Intel has started putting its new D-Series data center GPUs into customers’ hands for testing. The chips mark the company’s latest attempt to win a meaningful slice of the AI hardware market.
The pitch is not about building the biggest, fastest training monster on the block. Intel is aiming at AI inference, the part of the job where trained models actually answer questions.
What Intel is shipping
The D-Series was previously known by its codename, Crescent Island. It is built on Intel’s Xe3P architecture.
The flagship model is the D270. Intel is positioning it for AI inference and agentic workloads, meaning AI systems that take multi-step actions instead of answering a single prompt and stopping.
On paper, the D270 delivers a peak compute of approximately 1.3 PFLOPs at FP8 precision. FP8 is a lower-precision number format that trades a bit of numerical detail for much faster math.
That performance comes with a thermal design power of 350W.
Under the hood, the chip packs 32 Xe3P cores and 256 XMX engines. The XMX engines are Intel’s dedicated matrix-math units, which handle the heavy multiplication that neural networks run on. A unified 32 MB L2 cache ties the design together.
The memory bet
The most interesting choice is the memory. Instead of HBM, the high-bandwidth memory used in many competing accelerators, the D270 uses LPDDR5X.
Intel’s reference configuration ships with 160 GB of LPDDR5X. Partners can expand that to 480 GB. Peak memory bandwidth is estimated at around 1.5 TB/s.
Intel says it is targeting long-context and agentic AI applications with a value-oriented approach. Intel’s argument centers on high memory capacity relative to power draw and cost, rather than chasing peak bandwidth numbers.
Built to slot into existing racks
The D270 uses an air-cooled design and a PCIe Gen5 form factor. PCIe cards plug into standard servers, so data center operators can add them without rebuilding their infrastructure around exotic cooling or proprietary boards.
Crescent Island emphasizes scalability and integration over sheer training performance.
Timeline and the Xeon connection
Customer sampling for the D270 is targeted for the second half of 2026. Intel planned previews at Computex and Hot Chips 2026 ahead of broader availability.
The D-Series is also part of a bigger strategic play. Intel wants to pair its GPUs with its Xeon processors to build more complete AI inference systems. Intel’s OEM partners include ASUS, Dell, and HPE.
What this means
Intel is framing the D270 as a cost-effective option for AI workloads that competes with offerings from NVIDIA and AMD.
The risks are real. LPDDR5X offers capacity, but its bandwidth sits well below what HBM-based designs can deliver, and some workloads are bandwidth-hungry.
The thing to watch is customer feedback from the sampling phase in the second half of 2026. Early reactions should show whether enterprises see the D270 as a genuine alternative or a niche option.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.