Intel Crescent Island leaked PCB reveals massive Xe3P GPU with 160GB LPDDR5X to dodge HBM crisis

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

Intel is going with one massive GPU for its next-generation Xe3P AI GPU, and LPDDR5X memory rather than HBM. Pictures of the PCB of Intel's upcoming Crescent Island data center GPU, announced last year, have surfaced. The images, which come by way of @yuuki_ans on X, show the front and back of the Crescent Island PCB and reveal the layout of the board's components, including the GPU, VRAM, and power delivery system. The PCB shots confirm that Intel has opted to go with a single-GPU setup for Crescent Island, rather than a dual-GPU setup. A vast majority of the PCB's real estate is consumed by a massive GPU socket in the middle, taking up the width of almost the entire PCIe x16 slot below. There are pads for twelve LPDDR5X modules to the top, left, and right of the GPU socket and an additional eight on the back of the PCB, totaling 20. This confirms the GPU will use 32GB (8 GB) modules, the highest capacity modules out there for LPDDR5X currently on the market. The power delivery system consists of a single 16-pin 12V-2x6 power connector on the right and 19 power phases. Crescent Island is the codename for Intel's next-generation data center GPU powered by its Xe3P GPU architecture. Its most noticeable trait is the inclusion of LPDDR5X memory over HBM, which should make Crescent Island the world's first AI GPU to use LPDDR5X. This will give Crescent Island significantly inferior memory bandwidth, but undoubtedly, Intel is opting to use slower memory to save on production costs, thanks to the memory shortage. (HBM memory sits at the center of the outgoing memory shortage crisis, as it's the hardest to secure for GPU manufacturers .) Assuming the GPU comes with a 640-bit memory interface and uses 10.7Gbps memory, maximum memory bandwidth will be well under 1TB/s -- a far cry from Nvidia's older H200 GPUs, which have nearly 5TB/s of memory bandwidth. Memory bandwidth is one of the most important aspects of an AI GPU's performance and how quickly it can execute machine learning workloads, so it will be interesting to see how much LPDDR5X impacts Crescent Island's performance compared to HBM-equipped cards. Intel's new AI-GPU will be targeted at air-cooled servers and will be a competitor to existing Nvidia and AMD cards in the same target demographic, such as the recently released AMD MI350P with 144GB of HBM3E and the Nvidia H200 NVL with 141GB of HBM3. Intel plans to start sampling Crescent Island to its customers in the second half of 2026. Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

Intel Crescent Island PCB leak gives us our first look at the Xe3P GPU and its 160GB LPDDR5X memory design

Back in February 2026, Intel released an introductory video indicating that its next-generation Xe3P architecture would follow the Xe2 architecture. Later, it was reported that Xe3P would not appear in consumer Celestial gaming graphics cards, and Intel canceled the product entirely. Instead, Intel plans to use the architecture in Crescent Island. Crescent Island is Intel's next data-center GPU, targeting AI inference rather than a consumer Arc graphics card. Intel confirmed the product in October, featuring the Xe3P graphics architecture, 160GB of LPDDR5X memory, and targeting air-cooled data centers and workstations optimized for AI inference workloads. Now, the upcoming Crescent Island accelerator has appeared in a first PCB leak from YuuKi_AnS, spotted by Wccftech. The leaked PCB gives us a look at the large Xe3P GPU die and its LPDDR5X memory configuration. The GPU chip itself occupies a significant portion of the PCB and is notably larger than Intel's current flagship, the Xe2-based BMG-G31. The leaked PCB reportedly features 20 LPDDR5X memory sites - 12 on the front and 8 on the back - supporting a total of 160GB of LPDDR5X memory. Intel chose LPDDR5X over HBM as a cost-effective alternative that avoids the power-supply strain HBM would impose on the system. NVIDIA and AMD are already shipping data center AI hardware with high-end HBM memory like HBM3E, and neither is shy about teasing HBM4 for upcoming chips like Rubin and MI400. But HBM is getting harder to come by thanks to surging demand, and prices are climbing. Leveraging LPDDR5X memory could give Intel a significant cost advantage without sacrificing the performance needed for inference workloads. The new architecture is also built to handle a wide mix of data types, which should be a win for tokens-as-a-service providers and anyone running inference at scale. Other than that, the pictures show a high-end PCB design with 18 VRM positions, 13 of which appear populated. A USB-C port is visible on the side, presumably for testing, and a 12V-2x6 power connector is located at the rear of the PCB. The PCB leak does not confirm final clocks, power limits, or performance figures. Intel has not announced pricing or launch timing, other than confirming that customer sampling is planned for the second half of 2026.

