Home
Menu
Subscribe
Banner reading XPU dot pub
photo of Intel Meteor Lake chip with no lid and die layout emphasized

Is Intel Meteor Lake a New Centrino Moment?


One can imagine the sighs let out in Santa Clara when Intel’s long-delayed Core Ultra (Meteor Lake) PC processors were officially launched in December 2023. CEO Pat Gelsinger refers to Meteor’s launch as a Centrino moment, harkening back two decades to a time when Intel’s PC-processor strategy was getting back on track and a new technology called Wi-Fi was capturing users’ attention. In this era it’s been process technology, not microarchitecture, that’s been off track and the new technology is AI acceleration. The comparison is apt but overblown as Meteor’s CPUs are more power efficient but otherwise nearly identical to the previous gen’s cores and AI adoption not falling short of 802.11’s two decades ago.

Intel Meteor Lake Notables

  • AI—Meteor is Intel’s first PC processor with significant AI acceleration, and its release coincides with Microsoft’s pushing Copilot and adding a special keyboard key. By contrast with the Centrino era’s new technology, Wi-Fi, which had burgeoning grass-roots adoption before Intel’s push, AI is of uncertain value to ordinary PC users. The neural processing unit (NPU) accelerating AI also needs software to use it, and developers need a large installed base to target. It will be another year before Intel integrates NPUs across all new PC processors, and AI add-in boards are unlikely to be warmly received by OEMs and consumers in the meantime. Thus, the installed base will take a few more years to build.
  • Manufacturing—After years of struggling to advance its process technology, Intel is reaching an important milestone by finally employing EUV lithography as Samsung and TSMC have done for the past few years. Meteor is Intel’s first chip built using the Intel 4 EUV-based process.
  • Chip packaging—Concerns about Intel 4 likely drove Intel to employ chiplets in Meteor, using the new process only for CPUs and TSMC-made dice for other functions. It’s an expensive approach that doesn’t offer the composability of AMD’s chiplet strategy, but it minimizes the risk associated with a monolithic Intel 4 design. Intel must revisit its chiplet strategy to reap economic benefits from the approach.
  • CPU architecture—Whereas Centrino marked a shift from GHz as a marketing metric to performance and performance per watt, accomplished by rebooting Intel’s CPU strategy, Meteor integrates similar CPUs as Intel’s recent PC generations. The low-power efficiency cores are newish but of modest significance, potentially reducing chip power when workloads are light.
  • Chip configuration—Intel is still pursuing a small-medium-large product strategy instead of implementing more distinct configurations for different user types (video creators, photo and art creators, office drones, surfers, gamers, etc.). Moreover, even after a few generations of selling processors combining P cores and E cores, it’s unclear what value the E cores deliver, other than improving scores on multithread benchmarks. Cell phones, by contrast, have proven the value (power efficiency) of heterogeneous cores. In a mature market, segmentation and product differentiation can increase profits.

Competition

Overall, Meteor reinforces the competitive landscape. Had its schedule slipped further, Intel would’ve lost credibility (and share) as a supplier. On the other hand, power, CPU performance, and graphics performance are competitive and don’t catapult Intel ahead of AMD. Intel’s NPU-enhanced PC processor is coming to market a few quarters after AMD’s but in a timeline aligned with Microsoft, thus effectively yielding nothing in timeliness for the hot new feature. AMD, however, has bumped up the AI performance of its competing chip, giving it an edge. When it comes to AI and other attributes, Apple benefits from its vertical integration, synchronizing hardware and software releases to deliver functionality improvements. As for Qualcomm and the other Arm for Windows suppliers in the wings, they have a lot of work ahead of them before they’re a factor.

Customers

Intel had a $250 million promotion budget for Centrino, which went to things like logo stickers for cafes offering Wi-Fi to logo stickers for laptops. Such programs curry favor with OEMs, but Intel isn’t backing Meteor and the AI PC in the same way. Nonetheless, OEMs are launching Meteor-based systems, suggesting the chip is well received, likely owing to its improved graphics and power efficiency. These same factors should appeal to end users, with an NPU nice to have for future-proofing the system.

Bottom Line

Meteor Lake does what Intel needed it to do: prove Intel 4 and deliver better power efficiency. It may not be as economical as a monolithic design, but it’s a competitive chip that will help Intel maintain its PC-processor lead. Intel’s mass production of its Foveros chiplet technology and Intel 4 process node, the addition of the AI processing unit, and the new low-power E cores are all advancements and will be improved upon in future generations.


Posted

in

by


error: Unable to select