Marvell Technology announced on April 24, 2026, that it has been selected by Google to design and manufacture next-generation custom artificial intelligence inference chips. This partnership marks a significant expansion of Marvell’s custom compute business, as the company will provide specialized silicon tailored to Google’s specific data center requirements for large-scale AI model execution. The collaboration focuses on developing high-performance, energy-efficient application-specific integrated circuits (ASICs) that utilize Marvell’s advanced 2-nanometer process technology and sophisticated chiplet-based packaging solutions.

According to official statements from Marvell, the new inference chips are designed to optimize the processing of complex AI workloads, including generative AI and large language models, while minimizing power consumption. Google’s decision to partner with Marvell follows the tech giant’s ongoing strategy to diversify its hardware supply chain and enhance its internal compute capabilities. Amin Vahdat, Google’s Vice President and General Manager of Machine Learning, Systems, and Cloud AI, noted that the collaboration is intended to accelerate the deployment of AI-driven services across Google’s cloud infrastructure.

Simultaneously, Marvell announced the acquisition of Polariton Technologies, a pioneer in plasmonic electro-optic modulators based in Zurich, Switzerland. The acquisition is designed to bolster Marvell’s optical connectivity portfolio, specifically addressing the increasing demand for high-bandwidth data transmission within AI data centers. Polariton’s technology utilizes plasmonics to enable modulators that are significantly smaller and faster than traditional silicon photonics components, capable of reaching speeds exceeding several hundred gigahertz while maintaining high energy efficiency.

Matt Murphy, Chairman and CEO of Marvell Technology, stated that the integration of Polariton’s plasmonic technology will allow Marvell to overcome physical limitations currently facing optical interconnects. By combining these advanced modulators with Marvell’s existing digital signal processing and networking platforms, the company aims to provide a comprehensive connectivity solution for the next generation of AI clusters. The acquisition is expected to close by the end of the current fiscal year, subject to customary regulatory approvals.

While the financial terms of the Polariton acquisition were not publicly disclosed, Marvell indicated that the transaction will be funded through existing cash reserves. Regarding the Google partnership, Marvell executives projected that the custom silicon project would begin contributing to the company’s revenue in the latter half of 2027. These developments underscore Marvell’s strategic focus on the data center market, which now accounts for a substantial portion of the company’s total revenue. The company remains focused on delivering integrated silicon solutions that address the dual challenges of compute power and interconnect bandwidth in the evolving AI landscape.