Intel CEO Pat Gelsinger captured the attention of technology enthusiasts at Computex 2024 in Taiwan, demonstrating the strength of Intel’s resources in the emerging era of the AI PC. Embracing a narrative of continuous advancement, he declared that “everything that has been done, will be outdone,” showcasing the particularly strong design of Intel’s Lunar Lake, which features a groundbreaking architecture including new IP blocks for CPU, GPU, and NPU.
Intel stood out at the event by highlighting its readiness across the entire AI continuum, boasting a comprehensive suite of services and products that range from semiconductor fabrication to data centers and edge computing, with particular focus on the AI PC space.
Gelsinger set straight the record on performance, disputing claims that competitor technologies outpaced the X86 architecture. According to his speech, Intel’s Lunar Lake provides not only top-notch performance but also achieves remarkable power efficiency, claiming a 40% reduction in SoC power consumption compared to its predecessor, Meteor Lake.
Intel’s pride in its cutting-edge designs was palpable, as Gelsinger praised the Lunar Lake SoC for its enhancements, citing a 50% increase in graphics performance, a quadrupling of NPU AI compute, and 3.5 times the GPU AI compute over previous generations. This next-gen technology is poised to power an extensive lineup of AI PCs, with shipments expected to begin in the third quarter of 2024.
The role of collaborator TSMC in enabling these innovations was acknowledged, emphasizing the partnership’s effort not only in Lunar Lake but also in setting standards like the Universal Chiplet Interconnect Express (UCIe), marking a significant milestone in industry collaboration.
Gelsinger boasted about the optimization of over 500 AI models for Core Ultra, which will operate on these AI PCs. Furthermore, when comparing Intel’s Gaudi 2 & 3 AI kits to competitors’, Gelsinger hinted at substantial cost advantages and superior performance, such as Gaudi 3’s ability to train models faster and provide double the inferencing speed compared to rival Nvidia’s offerings.
Intel’s new-generation Xeon processors also made an appearance, illustrating their speed and efficiency for data center applications, a testament to Intel’s progress and continued innovation in this space.
Despite facing fierce competition from AMD, ARM, and Qualcomm, who are each expanding their market reach into the PC domain with their respective technologies, Intel shifted focus to the opportunities lying ahead. Gelsinger remained confident, citing over 8 million AI PCs shipped to date and projecting a surge to 40 million units by the end of 2024.
To further position itself as a leader and innovator, Intel highlighted its commitment to open innovation, contrasting with Nvidia’s closed ecosystem approach. The company is an active participant in consortia like the Ultra Ethernet Consortium (UEC) and UA Link alongside other industry giants like AMD.
Gelsinger also touched on Intel’s ambitious process roadmap, revealing the next-generation Panther Lake architecture would be manufactured using the 18A process, signifying a return to process leadership for the company.
Culminating his presentation, Gelsinger teased a major celebration for Intel’s 40th anniversary in Taiwan at Computex 2025, where Panther Lake is expected to take center stage, underscoring Intel’s enduring significance in the region and its forward-looking vision for the future of computing.
