Analyst(s): Olivier Blanchard
Publication Date: 09052024
Document #: MCNOAB202409
What is Covered in this Article
- Summary of Intel’s Lunar Lake Core Ultra announcement
- Lunar Lake Core Ultra specs and performance highlights
- SITREP of the AI PC competitive landscape
- Impact of the Transition to Windows 11 on the PC Refresh Supercycle
- Intel vs AMD vs Qualcomm: How and why the AI PC race is about to get very hot
The News: Intel (NASDAQ: INTC) announced its first set of next-generation Core Ultra processors, also known as “Lunar Lake” in Berlin, just ahead of the IFA conference. Formally known as “Intel Core Ultra (Series 2)” is the successor to Intel’s Meteor Lake Core Ultra CPUs. The new platform promises incremental CPU performance improvements but significant performance per watt and battery life improvements, noticeably faster graphics performance, and a new NPU (neural processing unit) that meets Microsoft’s requirements for Copilot+ PCs for on-device generative AI capabilities. Intel’s 9 new Core Ultra variants will be available in PCs beginning on September 24, with a Microsoft Copilot+ software update scheduled for November.
Intel Sticks the Lunar Lake Landing as the AI PC Race Heats Up
Analyst Take: As promised at the Computex conference in Taipei this past June, Intel formerly introduced its new Lunar Lake chipsets – Core Ultra Series 2 – at IFA on September 3 in Berlin. This new x86 PC platform is a vital play for Intel in the PC segment and comes at a critical time, as the segment enters a new refresh supercycle driven by next-gen on-device AI capabilities. To best understand the significance of this announcement, however, let’s quickly review how we got here.
January 2024: The first AI PC evolutionary leap (with Meteor Lake Core Ultra)
Intel entered the AI PC market last year with its first generation of Core Ultra chips, effectively kicking off the AI PC race. While the Meteor Lake Core Ultra CPUs were solid and a great first step in the transition from traditional to AI-capable PCs, the platform failed to capture the essence of what AI PCs could deliver. The result: Lukewarm excitement from the market, and, in my view, a bit of a false start for both Intel and the AI PC category. In the months that followed, Qualcomm capitalized on the opportunity gap left open by Intel and introduced its new Arm-based Snapdragon X PC platform in partnership with Microsoft in June, ushering in a new era for the AI PC with the launch of the new Copilot+ PC category. The launch was a remarkably well orchestrated event, with every major PC OEM (Acer, Asus, Dell, HP, Lenovo, Microsoft, and Samsung) introducing their first Copilot+ PCs, all powered by Snapdragon X Elite and X Plus chipsets.
May 2024: The Copilot+ PC ecosystem launch gives Snapdragon X a head start
While Snapdragon X-powered Copilot+ PCs were effectively the first exciting manifestation of AI PCs (with 45+ TOPS on the NPU and impressive performance per watt specs), the sense I got from the launch event in Redmond was that the initial positioning focus was more a Windows and PC OEM play against Apple than a hard push for new AI PC experiences. One thing at a time. In fairness, AI use cases and apps are still in their very early stages, so it made sense to prioritize new performance benchmarks over on-device AI UX. Our own research into the impact of that launch on excitement for the AI PC category reveals that it was extremely successful. Pre-orders from retailers for Copilot+ PCs exceeded expectations, and ITDMs on the commercial side definitely leaned in and accelerated their pilot schedules for the platform.
Microsoft was careful during its Copilot+ launch event to remind the market that while Qualcomm’s Snapdragon X platform was the only chipset capable of delivering Copilot+ experiences, they were working as fast as they could to get AMD and Intel into the ecosystem, and hinted at an end-of-year schedule for that milestone. Less than two weeks later, both AMD and Intel took to the stage at the Computex conference to give that timeline more gas: AMD introduced its own AI PC chips, and joined PC OEM partners with well coordinated launches of their own which continued well into the summer. (ASUS and HP especially come to mind.) Intel, for its part, confirmed that its Lunar Lake Core Ultra chips would be ready in September and would ship by year-end – months ahead of what the market initially expected. The early release speaks to the urgency the company must have felt in recent months, thanks to pressure from Qualcomm and AMD, and the importance of getting this product out as fast as possible in light of the Windows 10 sundowning countdown. (Support for Windows 10 is scheduled to end next fall.)
