Intel Is Already Mass-Producing Chips on the Intel 3 Node. Its Competition With TSMC and Samsung Has Begun

  • Intel’s factories in Oregon and Ireland are already mass-producing chips on the Intel 3 node.

  • As manufacturing consolidates on the 18A node, Intel will fine-tune the 14A node (1.4 nm).

Juan Carlos López

Senior Writer

An engineer by training. A science and tech journalist by passion, vocation, and conviction. I've been writing professionally for over two decades, and I suspect I still have a long way to go. At Xataka, I write about many topics, but I mainly enjoy covering nuclear fusion, quantum physics, quantum computers, microprocessors and TVs. LinkedIn

Intel is delivering on its promises. In its revised roadmap published at the end of February, the company said it would begin large-scale production of integrated circuits on its Intel 3 node by mid-2024. Today, its factories in Oregon and Ireland are already mass-producing these semiconductors. TSMC and Samsung started making 3 nm chips in 2022, so it’s evident that they’ve gained experience with this integration technology that Intel still doesn't have.

Before continuing with the article, there's one thing worth mentioning. Nanometers no longer faithfully reflect the length of logic gates or any other physical parameter, such as the distance between transistors. Each chipmaker uses them freely, leading to an almost complete disconnection between the nomenclature and the physical reality of integrated circuits. Nevertheless, TSMC, Intel, and Samsung produce “3 nm class” semiconductors on ASML’s extreme ultraviolet lithography (EUV) equipment, and they're destined to compete with each other.

The chips Intel produces on its Intel 3 node in its factories in Oregon and Ireland are specifically for data centers, so they won't reach the consumer market in the short-term. However, it's interesting that Intel has promised that this new node will deliver a performance that's 18% higher with the same power consumption and transistor density as the Intel 4 node. On the other hand, during the second half of this year, the company will start manufacturing chips with the Intel 3-T variant, which will use Foveros Direct 3D packaging.

The First Chips Made on the High Aperture EUV Machine Will Arrive in 2026

Intel intends to consolidate its position as the world’s second-largest semiconductor manufacturer by the end of this decade. That sentence may not seem to say much, but nothing could be further from the truth. It says a lot. And it's because it makes reality into something official. The reality is that the gap that the Taiwanese company TSMC, the world’s largest integrated circuit manufacturer, maintains over its two most advanced competitors, Intel and Samsung, is insurmountable in the short- and medium-term.

In the second half of this year, the company plans to be ready to manufacture integrated circuits on the 18A node.

At this point, Intel has turned its attention to Samsung. Currently, TSMC’s global market share is just over 50%, while Intel’s and Samsung’s are in the 17-20% range. A realistic plan requires paying attention to the closest competitor and taking steps to differentiate and surpass them. Reaching this milestone will require Intel to develop competitive lithography nodes, and its plan calls for the Intel 20A node to be ready in the first half of 2024. In the second half of this year, the company plans to be ready to manufacture integrated circuits at the 18A node. But as we can see in the image below, the roadmap doesn’t end there.

Once Intel consolidates its chip production on the 18A node, it will fine-tune the 14A node (1.4 nm). The roadmap above doesn’t reveal when this integration technology will be ready. Still, it’s reasonable to assume that it'll be in 2026 because in 2027, as the image shows, Intel anticipates that 14A-E lithography will be ready, which will be a revision of the original 14A integration technology.

In any case, one should pay attention to another critical fact: Lithography 14A will be the first time where Intel uses the new and expensive EUV and High Aperture (High-NA) equipment manufactured by ASML. Each machine costs about $375 million. Intel is currently testing one of these machines at its Hillsboro facility in Oregon, so it makes sense that its engineers will spend two years mastering and optimizing the processes involved in operating them.

Image | Intel

More info | Tom’s Hardware

Related |  All the Data on Earth Can Fit in a Cup Full of DNA. This Is MIT’s Jurassic Park-Inspired Project

See all comments on https://www.xatakaon.com

SEE 0 Comment

Cover of Xataka On