Loading...
Loading...
Intel is pushing a China-focused strategy to reduce costs and accelerate laptop production by importing smartphone manufacturing practices into its upcoming Wildcat Lake platform through Project Firefly. The initiative includes a reference-design thin-and-clean “Clean-D” chassis—first adopted by Lenovo’s Lai Ku Air 14—to enable slimmer, lower-cost notebooks built with standardized interfaces and barebone efficiencies. Concurrently, Intel’s mobile CPU roadmap advances with Razor Lake AX (RZL-AX), which reportedly offers much larger integrated Xe3P GPUs—16- and 32-core variants—indicating a push to boost on-chip graphics and better compete with AMD. Together, these moves reflect Intel’s dual focus on supply-chain localization and higher-end integrated GPU performance for future laptops.
Intel's Project Firefly and larger Xe GPU designs affect laptop OEM cost structures and system architectures, influencing supply-chain and design choices for device makers. Tech professionals must track these shifts for platform planning, performance expectations, and component sourcing decisions.
Dossier last updated: 2026-05-20 01:15:47
Leaked PCB images, shared by @结城安穗-YuuKi_AnS and reported by IT Home, reveal Intel’s data-center AI inference GPU 'Crescent Island.' The card uses a PCIe Gen5+ connector, a large central GPU pad, and 20 LPDDR5X memory pads—implying 160 GB total if each pad holds an 8 GB package—plus 12V 2×6 power connectors. BenchLife.info notes the board appears to have a 15-phase core and 3-phase memory VRM, a BMC chip location, and a side USB-C likely for debugging. Intel says Crescent Island targets air-cooled servers, uses the Xe3P microarchitecture, supports many numeric formats, and is expected to sample to customers in H2 2026, signaling a competitor in AI inference accelerators.
Leaked photos of Intel’s Crescent Island PCB reveal a large Xe3P GPU module, a 16-pin power connector, and onboard 160 GB of LPDDR5X memory—signaling Intel may bypass scarce HBM supply for this accelerator design. The board appears to integrate multiple memory packages and a beefy power delivery system, implying a high-performance compute target likely aimed at AI/data-center workloads. If accurate, using LPDDR5X instead of HBM could lower costs and ease supply-chain constraints but may trade off memory bandwidth and efficiency compared with HBM-based accelerators. The leak matters because it hints at Intel’s architectural choices for upcoming accelerators, competitive positioning versus NVIDIA/AMD, and how chipmakers adapt to memory shortages in the AI hardware market.
Intel announced Project Firefly at its third-generation Core processor briefing, aiming to integrate China’s mature mobile supply chain into the price-sensitive Wildcat Lake platform to cut costs and scale production with standardized interfaces. Intel China VP Gao Song showcased a reference-design laptop with an orange “intel color” lid and a Clean-D bottom cover featuring no large ventilation openings. Lenovo’s Lai Ku Air 14 will be the first mass-produced product adopting the Firefly reference design and the thin-body Clean-D approach. The plan targets barebone-level efficiency gains by leveraging smartphone manufacturing expertise for slimmer, lower-cost Wildcat Lake notebooks, signaling deeper localization of Intel’s laptop ecosystem in China.
Intel's upcoming mobile processor family Razor Lake AX (RZL-AX) will ship with two integrated GPU configurations: a 16 Xe3P-core variant and a 32 Xe3P-core variant, according to leakers @jaykihn0 and @Haze2K1. The reported graphics module area is about 162.84 mm², which aligns with the larger 32-Xe configuration when compared to Panther Lake's 12Xe ~54 mm² GPU die. Razor Lake succeeds Nova Lake in Intel's mobile roadmap and is expected to compete with AMD's Medusa Halo in integrated GPU performance. The details suggest Intel is scaling up Xe3P GPU complexity for stronger integrated graphics in next-generation mobile chips.