According to reports, UK chip designer Arm Holdings has recruited a top chip engineer from Amazon Web Services (AWS).
Rami Sinno previously led Arm’s engineering team before joining Amazon, where he was responsible for driving the development of the Trainium and Inferentia chips for the cloud computing giant’s Annapurna Labs.
According to sources cited by Reuters, Sinno will return to Arm to help the company transition from a traditional IP licensing designer to a chip manufacturer.
Sinno has extensive experience in physical chips and machine learning infrastructure, but it is currently unclear whether his work at Arm will involve dedicated AI accelerators like Trainium and Inferentia, or primarily focus on Arm’s core business—CPUs.
Arm declined to comment on the matter. El Reg also reached out to Sinno, but had not received a response by the time of publication.
Historically, Arm has not produced chips itself but has allowed companies like Apple, Qualcomm, MediaTek, Amazon, Google, Microsoft, and NVIDIA to create their own chips based on Arm-designed cores through licensing processor architectures.
In addition to CPU cores, Arm also designs GPUs and NPUs, primarily for edge computing and smartphone applications. While Arm has not yet directly challenged AMD and NVIDIA in the data center GPU market, it is not unfamiliar with the AI infrastructure space.
For example, Arm’s Neoverse V2 architecture is currently at the core of all NVIDIA GB200 and GB300 NVL72 rack systems, and NVIDIA’s upcoming Vera CPU will also be based on Arm technology.
However, Arm’s ambitions seem to extend beyond architecture licensing or CPU cores. Last month, Arm CEO Rene Haas stated that the company is exploring the possibility of bringing its designs directly to silicon, potentially in the form of chiplets or complete CPU packages.
He mentioned during the company’s Q1 earnings call, “We are exploring possibilities beyond our current platforms, including more computing subsystems, chiplets, and even complete end-to-end solutions.”
Meanwhile, Amazon is also making significant bets on AI. For instance, its supercomputing cluster Project Rainier built for Anthropic, as well as the next-generation Trainium2 chip, demonstrate AWS’s strategic focus on AI silicon. In fact, Amazon currently deploys over half of the Arm server CPUs globally.
In recent years, Arm has been pushing for more complete chip design solutions, such as computing subsystems (CSS), which include everything needed to bring a chip to market, allowing companies to either use them directly or customize them. For example, Microsoft’s Cobalt CPU is believed to be based on this “ready-made solution.”
Considering that Arm’s customers are already accustomed to directly adopting its Cortex and Neoverse cores, chipletization is seen as a natural evolutionary direction, providing standardized modules while allowing customization in areas like I/O, memory, and packaging.
However, if Arm were to launch more complete CPU or SoC package products, it could directly compete with its largest customers, a move that might backfire and force some customers to turn to alternative computing architectures like RISC-V.