There is a lesson here: If you are skeptical about AI, you don’t have to think about the impending Windows PC refresh cycle in terms of on-device AI being the primary driver. There are two other drivers that may actually be more critical in the short term: (1) performance improvements accelerated by advances in semiconductor architecture and NPU design (like battery life improvements and faster processing) and (2) the transition from Windows 10 to Windows 11.
September 2024: Did Intel’s Lunar Lake Core Ultra stick the landing?
Fast-forward to this week at IFA in Berlin, Intel delivered on that promise with a succinct but effective press conference introducing the new Lunar Lake Core Ultra platform, complete with PC OEM partner endorsements and product teases.
First impressions: Usually, new product announcements from Intel are straightforward, matter-of-fact affairs, mostly centered around spec and performance numbers that speak to Intel’s incumbency and leadership in the PC segment. Not this time. In Berlin this week, the energy in the oppressively warm press room felt different. Intel came out swinging, taking direct shots at AMD and Qualcomm, and dropping a very large hammer on current performance expectations for Copilot+ PCs: With the Lunar Lake Core Ultra platform, Intel promises to not only meet or exceed the 40+ TOPS on the NPU and minimum 16GB of RAM and 256GB of storage requirement for Copilot+ PCs, but also bring a powerful GPU to bear to deliver an additional 60+ TOPS, while delivering battery life that can compete against the Snapdragon X.
Intel’s rule of thirds for AI PC system design approach: Don’t sleep on the GPU
On paper, the Core Ultra 200v, as it is formally called, is a beast, but what caught my eye wasn’t the combination of 4 P-cores (Lion Cove architecture) and 4 E-cores (Skymont architecture). It’s the fully enabled Arc 140v GPU, based on Intel’s new Battlemage architecture: Expectations are that it will deliver 31% faster performance than Meteor Lake’s Arc GPU. Intel claims a 68% speed advantage over Qualcomm’s Adreno GPU in the Snapdragon X Elite platform and a more modest 16% advantage over AMD’s Radeon 890M in the Ryzen AI 9 HX 370 chip. We will of course have to validate these claims, but if they prove accurate, Intel will have regained the advantage in the AI PC race.
Several reasons for that design choice come to mind – all of them good. First, Intel probably wants to take an approach of thirds, in which it expects that PC workloads will essentially be split along a 1/3 CPU, 1/3 NPU, and 1/3 GPU ratio – a reasonable way of thinking of system design for AI PCs. Second, Intel also knows that not every application is designed for the NPU, and so it needed to bring a solid GPU into the mix. Third, if Lunar Lake Core Ultra is to be competitive against AMD, particularly in GPU-intensive applications, it might as well bring its a-game to conversation.
Circling back to some of the details I glossed over a moment ago: both the Lunar Lake Core Ultra 9 and 7 are basically the same chip, with 12MB of cache, 8 Xe cores for the integrated Arc 140V GPU, and 6 cores for the NPU, with the Core Ultra 7 clock speeds for the CPU, GPU, and NPU being slightly lower. Core Ultra 5 sports the same number of CPU cores but with 8MB of cache instead of 12, and its Arc 130V GPU only has 7 Xe cores while the NPU sports only 5 cores. Intel confirms that even with its lower specs, Core Ultra 5 can deliver the 40 TOPS needed to meet Copilot+ requirements.
Intel vs AMD vs Snapdragon: Scoping obvious differentiators across the Copilot+ platform ecosystem
My guess is that Intel’s cost/benefit equation is basically this: Running workloads on the GPU will be more thermally expensive than running them on an NPU, meaning a shorter battery life, but there is more value for Intel to have a well-rounded product that can do it all than to prioritize battery life over overall agility. And that perspective on system design (and purpose) may be one of the clearest differentiators yet between the Lunar Lake Core Ultra and Snapdragon X platforms, with the latter prioritizing performance per watt and battery life over Intel’s more heavy-handed do-it-all approach.
Where does that leave AMD? For the most part, AMD still delivers differentiated premium x86 performance at a lower price. A lot of commercial and consumer PC users prefer AMD, for a variety of reasons, particularly in Europe, and so AMD’s lane remains fairly clear for loyal PC users.
