Intel is changing its manufacturing roadmap to save money

Under CEO Pat Gelsinger, Intel (INTC 0.95%) set a plan in 2021 to develop five new semiconductor process nodes within four years. The final node, the Intel 18A, was meant to catch up with the company TSMC from the point of view of process technology.

Intel’s 18A process remains on track to be ready by the end of the year, with high-volume production slated for 2025. Several future first-party products, including Panther Lake PC processors and Clearwater Forest server processors, will be built on Intel. 18A and are scheduled to be released next year.

The Intel 18A will feature some major innovations, notably a new transistor architecture and rear power delivery. Intel’s plan was to first introduce these technologies into the Intel 20A process, which the company was originally supposed to use for this year’s PC CPU products. Given Intel’s financial difficulties, it instead effectively destroys the Intel 20A process.

I’m switching to TSMC

Intel revealed a few months ago that it will use TSMC’s 3nm process to produce Lunar Lake, the latest laptop processors shipping now, instead of Intel’s 20A. Based on Intel’s numbers, Lunar Lake systems should offer best-in-class battery life and compete effectively with AMD and Qualcomm.

Arrow Lake, Intel’s upcoming family of desktop processors, was until recently expected to use Intel’s 20A process for its motherboard. That plan is now out the window as well. The company revealed last week that it will move Arrow Lake to “external partners” for production, while also shifting resources from Intel 20A to Intel 18A.

The Intel 20A process was a bit strange. It was never going to be available to Intel’s foundry customers, and Arrow Lake was the only product slated to use it once the decision was made to outsource Lunar Lake. Spending to ramp up production of Intel 20A for a single product family would have been possible when Intel’s finances were in better shape, but certainly not now.

Intel expects to save about $500 million in capital expenditures by not expanding Intel 20A production. The company has already announced layoffs and a dividend suspension to cut expenses and may sell some non-core businesses to raise cash. Given that Intel needs to reduce capital expenditures without undermining its long-term foundry ambitions, it makes sense to put the Intel 20A on the chopping block.

Not a complete loss

While the Intel 20A will never see the light of day, the process was critical as Intel refined the core technologies that would define the Intel 18A process. In its article revealing that Arrow Lake will move away from the Intel 20A, the company noted that it has successfully integrated its new transistor design and rear power delivery. The first commercial use of these technologies will now be Intel 18A rather than Intel 20A.

Intel 18A is important for Intel products, but it is absolutely critical for the foundry business. Whether Intel has a chance to become the world’s second largest foundry by 2030 depends on customers signing up for Intel 18A. The company has already messed up Microsoft as a future Intel 18A customer, but that was the last significant customer win announced by Intel.

By shifting resources to Intel 18A earlier than originally planned, the company can better position the all-important process for the 2025 launch and production ramp to follow. Intel needs to win customers, but it also needs to scale production efficiently. What Intel 20A process development learned will help the cause.

While the Intel 20A process is dead, Intel 18A will stand on its shoulders as it drives the company’s product and foundry businesses into 2025 and beyond.