Marvell Locks TSMC 1.4nm with $1B Deal

According to industry reports, U.S.-based chip designer Marvell Technology is deepening its strategic collaboration with leading foundry TSMC, betting that surging demand for high-speed connectivity in artificial intelligence (AI) data centers will drive its next phase of growth.

Marvell President and Chief Operating Officer Chris Koopmans stated that the company has been in discussions with TSMC to adopt its upcoming A14 process technology, expected to correspond to an advanced 1.4nm-class node, for future-generation semiconductor products. TSMC is currently targeting 2028 for volume production of the A14 process.

To secure long-term manufacturing capacity amid intensifying competition for advanced nodes, Marvell has also committed to prepaying approximately $1 billion.

High-speed connectivity chips play a critical role in enabling efficient data movement within and between AI data centers. Koopmans emphasized that advanced interconnect technologies have become as essential as high-performance processors and cutting-edge memory in AI supercomputing systems, as they enable higher bandwidth, lower latency, and improved system efficiency.

Marvell has been actively expanding its presence in this segment through next-generation solutions. The company has developed a digital signal processor (DSP) for the industry’s first 1.6-terabit interconnect platform, manufactured using TSMC’s 3nm process. It has also become an early adopter of more advanced nodes, including 2nm-class technologies, for DSPs and data center interconnect (DCI) pluggable solutions aimed at enabling ultra-high-speed data transmission in future AI infrastructure.

The company’s momentum has been further reinforced by strategic industry partnerships. NVIDIA has invested $2 billion in Marvell to strengthen collaboration in AI infrastructure development. Together, the two companies are working on high-speed interconnect solutions and custom processors designed to integrate with NVIDIA’s NVLink ecosystem, supporting next-generation AI computing platforms.