Intel recently hosted a marquee event tied to the IFA tradeshow in Berlin to launch its new series of Lunar Lake mobile processors—a launch that has a lot riding on it at a time when Intel faces challenges on seemingly every front. I have written extensively about Lunar Lake since back in June when Intel took the wraps off of it at Computex in Taiwan. Even then, Lunar Lake felt like it was a resounding answer to the threat in PC processors posed by Apple, AMD and Qualcomm.
Intel’s dominance of the PC market has been under pressure from AMD’s and Qualcomm’s new generations of AI PC chips, which themselves are also a response to Apple’s M-Series. In this context, Lunar Lake is Intel’s demonstration of what it can achieve when it focuses.
The New Core Ultra Lineup
Lunar Lake is one of the most important parts of Intel’s Core Ultra lineup and brings this family of chips into its second generation, hence the 200-series designation. With these new parts comes a new architecture that has been overhauled across the board. In my Computex coverage of the initial Lunar Lake launch, I detailed the many changes to the Lunar Lake architecture that make it significantly better than its predecessor, Meteor Lake.
The big developments since the Computex launch are finalized frequencies, SKUs, platform details and plenty of benchmarks. To summarize all of that, Intel has significantly optimized the entire platform for low power and beefed up its low-power cores to the point where they perform like last generation’s high-performance cores. This allows these chips to keep as many workloads on the four low-power cores as long as possible without lighting up the four high-performance cores. Those CPU improvements are accompanied by significant improvements to GPU and AI performance to make Lunar Lake a Copilot+ PC-capable platform.
One interesting development I noticed from this SKU lineup is that all processors except one have 17 watts of base power and 37 watts of maximum turbo power. The outlier is the Core Ultra 9 288V (the highest spec in the lineup), which has a base power of 30 watts with a minimum power of 17 watts. This is likely because it is the top-of-the-line performance benchmarking part. (Something similar happens with Qualcomm’s top SKU.)
Compared to Intel’s own Core Ultra 165H Meteor Lake chip, which isn’t even Intel’s highest-power platform, the company is showing anywhere from 33% to 50% lower package power for a unit that includes 32GB of on-package memory. In addition to that, Intel has made comparisons to Qualcomm’s platform using the same OEM and the same chassis (likely a Dell platform), claiming better battery life in UL Procyon and Microsoft Teams benchmarks. While we will have to verify these claims ourselves, it’s worth noting that nobody could have imagined just a few years ago that Intel would have an answer (or need to have an answer) to Qualcomm’s best low-power, high-performance platform. Intel also took its best shot at AMD using the same battery-life test and the same OEM (in this case, likely ASUS).
Intel also noted differences in AI and GPU performance, namely that its competitors are having trouble getting much software to run on their platforms. A good example of that was provided in testing via Geekbench AI, which recently came out as the most comprehensive AI performance benchmark—and showed some performance deficiencies in Qualcomm’s GPU driver and AMD’s NPU. Both AMD and Qualcomm are going to need to invest heavily in both areas if they want to compete with Lunar Lake in real-world performance.
Intel clearly has invested the most in enabling developers to build for its entire SoC, even if its performance hasn’t always been up to par. We saw Intel struggle on GPU for quite some time, but the company has continued to invest to the point where it’s no longer a weakness.
New Evo Program
With a new platform and OEM systems come new expectations from Intel on how OEMs should build and test their systems to maximize the potential of Lunar Lake. The best systems that meet Intel’s 12-month-long process get the Evo badge, which is displayed prominently on the system as an indicator of a superior user experience. As part of this, Intel sets responsiveness requirements on battery power, as well as instant-on and authentication requirements of 1.5 seconds or less. There is also a requirement to hit 11 or more hours of real-world battery life with full-HD displays and 4.5 or more hours of battery life from a 30-minute charge. Intel is also requiring secured-core PC features from Microsoft for added security.
Intel is also requiring Wi-Fi 7 performance and latency, ensuring that Wi-Fi 7 on OEM systems delivers 60% better latency and at least 5Gbs better performance than Wi-Fi 6. Intel is also requiring Intel’s Connectivity Performance Suite for smart Wi-Fi connection and optimization. It’s even mandating better camera quality, which makes sense when you consider how much of a leap Qualcomm’s platform brought to camera quality for the initial Copilot+ PCs.
OEM Systems
At Intel’s event, Michelle Johnston Holthaus, who leads Intel’s client PC business, and her lieutenant Jim Johnson trotted out all of the company’s OEM partners to talk about the numerous systems they would all be bringing to market. The first and closest partner was Lenovo; they brought onstage Luca Rossi, president of the Intelligent Device Group at Lenovo, to talk about the multi-year effort to co-engineer Lunar Lake together.
Following Rossi’s discussion with Johnson about their close partnership, Dan Rogers, head of the Client Performance Marketing Lab at Intel, talked about its many free and open-source AI tools and APIs. This was followed by Holthaus reiterating Intel’s performance advantages and bringing out Microsoft’s vice president of Windows+ devices, Pavan Davuluri, to talk about Copilot+. They confirmed that Microsoft has set an official date of November for capable x86 systems to be updated to support Copilot+ features.
Holthaus also brought on stage Dell’s Sam Burd to talk about the new XPS desktop powered by the Core Ultra 200 Series, as well as HP’s Alex Cho to do the same for HP’s new systems. Wrapping up the event, Holthaus talked about pre-orders, which kicked off that day, and systems being available starting September 24. (The 24th was supposed to be Intel’s big day with the Innovation event it had planned, but it cancelled the event due to public pressure connected to its financial struggles and layoffs.) Intel also communicated that while this launch is happening this year for consumers, commercial systems will be available next year, just in time for many IT departments to transition from Windows 10 to Windows 11.
Intel’s Response To Competition Is More Than Adequate
One of Intel’s strongest messages is that a great AI PC starts with a great PC. This speaks to the fundamentals: a fast CPU, fast GPU, good battery life and overall great experience are just as important as supporting AI features. This is especially true when you consider how few on-device AI apps there are today, and even how few of those apps’ features actually use the NPU for acceleration.
I believe that Intel’s response to Qualcomm’s and AMD’s competition has shown that its PC division is capable of keeping the company afloat and keeping OEM partners happy with innovative platforms. I do believe that Intel and its partners will have a harder time juggling Arrow Lake and Lunar Lake systems later this year and next, but I believe that Intel’s client business has responded well to the competition. Lunar Lake is very real, and when I got a chance to play with numerous Lunar Lake-based systems in Berlin, their performance confirmed many of Intel’s claims. Now we just need to get systems in hand to verify these claims in detail ourselves—which shouldn’t be too long from now.
