Intel has provided another foundry update on its IDM 2.0 strategy, promising process leadership with its 18A node by 2025. Recalling its past fortunes, the chipmaker acknowledged falling behind TSMC because of its 10nm woes. Intel 3 and 4 will compete against TSMC’s 5nm and 3nm processes, albeit unfavorably, ushering in the Angstrom era.
The 20A process will be roughly on par with TSMC’s 3nm node, mostly powering client lineups such as Arrow Lake and Lunar Lake in 2024-25. It will only be with the release of the 18A node in late 2025 that Intel will regain process leadership after a decade of lagging behind its East Asian rivals.
As part of IDM 2.0, Intel is outsourcing certain products to TSMC to relieve its internal foundry. The Arc GPUs and the iGPU tiles of the Core processors are the first to get this treatment, with the 15th Gen Arrow Lake-U CPUs a potential second.
With the coming node, Intel will treat its internal units like its external fabless clients. This means the internal and external fabless clients will have their P&L. In the past, internal customers have exploited their unique position with the IFS by making small iterations on chips and getting away with minimal prices.
Intel expects its first foundry customers later this year. While external customers will be charged the same as existing internal clients for cutting-edge nodes, the latter will get priority pre-volume ramping for flagship products.