Intel leans on LPDDR5X to dodge global HBM crisis, leaked Crescent Island AI GPU pics reveal massive Xe3P core — chip sidesteps HBM shortage with 160GB of cheaper memory
Images of the PCB Intel will be using with Crescent Island has been published on X, showing a single-GPU design and pads for the graphics card's 20 LPDDR5X modules.
The increasing demand for AI compute and the ongoing HBM supply crunch are forcing manufacturers to innovate and seek alternative memory solutions.
This move by Intel suggests a significant strategic pivot in high-performance computing memory architecture, potentially bypassing a critical supply bottleneck and lowering costs for AI accelerators.
Intel's decision to use LPDDR5X for its AI GPU signals a potential diversification in memory architectures for AI, offering an alternative to HBM and impacting market dynamics for both memory types.
- · Intel
- · LPDDR5X manufacturers
- · AI compute cost-conscious buyers
- · HBM manufacturers (short-term)
- · Competitors reliant solely on HBM
Intel gains a competitive advantage in the AI GPU market by mitigating HBM supply risks and potentially offering more cost-effective solutions.
Increased adoption of LPDDR5X in high-performance computing could drive further investment and innovation in this memory technology, reducing its power consumption and improving bandwidth.
This could lead to a bifurcation of the AI accelerator market, with some players optimizing for raw HBM bandwidth and others for LPDDR5X cost-effectiveness and volume, ultimately impacting the overall cost of AI infrastructure.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at Tom's Hardware