Marvell Technology, Inc. develops data infrastructure semiconductor solutions for systems that move, store, process and secure data. The company is a fabless chip designer serving data center, carrier, enterprise and consumer end markets with processors, optical and copper transceivers, switches and storage controllers.
MRVL news regularly covers product launches and portfolio expansion across AI data-center connectivity, optical DSPs, silicon photonics, SerDes, PCIe and CXL switching, near-memory acceleration, telemetry and custom silicon. Company updates also include financial-results conference calls, quarterly dividend declarations, acquisitions that add optical or interconnect technology, and ecosystem demonstrations for cloud and carrier architectures.
Marvell Technology (MRVL) announced two additions to its AI connectivity portfolio, with both devices now sampling to customers. The Alaska E 3.2T retimer, which restores signal quality, supports Ethernet, UALink and ESUN through 16 lanes of 224G connectivity for on-board and backplane applications.
The second-generation Alaska A 1.6T DSP, a digital signal processor for active electrical cables, moves to 3nm technology. It supports copper connections exceeding six meters, doubling the first-generation product's reach. Marvell designed the devices to address bandwidth, reach and power-efficiency requirements in AI infrastructure.
Marvell Technology (MRVL), Microsoft and Oracle have established an Open Compute Project working group to standardize integrated hardware security for data centers.
The group operates under OCP’s Security Track and will define open specifications for a hardware security module (HSM), a component that protects cryptographic keys, integrated directly into servers. The effort targets the NIST FIPS 140-3 Level 3 security standard for cloud and hyperscale infrastructure. An initial technical specification draft is planned for the OCP Global Summit in October 2026.
Marvell Technology (MRVL) unveiled its second-generation Structera X 3406 and 3506 memory expanders, designed for AI inference infrastructure.
The controllers use CXL 3.2, a connection technology for sharing memory resources, and PCIe Gen 6 connectivity. Their six memory channels represent 50% more channels than the previous generation. The 3406 supports up to 9TB of DDR4 memory, while the 3506 supports up to 6TB of DDR5. Both support memory sharing across up to four hosts.
Marvell expects the controllers to sample in Q1 2027. They extend the architecture of the Structera X 2404 and 2504, which are already shipping to hyperscalers.
Marvell Technology (MRVL) unveiled its second-generation Structera X 3406 and 3506 memory expansion controllers for artificial intelligence infrastructure. The controllers use six memory channels, 50% more than the previous generation, with CXL 3.2 and PCIe Gen 6 connectivity. Supported memory capacity reaches up to 9TB of DDR4 for the 3406 or 6TB of DDR5 for the 3506.
Marvell expects samples in Q1 2027; its first-generation Structera X 2404 and 2504 controllers are already shipping to hyperscalers. The new controllers support memory sharing among up to four hosts. The updated announcement also adds a statement from Meta about its work with Marvell on CXL memory solutions.
Marvell Technology (MRVL) will showcase its AI data center infrastructure portfolio at OCP Global Summit 2026 from October 12 to 15. The event takes place at the San Jose McEnery Convention Center in San Jose, California.
Demonstrations at Booth B31 will cover switching, memory expansion and acceleration, optical and copper connectivity, storage controllers and hardware security. Marvell executives and engineers will present 14 sessions on open connectivity for AI infrastructure, including Cary Ussery's executive session, “Spearheading a New Security Standard for Data Centers.”
Marvell Technology (MRVL) declared a quarterly dividend of $0.06 per share, payable October 29, 2026.
The dividend covers common stock and preferred stock on an as-converted-to-common-stock basis. Stockholders of record as of October 9, 2026, will receive the payment.
Marvell (MRVL) will showcase industry-first 2nm optical interconnect technology demonstrations for AI data center infrastructure at ECOC 2026 in Málaga, Spain, from September 20–24, 2026.
The company plans 38 demos, including first 2nm 400G/lane optical PAM4, live 2nm 800G ZR/ZR+ OSFP pluggables with integrated MACsec, and 2nm 1.6T ZR and O-band coherent-lite technologies. Marvell will also feature a 102.4 Tbps co-packaged optics platform with 200G-per-lane silicon photonics, aimed at improving bandwidth density, power efficiency and security across AI data center networks.
Marvell (MRVL), Microsoft and Utimaco launched Azure Payment HSM v2, a cloud-native payments security service, on September 17, 2026.
The fully managed platform combines Marvell LiquidSecurity hardware security modules, Utimaco Atalla Payments Module software and Microsoft Azure to secure critical payment workloads and transactions. It targets banks and payment providers needing PCI security, compliance, audit and operational controls without deploying or managing physical HSM infrastructure, and is available in public preview in Western U.S. and Western Europe.
GlobalFoundries (GFS) and Marvell (MRVL) have expanded a multi-year agreement to increase capacity for GF’s silicon germanium (SiGe) technology at its Burlington, Vermont facility to support growing demand for next-generation optical connectivity in AI and cloud data centers.
The collaboration targets high-performance SiGe used in pluggable optical transceivers, Near-Packaged Optics (NPO) and Co-packaged Optics (CPO), and is expected to add significant capacity for Marvell’s requirements. GF’s SiGe portfolio currently supports 200G-per-lane optical links and has a roadmap for higher-speed generations as bandwidth needs grow. The deal builds on more than a decade of joint work between the companies.
Marvell (MRVL) will showcase its end-to-end AI data center connectivity and memory portfolio at AI Infra Summit 2026, held September 15–17 at the Santa Clara Convention Center in California.
Demonstrations at Booth #704 will cover Photonic Fabric technology for multi-rack memory sharing, the RELIANT interconnect telemetry platform for real-time link health monitoring, and the Teralynx T100 102 Tbps Ethernet switch for large-scale AI networks. Marvell will also highlight a 256-lane PCIe 6.0 scale-up switch, CXL-based memory expansion and pooling solutions, high-speed storage offerings including a PCIe 6.0 NVMe SSD controller, and DPUs with Ethernet switching for NVMe-over-network access. Company executives will give presentations and panels on next-generation AI data center connectivity, including a keynote titled “AI Inference at Scale.”