Data center interconnects boosted as CXL Consortium releases CXL 4.0 specification

The CXL Consortium has released the Compute Express Link (CXL) 4.0 specification, increasing interconnect bandwidth to address scaling and performance demands in data centers and emerging heterogeneous computing environments. The new CXL 4.0 specification, now publicly available, doubles the bandwidth from 64 giga-transfers per second (GT/s) to 128 GT/s without adding latency. The Consortium reports this release as advancing memory and accelerator connectivity in data center servers, enabling rapid data movement between CXL-attached components.

According to the CXL Consortium, CXL 4.0 maintains protocol features from CXL 3.x, including the 256-byte flit format, and introduces native x2 width to support higher fan-out in platforms. The update enables up to four retimers per channel to support greater reach and implements bundled port capabilities. This allows device ports between hosts and CXL-based accelerators (Type 1 and Type 2 devices) to be combined for increased connection bandwidth.

The new specification also includes memory Reliability, Availability, and Serviceability (RAS) enhancements, with the goal of improving error visibility and maintenance efficiency. CXL 4.0 is fully backward compatible with CXL versions 3.x, 2.0, 1.1, and 1.0. These features collectively target high-performance data center computing, storage, and memory devices as described by the Consortium.

CXL Consortium members will demonstrate CXL 4.0 and related technologies at Supercomputing 2025 (SC25) from November 16 to 21, 2025, in St. Louis. Technical sessions, including discussions on CXL memory pooling and sharing for artificial intelligence (AI) and high-performance computing (HPC) workloads, are planned.

“The release of the CXL 4.0 specification sets a new milestone for advancing coherent memory connectivity, doubling the bandwidth over the previous generation with powerful new features,” said Derek Rohde, CXL Consortium President and Treasurer, and Principal Engineer at NVIDIA. “This release reflects our member companies’ commitment to driving open standards that empower data center innovation, support compliance and interoperability, and enable the entire industry to scale for future usage models.”

For detailed specification access and technical deep dives, the CXL Consortium leadership and technical task force are hosting a public webinar on December 4, 2025.

Source: CXL Consortium

Get Data Center Engineering News In Your Inbox:

Popular Posts:

1765906506220
Tritium launches 800 VDC bidirectional inverter for data centers and renewable energy sites
Screenshot
Five AI data centers to reach 1 GW power capacity in 2026, new analysis shows
Water_From_Air_Data_Center_Rendering_Nov_2025
iMasons selects Water From Air for data center waste heat-to-water innovation
68e79d30a17eea847251fae6_img-home-product-liquidjet-main
Frore Systems updates LiquidJet direct-to-chip coldplate for 1,950 W NVIDIA Rubin data center GPUs
River Bend Rendering
Hut 8 announces $7 billion, 245 MW hyperscale data center IT lease with Fluidstack backed by Google

Share Your Data Center Engineering News

Do you have a new product announcement, webinar, whitepaper, or article topic? 

Get Data Center Engineering News In Your Inbox: