Intel Launches Lunar Lake Core Ultra 200V CPUs for AI PCs

Today, Intel is officially launching its Lunar Lake architecture at the Innovation For All (IFA) conference in Berlin. These CPUs announce the company’s arrival to the AI PC market thanks to the CPUs’ embedded NPU. They also mark a radical shift in design philosophy, with many notable firsts for the company. They’re the first Intel CPUs with a compute tile made on a TSMC node, the first to include embedded memory like Apple’s M-series chips, the first with its all-new P-core and E-core design, and the first CPU to feature its second-generation Battlemage GPU architecture. In other words, it’s practically an entirely new CPU from the ground up.

Intel first unveiled Lunar Lake in May at Computex, offering multiple deep dives into the architecture and some granular information about its GPU, NPU, and overall design. Today, it’s announcing the rest of the details, including the CPU SKUs, efficiency numbers, and performance claims. Overall, Intel thinks Lunar Lake will be a big deal for the thin-and-light notebook market. Its launch marketing includes phrases such as “historic x86 perf/w,” “up to 50% lower power,” and other headline-grabbing statements you typically see in a PowerPoint presentation. Despite the typical marketing bluster, Lunar Lake has much going for it, unlike Meteor Lake, which never seemed to blow anyone’s hair back.

Intel is positioning Lunar Lake as a forward-looking CPU with DP 2.1 and HDMI 2.1, as well as PCIe 5 for SSDs.
Credit: Intel

To recap, Lunar Lake features all-new performance cores dubbed Lion Cove and new efficiency cores named Skymont. Every Lunar Lake CPU features a 4+4 design with four P-cores and four E-cores, and either 16GB or 32GB of embedded LPDDDR5 memory. These new cores were made on TSMC’s 3nm process, making them the second chip family to utilize TSMC’s most cutting-edge node, with the other being Apple’s M3/M4 chips. For context, AMD’s Zen 5 laptop CPUs and Qualcomm’s Snapdragon X Arm-based CPUs are made on TSMC’s 4nm process, so Intel has a slight advantage here via purchasing a more expensive process.

Overall, Lunar Lake consists of nine SKUs, with a single Core Ultra 9 chip just for bragging rights, as it’s almost the same as the top-tier Core Ultra 7 CPU. Every Lunar Lake CPU features a 4+4 core configuration, and they all top out at 37W, so the main difference comes down to clock speeds, memory allotment, and performance of the embedded GPU and NPU. Memory varies between 16GB and 32GB, with clocks in the 4.5GHz to 5.1GHz range and GPU and NPU performance decreasing as you go down the stack.

There’s only modest changes between the CPUs due to the incredibly tight power envelopes they’re all working with.
Credit: Intel

The flagship CPU is the Core Ultra 9 288V, which sports a 5.1GHz maximum boost clock, a full-powered Xe2 GPU, 48 TOPS NPU, 32GB of LPDDR5 8533 MT/s memory, and a 17W minimum power draw. The chip below it—the Core Ultra 7 268V—has almost the exact same specs but slightly lower clocks for the CPU and GPU and a minimum power draw of just 8W. The lineup only includes Core Ultra 5, 7, and 9 branding, so no Ultra 3 chip exists.

Intel says Lunar Lake is superior to its Meteor Lake mobile chips, Qualcomm’s Snapdragon X Elite/Pro Arm CPUs, and AMD’s Zen 5 CPUs by offering better performance at less power. Overall, it claims a 1.2x performance-per-watt advantage over Qualcomm’s chips, a whopping 2.29x advantage over Meteor Lake, and better overall performance than Zen 5, depending on the metric.

Intel claims its all-new Lion Cove P-cores will offer sizable gains over its Arm and x86 rivals.
Credit: Intel

The built-in GPU is Intel’s first to use its all-new Battlemage architecture, which the company says offers class-leading performance for a thin-and-light notebook. Overall, Intel claims it has a 68% advantage over the Qualcomm Snapdragon X Elite chip and a 16% uplift compared with AMD’s Zen 5 Ryzen AI 9 HX 370 CPU. It also has 31% more performance than the Arc Alchemist GPU in the Core Ultra 7 155H CPU.

Intel previously demo’d the Xe2 GPU running a AAA title at 60fps using ray tracing, but we’ll have to see how it performs in the real world before we draw any conclusions.
Credit: Intel

Regarding AI performance, Lunar Lake lands between Qualcomm and AMD with a rating of 48 TOPS. AMD’s Zen 5 mobile CPUs boast a 50 TOPS rating, and Qualcomm is at 45. But Microsoft says just 40 are needed to be certified as a Copilot+ PC, so Intel is certainly competitive. How these chips perform in AI tasks depends on the benchmark, but unsurprisingly, Intel says it’s more powerful than its competition using a handful of AI benchmarks. This category is so new that it’s difficult to say if it will matter to end users. Still, it’s a new front in the PC arena, so there will be fierce competition for bragging rights here going forward.

We’re not sure when reviews will show up with these CPUs, but we expect them to be imminent. This is Intel’s first time trying out TSMC’s node instead of its own for its compute tile, so we’re very curious to see how it performs. We’re even more curious to see how these same Lion Cove and Skymont cores perform in the desktop CPUs named Arrow Lake, but we will probably have to wait for October for that news to arrive.