![](https://www.techmonitor.ai/wp-content/uploads/sites/29/2023/10/shutterstock_1954818265.jpg)
AMD plans to buy AI chip start-up Nod.ai as part of its bid to more effectively compete with Nvidia by boosting its software capabilities. The chipmaker has previously said it intends to invest heavily in software alongside its most advanced chips to improve performance and functionality. Nod.ai builds open-source compiler-based systems that can auto-schedule compute and communications resources over CPUs, GPUs and accelerators.
![AMD employs more than 1,500 engineers in its AI Group, mainly focused on software and plans to expand this through acquisitions (Photo: JHVEPhoto / Shutterstock)](https://techmonitor.ai/wp-content/uploads/sites/4/2023/10/shutterstock_1954818265-1024x682.jpg)
The chipmaker says Nod.ai will bring an experienced team into the AMD fold that has already developed industry-leading technology designed to run on AMD data centre chips. AMD says it will continue to provide the software as part of an open ecosystem that lowers the barriers to entry through developer tools, libraries and models.
The move comes as competition in the AI chip space is set to increase, with OpenAI actively investigating whether to build its own chips for training next-generation models and SoftBank’s Arm pursuing a data centre strategy fresh from its IPO. Nvidia continues to dominate in the space with demand for its GPUs in training AI models causing a shortage.
AI companies need thousands of chips to run their systems, with the AI supercomputer built by Microsoft for OpenAI powered by 10,000 of Nvidia’s A100 GPUs, which come at a cost of $10,000 a time. The H100, its recently released successor, is even more expensive, at some $30,000 per chip, and with Nvidia only expecting to ship 550,000 this year, supply problems are likely to be an ongoing issue for buyers. OpenAI CEO Sam Altman has openly complained that a lack of GPU availability is hindering his company’s progress.
Can AMD differentiate through software?
Software is emerging as a key differentiator for AMD compared with Nvidia. It has previously claimed that through automation and software, it can achieve 80% of Nvidia’s performance on similar-age chips when training large language models. “The acquisition of Nod.ai is expected to significantly enhance our ability to provide AI customers with open software that allows them to easily deploy highly performant AI models tuned for AMD hardware,” said Vamsi Boppana, senior VP of the AI Group at AMD.
Nod.ai produces software already in use by cloud hyperscalers. It includes the SHARK package, which reduces the need for manual optimisation and cuts the time required to deploy a highly performant AI model across a portfolio of data centre, edge and client platforms. Under the AMD AI Group, the Nod.ai team will more closely integrate its software into the AMD architecture, and enhance existing AI software products.
“At Nod.ai, we are a team of engineers focused on problem-solving – quickly – and moving at pace in an industry of constant change to develop solutions for the next set of problems,” said Anush Elangovan, co-founder and CEO, of Nod.ai. “Our journey as a company has cemented our role as the primary maintainer and major contributor to some of the world’s most important AI repositories, including SHARK, Torch-MLIR, and OpenXLA/IREE code generation technology. By joining forces with AMD, we will bring this expertise to a broader range of customers on a global scale.”
AMD has not revealed the value of the deal to acquire Nod.ai but the California-based company raised $36.5m so far and the employees will fall within the 1,500-strong AMD AI group, which was formed earlier this year to house its acquisitions. AMD says it will add an additional 300 hires to the group this year and more next year as it continues to focus on using software to drive growth.
Boppana added: “The addition of the talented Nod.ai team accelerates our ability to advance open-source compiler technology and enable portable, high-performance AI solutions across the AMD product portfolio. Nod.ai’s technologies are already widely deployed in the cloud, at the edge and across a broad range of endpoint devices today.”