Marvell Technology, the fabless semiconductor firm christened the “next trillion-dollar company” by Nvidia CEO Jensen Huang, has poached AMD’s Epyc and Instinct platform engineering lead.

Jay Kirkland had been at AMD since 2020, but joins Marvell as SVP of custom silicon engineering. The exec said he was drawn to Marvell by the “opportunity to help solve what I believe are some of the biggest engineering challenges our industry has ever faced.”

Kirkland joins a Californian chip firm developing custom specialized processing units (XPUs) in addition to optical interconnect offerings. Marvell is already hard at work on custom chip lines that can be integrated with Nvidia hardware following a $2 billion investment back in March. The firm is also believed to be in talks with Google to break Broadcom’s stranglehold on designing the hyperscaler’s custom Tensor Processing Units (TPUs), with reports suggesting Marvell is in line to develop a version specifically for running AI models.

Kirkland’s role will see him support Marvell’s custom silicon push, with the SVP suggesting such opportunities in AI are “expanding faster than our industry’s ability to deliver them.”

“Success will depend on bringing together exceptional talent, deep customer partnerships, and a broad set of technologies to build the infrastructure that enables what’s next,” Kirkland said in a post on LinkedIn. “Marvell’s engineering excellence, trusted customer relationships, and breadth of technologies across AI infrastructure create a unique opportunity to help customers solve challenges that no single technology can address alone.”

Kirkland spent a little over six years at AMD, where he built and led customer engineering, platform engineering, and AI enablement organizations, supporting both hyperscale and AI customers.

Kirkland also served for more than two decades at Intel across various engineering leadership roles. Among his accomplishments at the chip giant were the establishment of the Board Engineering and Support Division (BESD) as part of Intel Data Center Group’s platform engineering team, which encompassed some 800 hardware and software developers.

Prior to joining AMD, Kirkland served as VP for platform engineering in Intel’s Programmable Solutions Group, and would go on to play a pivotal role in bringing multiple generations of AMD’s EPYC server platforms to life.

Upon his AMD exit, the exec said in an earlier post: “I am proud of what we accomplished together and grateful for the relationships, experiences, and lessons gained along the way. AMD’s future is bright, and I look forward to watching the company continue its remarkable momentum.”

Kirkland joins a company on the up, with Huang lauding the brand at Computex, the cherry on the icing on the cake of a year in which its stock price is currently up almost 207 percent.

In addition to custom silicon, networking hardware looks to be central to Marvell’s growth, with its Taiwan trip seeing it unveil the Teralynx T100, a 102.4 Tbps (102.4T) switch customized for AI and cloud data center infrastructure.

The company also recently acquired silicon photonics startups Polariton Technologies and Celestial AI, the latter of which it paid $3.25 billion for, with it betting on optical interconnects to boost customer data center connectivity.

Kirkland is the second senior executive to join Marvell of late, following Dan Durn, who took on the CFO role in June. Durn joined from Adobe, replacing Willem Meintjes, who recently resigned, though he will stay on in an advisory capacity through April 2027.