Intel's Crescent Island PCB Leaks, Showing a Massive Xe3P GPU, 16-Pin Connector, 160GB LPDDR5X as Intel Sidesteps the HBM Shortage

Intel's next-generation Crescent Island PCIe graphics card has been pictured in the first PCB leak, giving us a look at the large Xe3P GPU & support for LPDDR5X memory. The first pictures of Intel's Crescent Island PCIe accelerator's PCB have been leaked by YuuKi_AnS. Crescent Island is the latest and upcoming Inference accelerator designed for AI workflows, offering competitive value. The leaked PCB gives us an idea of what the graphics card will look like. Starting first with the GPU itself, which looks massive and much larger than Intel's current flagship, the Xe2-based BMG-G31. The Crescent Island GPU is based on the Xe3P architecture, the one following the current Xe3 architecture. The BGA pad shows the massive scope of the chip. Surrounding the GPU are 12 sites for the memory, which are much smaller than standard GDDR modules. This is due to the fact that Crescent Island features LPDDR5X memory & is designed to be a cost-effective option versus the more expensive HBM standards. There are 12 sites on the front & 8 sites on the back for a total of 20 LPDDR5X sites, totalling up to 160 GB capacity. That's 8 GB per module. Other than that, you are looking at a high-end PCB design with 13 VRMs, which seem to be the ones that will be populated, while the actual total count is 18. Power is provided through a single 16-pin connector featured on the back of the board. YuuKi provides additional details, such as a side USB Type-C port, which seems to be for testing purposes. Currently, the Intel Crescent Island GPU is still a few months away from launch, but based on this PCB looks much closer to the final version. The Intel Crescent Island GPU is based on the brand-new Xe3P architecture, which is the same graphics architecture that was teased by the company last week during its Panther Lake and Xe3 deep dives. The new architecture will be a further upgrade over the Xe3 architecture, and for clients, the architecture will be featured on a next-gen Arc family, the Arc C-Series. But Xe3P is going to be even more scalable, from client iGPUs to data center AI GPUs. The new data center GPU code-named Crescent Island is being designed to be power and cost-optimized for air-cooled enterprise servers and to incorporate large amounts of memory capacity and bandwidth, optimized for inference workflows. Key features include: Intel Crescent Island will be both power- and cost-optimized. It will be targeted at air-cooled data center solutions and will be aimed at AI inference workloads. According to Intel, the Xe3P graphics architecture used for Crescent Island will be optimized for performance per watt. The card itself will feature a massive 160 GB memory capacity based on the LPDDR5X standard. Interestingly, Intel is going with LP5X. Competitors such as NVIDIA and AMD are offering their data center AI solutions with top-grade HBM memory, such as HBM3E, and already talking about HBM4 for next-gen parts such as Rubin and MI400. But at the same time, sourcing HBM has become difficult due to increased demand, and that has also led to higher prices. Leveraging LPDDR5X memory can give Intel a big edge in the cost/performance segment. Furthermore, the architecture will support a broad range of data types that are ideal for "Tokens-as-a-service" providers and inference use cases. Intel is already evaluating its open and unified software stack for heterogeneous AI systems with its existing Arc Pro B-series lineup, so future iterations will be able to access these optimizations early on. Intel is currently targeting customer sampling for its Crescent Island GPU for the 2H of 2026, so we'll definitely learn more about the GPU in the coming months.