Lunar Lake Core Ultra also appears to be competitive on the battery life front, but with some caveats
Intel shared a few benchmarks that suggest that Lunar Lake Core Ultra PCs may be competitive with Snapdragon X in terms of battery life, and even in some narrow use cases, more efficient. Well played: Pushing back against Snapdragon’s impressive battery life performance is an important objective for Intel for a number of reasons, from UX to TCO.
Core Ultra 5 and 7 chipsets are designed for a minimum of 8 watts with default power set to 17 watts. The Core Ultra 9’s minimum is 17 watts and its default base power is set to 30 watts. All of the Lunar Lake Core Ultra max out at 37 watts. Even with those specs, I suspect that Snapdragon X PCs will still deliver noticeably better battery life than any x86 Windows PC, even with Lunar Lake Core Ultra’s impressive improvements. There are several reasons for that, but the most obvious is that GPUs are thermally expensive. And if Lunar Lake Core Ultra PCs end up running more workloads on the GPU, battery life will take a hit.
Having said that, Lunar Lake Core Ultra’s considerable architecture upgrades may deliver battery life improvements solid enough for most PC users looking for a well-rounded PC that will last a whole day without needing to be plugged in.
Is Intel’s emphasis on Lunar Lake Core Ultra’s gaming performance an oversell?
Intel spent a good deal of time focusing on gaming performance during its announcement. This is a useful competitive advantage narrative that will obviously resonate with even casual gamers. Bonus: Even if the average PC user (and enterprise buyer) doesn’t care about gaming, those system improvements will trickle down to deliver more premium experiences across a broad swath of applications, such as video editing and other media-centric workloads. It also serves as an additional point of differentiation against Snapdragon X, which has, at least until now, struggled a bit in the gaming department. The subtext here isn’t just “better for gaming” but rather “if better for gaming, perhaps better overall.”
Made by TSMC: Pros and cons
Intel outsourcing at least some of Lunar Lake Core Ultra’s manufacturing to TSMC feels a bit awkward, given Intel’s fab and packaging investments. I am sure that Intel would much rather enjoy different optics. But the more I think about it, the more I think this might actually be a plus for Lunar Lake, and here is why: One of the big unknowns relative to the new Core Ultra shipping to OEMs was whether or not Intel would be able to handle production at pace and at scale with minimal quality issues. I feel that working with TSMC on this addresses that concern, or at the very least mitigates the quality control challenges that Intel might have encountered along the way. Supply chain delays notwithstanding – TSMC is in very high demand lately – execution should be a lot smoother and with far fewer headaches than the alternative.
Yes but…
One last caveat to all of this: As good as it appears to be, the part is expensive to make. That cost challenge may create additional opportunities for both AMD and Qualcomm, whose Copilot+ AI PC chips offer comparable performance at attractive price points. (Qualcomm notably just launched an 8-core version of its Snapdragon X Plus to address the $600-$700 price tier, broadening the range already covered by the 10-core Snapdragon X Plus and the 12-core Snapdragon X Elite.) Something to keep an eye on as PC OEMs begin to roll out their new lines of Copilot+ PCs powered by Snapdragon, AMD, and Intel across multiple price tiers.
In summary…
If Intel’s job was to show that Lunar Lake isn’t just an also-in PC chipset limping in late to the Copilot+ PC party, it understood the assignment and delivered on that objective: the Core Ultra 200v makes a great first impression. It’s difficult not to be impressed by what Intel accomplished here, especially given the headwinds the company has had to overcome to get here in the past year. On paper at least, this product exceeds my expectations and delivers to the PC segment an extremely well rounded x86-based solution for AI PCs that may very well help Intel shift the competitive balance of power back in its favor. I look forward to validating Intel’s benchmarks and seeing how Lunar Lake’s 9 Core Ultra variants perform in the real world.
Disclosure: The Futurum Group is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of The Futurum Group as a whole.
Other insights from The Futurum Group:
Intel and IBM Show Joint AI Moxie in the Cloud
Computex 2024: Intel Catalyzes AI Everywhere with Key AI Portfolio Innovations
Intel Vision 2024: Intel Unleashes Gaudi 3 Led Enterprise AI Strategy
Author Information
Olivier Blanchard has extensive experience managing product innovation, technology adoption, digital integration, and change management for industry leaders in the B2B, B2C, B2G sectors, and the IT channel. His passion is helping decision-makers and their organizations understand the many risks and opportunities of technology-driven disruption, and leverage innovation to build stronger, better, more competitive companies. Read Full Bio.
