Applied Materials Inc. has begun selling a chipmaking machine designed to decrease the industry’s reliance on ASML Holding NV and more cheaply produce the kind of ultra-powerful semiconductors that can handle artificial intelligence tasks.
(Bloomberg) — Applied Materials Inc. has begun selling a chipmaking machine designed to decrease the industry’s reliance on ASML Holding NV and more cheaply produce the kind of ultra-powerful semiconductors that can handle artificial intelligence tasks.
The company’s Centura Sculpta machine — a so-called pattern shaping system — lets customers reduce the amount of time they spend on lithography, the process of using light to burn lines into silicon. Lithography has become increasingly complex and expensive, and the new approach will help streamline chip production while reducing waste, Applied Materials said Tuesday.
The move threatens to disrupt a lithography market dominated by ASML’s machines. Though Applied Materials isn’t challenging that company directly, it’s attempting to rethink the way the industry manufactures chips — the tiny electronic components that are built by depositing materials on disks of silicon.
Applied Materials gained 4.3% to $116.86 as of 3:18 p.m. in New York. ASML fell 1.9% to €584 ($618) in Amsterdam.
The patterns that allow chips to store and process data are burned onto wafers and then later filled with metal and other substances. The lithography stage of the process has become increasingly difficult as circuits shrink down to dimensions that are smaller than the wavelength of most light sources.
ASML’s extreme ultraviolet machines are able to handle these tasks for cutting-edge semiconductors, and it’s a lucrative source of revenue for the Dutch company. Sales have nearly doubled since 2019, and ASML is expected to post revenue growth of 25% this year. Compare that with the rest of the chip equipment industry, which is bracing for a sales contraction as customers cope with an inventory glut.
ASML’s machines are some of the most expensive in an industry where equipment can easily cost more than $10 million apiece. But ASML’s approach has drawbacks. Some parts of the chip require what’s called multiple patterning — taking two or more runs through the process to achieve the desired results.
Applied Materials’ new machine will remove the need for some multiple patterning by using an electrically activated chemical technique to complete that part of the process.
The idea is to “complement” ultraviolet lithography, “while also tackling the growing economic and environmental challenges of advanced chipmaking,” Applied Materials executive Prabu Raja said in a statement.
The new tool could add $200 million to $300 million in annual sales for Applied Materials, but won’t dent ASML’s extreme ultraviolet revenue in the near term, according to Krish Sankar, an analyst at Cowen & Co.
Chipmakers that adopt the system will save as much as $250 million in capital costs if they’re running a line that processes 100,000 wafers per month, according to Applied Materials. Each wafer will cost about $50 less and require 15 fewer kilowatt hours of energy. It also will need 15 fewer liters of water and result in 0.35 kilograms less in carbon dioxide emissions, the company said.
The new offering is primarily aimed at the production of what’s called advanced logic — massive, complicated chips such as Nvidia Corp.’s graphics processors used in AI. Only three companies specialize in that type of production: Taiwan Semiconductor Manufacturing Co., Samsung Electronics Co. and Intel Corp.
Combined, those companies make up the majority of spending on chipmaker gear and account for more than half of sales for both Applied Materials and ASML.
Applied Materials cited Intel as a company that has already embraced the new approach.
“Intel will be deploying pattern-shaping capabilities to help us deliver reduced design and manufacturing costs, process cycle times and environmental impact,” Intel executive Ryan Russell said in the statement.
(Updates shares in fourth paragraph.)
More stories like this are available on bloomberg.com
©2023 Bloomberg L.P.