-16 weeks. It is nearly sold out for 2027 and has significant orders for 2028. Guidance has been raised to €43-45B, with a planned 30% boost in capacity.

      ASML is aiming to shorten the time needed to manufacture and test its extreme ultraviolet lithography machines by about one-third, according to CFO Roger Dassen speaking to reporters on Wednesday. The cycle time, which is the duration from beginning production in ASML’s clean rooms to shipping, was approximately 22 weeks a few quarters ago. “We’re now aiming to reduce that to 15 to 16 weeks,” Dassen mentioned.

      The Dutch firm is "close to being fully booked" for EUV machines in 2027, with a "significant number" of orders for 2028. Dassen noted that it is "rare" to have such advance orders, indicating the aggressive capacity expansion by chipmakers for AI purposes. ASML has updated its full-year sales forecast to between €43 billion and €45 billion ($49.2 billion) and intends to boost capacity by 30% for 2027, while also "exploring" an additional 30% increase for 2028.

      The company plans to produce around 65 units of its low-NA EUV machines this year. To achieve higher production numbers, ASML is streamlining testing protocols while ensuring quality, reorganizing clean room spaces to prioritize output over R&D, and collaborating with its supply chain to eliminate bottlenecks. “We see opportunities to lessen the testing protocol while still maintaining quality,” Dassen stated. “We can produce more tools, and customers are receptive to that.” Demand for AI infrastructure is leading to long-term supply commitments at every level of the semiconductor industry, from memory to the machines that manufacture chips.

      EUV machines take over a year to deliver once an order is placed, and ASML is the only producer of these machines globally. Every advanced chip from TSMC, Samsung, and Intel relies on ASML’s equipment. The capacity limitations mean that the speed at which AI data centers can be established is ultimately dependent on ASML’s ability to deliver lithography tools. The AI memory shortage is already increasing consumer electronics prices, while ASML's production bottleneck exists further upstream, affecting the machines that create the chips used in various products.