Home Web timeline Apple and Intel are likely to be the first to use TSMC’s 2nm node – in 2025 • The Register

Apple and Intel are likely to be the first to use TSMC’s 2nm node – in 2025 • The Register

0

Apple and Intel will likely become the first customers of TSMC’s advanced 2nm manufacturing process when the node enters production in late 2025, according to new reports.

The reports, which come from Taiwanese outlets DigiTimes and UDNare completed by a financial analyst suggestion that Intel plans to use TSMC’s 2nm process node, officially known as N2, for the graphics tile of its next-generation client processor, named Lunar Lake. Intel will also use its own 18A node, which the chipmaker says is equivalent to a 1.8nm process, for the Lunar Lake CPU title.

More broadly, Intel plans to use TSMC processes for GPUs and various system-on-chips, DigiTimes said. This matches previous statements from Intel, including one regarding its plans to use 18A and an external process for Lunar Lake.

Intel has said Lunar Lake is planned for 2024, so we don’t yet know how TSMC’s late 2025 high-volume production schedule for 2nm fits into that, besides knowing that the 2nm node will go into production at risk by the end of 2024.

We don’t yet have any indication of which chips Apple will manufacture using TSMC’s 2nm node.

These reports came shortly after TSMC revealed its production schedule for 2nm, which, at least on paper, shows that Intel could catch up to TSMC in peak nodes over the next two years. Indeed, Intel recently revealed that production of its 18A node will begin in the second half of 2024, months ahead of the previously given 2025 timeline.

Meanwhile, AMD, Broadcom, Nvidia and MediaTek will apparently have to jockey for their own allocation of TSMC’s 2nm node, according to the DigiTimes report, which says fabless chip designers are expected to start discussions about it next year. .

Intel’s plan to rely on foundries like TSMC in addition to its own fabs is part of the chipmaker’s new IDM 2.0 strategy, an evolution of its embedded device maker model. Its goal is to manufacture the highest performing chips possible, whether based on in-house or external fabs, after the company experienced setbacks with its 10nm and 7nm processes in recent years.

“This will provide us with the increased flexibility and scale we need to optimize our cost, performance, schedule and supply roadmaps, giving us a unique competitive advantage,” said the CEO of Intel, Pat Gelsinger, last year.

At the same time, Intel is dramatically increasing its manufacturing capacity, not only for its own products, but also for its revitalized foundry business, which competes with TSMC. It’s a weird dynamic, sure, but apparently TSMC CEO CC Wei isn’t concerned, based on recent comments.

“We don’t foresee any problems. And looking ahead, in fact, this IDM might bring their business back to their own home,” Wei said in reference to Intel during TSMC’s recent earnings call. [PDF]. “We have already factored this into our capacity planning.” ®