What Intel Investors Need to Know About Lunar Lake | The Motley Fool

It’s only been about seven months since Intel (INTC 1.05%) launched its Meteor Lake laptop CPUs (central processing units). Meteor Lake chips were the first PC CPUs from Intel to include built-in artificial intellience (AI) accelerators, the first to use the Intel 4 manufacturing process, the first to opt for a disaggregated multitile architecture, and the first to use Intel’s Foveros advanced packaging technology at scale.

Meteor Lake was a stepping stone for Intel as it worked to regain its edge against rival AMD, ward off threats from chips based on Arm Holdings‘ technology, and plant its flag in the AI PC market. Intel’s Lunar Lake laptop CPUs will build on that foundation when they become available in the third quarter, early enough for dozens of systems to flood the market in the lead-up to the holiday season.

Intel is moving fast by launching Lunar Lake less than a year after Meteor Lake. While Meteor Lake was a solid product, Lunar Lake has the potential to be a game changer for Intel.

Massive performance gains

Lunar Lake is focused on power efficiency, but the chips will also pack a performance punch. Like the last few generations of Intel’s PC chips, Lunar Lake will feature a mix of powerful P-cores and power-sipping E-cores. Both are getting a big upgrade.

Intel has overhauled the design of its P-cores for Lunar Lake, doing some major rearchitecting to maximize performance. Intel claims that its new P-cores deliver a 14% increase in instructions per clock, compared to the P-cores in Meteor Lake.

The E-cores are getting an even bigger makeover. Depending on the type of workload, Intel claims that its Lunar Lake E-cores deliver a gain of between 38% and 68% in instructions per clock, compared to the low-power E-cores in Meteor Lake.

Intel is dropping hyperthreading, a feature that allows multiple threads to run on a single CPU core, from Lunar Lake. Hyperthreading can help performance in multithreaded workloads but also takes up space on the chip. Intel found that by removing hyperthreading completely, it was able to boost performance efficiency by 15% and performance per area by 10%.

Graphics and AI are also getting big upgrades. Intel claims Lunar Lake will deliver a 50% jump in graphics performance, compared to Meteor Lake, by utilizing the new Battlemage architecture. Meanwhile, the AI processor will deliver as much as a fourfold improvement in performance over the AI processor in Meteor Lake.

Turning to TSMC

Intel will manufacture very little of its Lunar Lake chips. While the compute tile for Meteor Lake used the Intel 4 process, Intel has opted for a 3-nanometer process from TSMC for Lunar Lake’s compute tile. This could just be an issue of timing.

Right now, TSMC’s 3nm process is the best available. This situation might change as Intel rolls out Intel 20A, which will be used by its Arrow Lake desktop PC CPUs later this year, and Intel 18A. But with Intel pushing Lunar Lake out the door as quickly as possible, TSMC’s 3nm process likely made the most sense for maximizing performance and efficiency.

Intel’s splitting of its manufacturing operations from its product groups enables these kinds of decisions. The product groups are no longer tethered to manufacturing, meaning that manufacturing delays will no longer cause cascades of product delays and derail Intel’s product roadmap. Over time, Intel plans to bring more manufacturing in-house as it launches new process nodes and boosts capacity.

Lunar Lake will be up against AMD’s upcoming Ryzen AI 300 chips based on its Zen 5 architecture, as well as Qualcomm‘s new Arm-based PC chips. Lunar Lake has the potential to be a groundbreaking product, although competition will be stiff as all three companies jockey for position in the AI PC market.