A ‘Big Battlemage’ iGPU: 32 Xe3P Graphics Cores on Tap
Razor Lake-AX is shaping up to be Intel’s most ambitious integrated graphics platform yet. Leaks suggest the chip will feature up to 32 Xe3P graphics cores, effectively matching the scale of Intel’s largest Battlemage-class GPU dies. That is a dramatic jump from Panther Lake’s 12 Xe3 cores, representing roughly a 166% increase in core count for the top configuration. Early reports point to two Razor Lake-AX GPU options: 16 Xe3P cores for performance-focused thin-and-lights, and a full 32-core variant positioned for gaming laptops and compact workstations. For context, Intel’s workstation-oriented Arc Pro B70 already delivers performance comparable to an RTX 5060 Ti while also offering 32 Xe2 cores. Moving to a similar core count with a newer Xe3P architecture inside a processor package sets the stage for integrated graphics performance that could rival mid-range discrete cards in everyday gaming and content creation workloads.
On-Package Memory Returns to Supercharge Integrated Graphics Performance
Equally important to the Razor Lake-AX GPU is its memory architecture. Intel is reportedly reviving the on-package memory strategy first seen in Lunar Lake, after skipping it for Panther Lake. By placing LPDDR-class memory directly on the CPU package, Razor Lake-AX shortens the electrical path to RAM, reducing latency and simplifying signal integrity at very high data rates. This design is particularly beneficial for a wide memory bus feeding a large integrated GPU, where conventional motherboard traces can become a bottleneck. While exact specifications remain unconfirmed, leaks suggest high-speed LPDDR5X or next-generation LPDDR6, and even reference Intel’s in-house Z-Angle Memory (ZAM) as a potential alternative. The result should be significantly higher effective bandwidth for the Xe3P graphics cores, translating into smoother frame rates, faster texture streaming, and more consistent performance when compared to traditional integrated graphics reliant on external system memory.
An RTX 5060 Ti Rival Without a Discrete Card?
Pairing a 32-core Xe3P GPU with high-bandwidth on-package memory raises an intriguing possibility: an integrated GPU that behaves like a mid-range desktop graphics card. Today, Intel’s Arc Pro B70 with 32 Xe2 cores is already in RTX 5060 Ti territory, and Razor Lake-AX is expected to surpass this with architectural and frequency gains. In a thin-and-light laptop, that could mean 1080p or even 1440p gaming with settings far higher than typical integrated solutions can handle, reducing or eliminating the need for an entry-level discrete GPU. Power and thermals will remain key constraints, but the tight coupling of compute and memory should help maintain efficiency under load. If Intel can deliver consistent, sustained performance, Razor Lake-AX may redefine integrated graphics performance expectations and blur the line between iGPUs and mainstream discrete cards in everyday gaming scenarios.
The Trade-Off: Fixed Memory Capacity vs Efficient, Compact Designs
On-package memory is not without its drawbacks. Because RAM is soldered alongside the processor, users lose the ability to upgrade or expand memory after purchase, limiting long-term flexibility. System builders also have less freedom to mix and match memory capacities or brands. However, this trade-off brings significant benefits for devices that prioritize thin form factors, long battery life, and strong integrated graphics performance. A unified package simplifies motherboard design, improves thermals by localizing hot components, and makes it easier to sustain wide memory buses at high speeds. For compact gaming PCs, handhelds, and premium laptops, these advantages often outweigh the downside of fixed memory. Razor Lake-AX appears tailored to these segments, where a balanced, highly integrated design can deliver a better overall experience than a more modular system constrained by power, space, and cooling limits.
Intel’s New Position in the GPU Landscape
Strategically, Razor Lake-AX signals Intel’s intent to compete head-on with AMD’s Medusa Halo-class APUs, Apple-style single-package designs, and Nvidia’s entry-level discrete GPUs. By scaling its integrated Xe3P graphics cores to ‘Big Battlemage’ proportions and feeding them with on-package memory, Intel is targeting high-end thin-and-light laptops, compact desktops, and mobile workstations where discrete GPUs add bulk, cost, and power draw. At the same time, Razor Lake-AX helps differentiate Intel’s roadmap from Panther Lake and Nova Lake, carving out an ‘AX’ tier focused on maximum integrated graphics performance and bandwidth. Looking ahead, this approach could pressure Nvidia’s lower-end desktop offerings and push AMD to respond with even more capable integrated solutions. If leaks hold true, Razor Lake-AX may not just be a faster iGPU—it could mark a turning point where integrated graphics become a default choice for serious gaming on portable and small-form-factor systems.