Logo

Marvell to acquire XConn, boosting AI data centre connectivity

Marvell Technology acquires XConn Technologies to strengthen its PCIe and CXL switching portfolio and advance AI infrastructure solutions.

Marvell Technology, a semiconductor company focused on data centre connectivity and infrastructure, has announced its agreement to acquire XConn Technologies, a provider of PCIe and CXL switching silicon. This acquisition aims to expand Marvell's existing switching portfolio by welcoming aboard XConn's PCIe and CXL products, and enhances Marvell’s Ultra Accelerator Link (UALink) scale-up switch team, by infusing it with XConn's engineering team.

As artificial intelligence workloads grow, there’s a noticeable shift in data centre system design, advancing from isolated rack solutions to much larger, interconnected multi-rack configurations. These emerging platforms often use an ultra-low latency and high-bandwidth scale-up fabric, such as UALink, to connect myriad XPUs and facilitate resource sharing.

UALink is an open industry standard designed with scale-up connectivity in mind, intended to enable high-speed communications and allow multiple accelerators to operate together as a single system. Building on PCIe ecosystem innovation, UALink employs high-speed I/O techniques designed to meet the bandwidth, latency, and reach requirements of contemporary accelerated infrastructure.

The union of Marvell and XConn amalgamates two teams poised to address the expanding UALink switching landscape, providing support to the growing roster of clients keen on developing future-focused AI platforms with Marvell.

In a statement, Matt Murphy, chairman and CEO of Marvell, highlighted the strategic value of the merger, stating that the combined platform will enhance Marvell’s connectivity strategy for future AI and cloud data centres. The inclusion of XConn’s PCIe and CXL switch products expands Marvell’s portfolio.

This acquisition complements Marvell’s ongoing initiatives, including its prospective acquisition of Celestial AI, and aims to support performance, flexibility, and architectural options for AI systems.

Scality's research reveals challenges in scaling AI infrastructure, highlighting storage's critical...
Schneider Electric collaborates with NVIDIA and AVEVA to develop new reference designs and AI...
Pure DC announces senior leadership appointments to support the development of hyperscale AI...
Jacobs introduces a digital twin solution for AI data centres, aiming to enhance planning,...
Corning partners with US Conec to enhance AI network connectivity through PRIZM TMT technology,...
Dell expands its security measures to help protect organisations against quantum computing and AI...
F5 and NVIDIA join forces to enhance AI infrastructures by enhancing token throughput, reducing...
Zadara's AI cloud platform aligns with NVIDIA guidelines, aiming to enhance multi-tenant AI...