Marvell Buys XConn for $540M to Deepen AI Data Center Reach
The acquisition bolsters Marvell's high-speed connectivity portfolio with key PCIe and CXL technologies, aiming to capture growth in AI infrastructure.
Marvell Technology Inc. (NASDAQ: MRVL) has agreed to acquire XConn Technologies for approximately $540 million in cash, a strategic move to fortify its leadership in the high-speed connectivity solutions powering the world’s most advanced AI and cloud data centers. The deal, announced Tuesday morning, sent Marvell's shares up nearly 3% in early trading.
Shares of the Wilmington, Delaware-based semiconductor firm climbed to $92.71, adding to a bullish trend that has seen the stock trade consistently above its key 50-day and 200-day moving averages. The acquisition targets the critical plumbing of AI infrastructure, focusing on the high-bandwidth, low-latency interconnects essential for training and deploying complex artificial intelligence models.
In a statement released before the market opened, Marvell outlined its plan to integrate XConn's technology and engineering talent. XConn specializes in next-generation Peripheral Component Interconnect Express (PCIe) and Compute Express Link (CXL) switches, open-standard technologies that are rapidly becoming the backbone of modern server architecture.
"This combination creates a compelling switching platform for accelerated infrastructure, advancing Marvell’s connectivity strategy for next-generation AI and cloud data centers," said Matt Murphy, Chairman and CEO of Marvell. "With XConn, we add proven PCIe and CXL switch products, IP, and engineering talent to expand our UALink scale-up switch team."
The acquisition is the latest step in Marvell's strategy to assemble a comprehensive portfolio for accelerated infrastructure. Murphy noted that the XConn deal, combined with its pending acquisition of optical interconnect firm Celestial AI, will better position Marvell to provide the performance and flexibility required as AI systems grow in size and complexity. With a market capitalization now exceeding $77 billion, Marvell is aggressively positioning itself as a one-stop-shop for the foundational hardware that underpins the AI revolution, a market largely dominated by Nvidia but with intense competition from players like Broadcom.
As AI workloads intensify, data centers require new ways to connect massive numbers of accelerators—like GPUs—and share pools of memory efficiently. This is where XConn’s technology becomes critical. The company provides advanced switching solutions based on PCIe and CXL, which allow data to move at high speeds between processors, memory, and accelerators.
CXL, in particular, is a crucial technology for enabling memory disaggregation, allowing servers to share resources more effectively to handle the enormous datasets and parameter sizes of large language models. The combination of Marvell's existing CXL memory-expansion controllers with XConn's CXL switches promises a comprehensive solution for these demanding AI workloads.
"At XConn we have built the industry’s highest-port-count advanced PCIe 5 and PCIe 6 switching portfolio to support the next generation of accelerated infrastructure," said Gerry Fan, CEO of XConn. "Marvell brings cutting-edge SerDes technology, a leading process roadmap, deep hyperscale customer relationships, and global scale."
Marvell stated the acquisition will expand its total addressable market by capturing the growing opportunity in the PCIe and CXL switch sector. According to market analysis shared by sources like Benzinga, this move is aimed squarely at dominating the AI connectivity space.
The financial impact of the transaction is not expected to be immediate. Marvell anticipates revenue contribution from XConn's products, which are already in production or sampling, to begin in the second half of fiscal year 2027. The company projects the deal could add approximately $100 million in revenue in fiscal 2028, signaling this as a long-term strategic investment rather than a short-term revenue boost.
The transaction is expected to close in the second half of calendar year 2024, subject to customary closing conditions. With this acquisition, Marvell is making a clear and costly bet that the future of AI will be decided not just by the power of chips themselves, but by the speed and efficiency of the connections between them.