AMD acquires Mipsology to ramp up AI inference rivalry with Nvidia
The acquisition is part of AMD’s larger strategy to challenge Nvidia’s dominance in AI computing
AMD has acquired French startup Mipsology to strengthen its AI inference software capabilities as the chip designer mounts its biggest challenge yet to AI chip powerhouse Nvidia.
The California-based company unveiled the acquisition saying that Mipsology, "a leader in AI software and long-standing AMD partner," will help the company "accelerate our customer engagements and expand our AI software development capabilities."
"Specifically, the team will help develop our full AI software stack, expanding our open ecosystem of software tools, libraries and models to pave the way for streamlined deployment of AI models running on AMD hardware," wrote Vamsi Boppana, senior vice president of AMD's AI group.
An AMD spokesperson declined to disclose financial terms or say how many employees Mipsology has.
Mipsology's expertise is in developing "plug-and-play" software that speeds up AI inference performance without requiring new tools or changes to neural network models, which are key to AI applications.
The software, called Zebra, is focused on enabling "world-class speed" on FPGAs, short for field programmable gate arrays, that AMD gained with its acquisition of Xilinx, according to its website.
Inference is a critical aspect of AI, allowing applications to make predictions and generate responses based on models that have been trained to recognise patterns or objects.
Supporting AMD's larger AI strategy
The acquisition is part of AMD's larger strategy to challenge Nvidia's dominance in the AI computing space with a broad portfolio of what AMD Chair and CEO Lisa Su has called "leadership GPUs, CPUs and adaptive computing solutions for AI inferencing and training."
Since last year, Su and other executives have articulated a comprehensive AI strategy that addresses what she said is a "multibillion-dollar growth opportunity for AMD across cloud, edge and an increasingly diverse number of intelligent endpoints."
In June, AMD revealed its biggest challenge yet to Nvidia's most powerful AI chips, the Instinct MI300 series.
The new line-up includes the MI300X, which Su said will provide better efficiency and cost savings for running large language models than Nvidia's flagship H100 data centre GPU.
The company also sees a significant AI opportunity with the FPGA-based products it gained from the Xilinx acquisition, which the company now refers to as adaptive computing chips.
These products target workloads anywhere from small endpoints and edge computers to data centres.
But for AMD to effectively challenge Nvidia in the AI computing space, the chip designer needs to attract software developers to build on its hardware.
That's why the company is consolidating previously disparate software development stacks for CPUs, GPUs and adaptive chips to provide a "cohesive AI training and inference interface" it has called the AMD Unified AI Stack.
It's here where AMD is hoping Mipsology, which was founded in 2015 by former employees at chip tool design vendor Synposys, will help the company make further improvements.
"AI is our top strategic priority and a significant driver of incremental silicon demand over the coming decade," Boppana wrote in the announcement.
"By welcoming the skilled Mipsology team to AMD, we will continue to strengthen our software capabilities to enable customers around the world to tap into the vast potential of pervasive AI."