Marvell Technology is a semiconductor company founded in 1995 with operational headquarters in Santa Clara, California (incorporated in Wilmington, Delaware). It was founded by Dr. Sehat Sutardja, his wife Weili Dai and his brother Pantas Sutardja; the company went public in June 2000. Matthew Murphy has served as CEO since July 2016, with Chris Koopmans as President and COO. Marvell develops and produces semiconductors and related technology with a focus on data infrastructure.
AI relevance
The portfolio includes data processing units from the OCTEON and ARMADA product lines, custom ASICs, and networking and storage solutions such as the Teralynx and Prestera Ethernet switch families. For AI data centers, Marvell is particularly relevant as a supplier of custom accelerator chips (XPUs) and of optical and electrical interconnect technology — including PAM4 and coherent DSPs — that enables data flow between AI servers. In December 2024, Marvell expanded its strategic collaboration with Amazon Web Services through a five-year, multi-generational agreement covering custom AI chips, optical DSPs, PCIe retimers and Ethernet switching solutions.
Business model
Marvell is a fabless supplier focused on chips for data centers, networking and storage. By its own estimates, the addressable market for data-center semiconductors will grow to $94 billion by 2028, of which roughly $55.4 billion is for custom devices and roughly $40.8 billion for custom accelerators, growing at an annual rate of about 47%.
Key figures
For fiscal year 2026 Wikipedia cites revenue of roughly US$8.19 billion and about 7,480 employees. In the first quarter of fiscal year 2026, Marvell reported revenue of $1.895 billion, up 63% year over year; the data center segment alone contributed $1.44 billion, up 76%, representing 76% of total revenue. The shares trade on Nasdaq under the ticker MRVL (ISIN US5738741041).
A specialist in data infrastructure
Marvell focuses on semiconductors for data infrastructure: chips that move, process and secure data in data centers, networks and storage systems. Through targeted acquisitions — including Cavium (2018), Inphi (2021) and Innovium (2021) — the company has expanded its portfolio in areas such as optical transmission, Ethernet and custom silicon. In 2025, Nvidia invested roughly $2 billion in Marvell as part of a partnership to integrate Marvell's custom chips and networking technology into Nvidia's NVLink Fusion interconnect ecosystem.
AI drives data traffic
As AI data centers grow, so does the need to move large amounts of data quickly between accelerators, servers and storage. Marvell addresses this need with interconnect technology and custom accelerator chips that it develops together with large cloud customers such as AWS. As a fabless supplier, Marvell relies on leading contract manufacturers. The business is cyclical and shaped by the investment behavior of data-center operators; custom projects can lead to larger but also concentrated revenue contributions. Marvell is thus a less visible but important building block of AI infrastructure.
This profile is a neutral description and is not investment advice.