Microsoft wants to develop their own AI chips for OpenAI to use with tools like GPT-4 and ChatGPT, and has just added a luxury ally: AMD. This is how he assures it Bloombergwhich reports that the companies reached an agreement to work together on the creation of the hardware in question.
The choice of AMD as a new partner is far from being casual, according to the aforementioned media. Those of Lisa Su seek to expand their presence in the market for components applied to artificial intelligence, to close the gap with NVIDIA. Thus, Redmond will provide financial support to Californians and in return will receive help in the development of Athenahis own AI chip.
But the alliance isn’t just looking to accelerate the design of a Microsoft-branded GPU for training and running AI models. Money from Satya Nadella’s firm will also allow that AMD move forward with its own initiatives in the matter. In fact, the CEO of the Santa Clara chipper herself recently indicated that among her future plans is to create partially customized chips for her main clients to use in their AI data centers.
The deal also makes it clear that proprietary component development is an area of growing interest at Redmond. It’s true that Microsoft has already worked on ARM-based hardware for Surface computers with Qualcomm; but what you are looking for in this alliance with AMD is on another level.
AMD will develop Microsoft chips to power ChatGPT
A short time ago Microsoft’s initiative to create its own hardware for artificial intelligence applications came to light. A project that was launched in 2019, but which gained real momentum in recent months with the rage for ChatGPT. However, until now the involvement of AMD was unknown.
Keep in mind that generative AI tools require hellish computing power to work, and the components required aren’t exactly cheap. For this reason, those from Redmond would seek to lower costs by producing their own GPUs and providing access to them to OpenAI.
Working with AMD would allow them reduce reliance on third-party vendors like NVIDIA. Although neither Microsoft nor Sam Altman’s startup would intend to cut ties with those led by Jensen Huang. In reality, what would be sought would be to have guaranteed access to the necessary hardware, no matter how much the demand for these components grows in the market.
As recently learned, the proprietary chips that Microsoft wants to use to power ChatGPT and other AI tools are already being tested internally. It is even rumored that this would be the first step in a more ambitious plan. Bloomberg mentions that the company has already earmarked more than 2,000 million dollars to its efforts in the field of semiconductors. Furthermore, it indicates that Athena’s development has become the focus of the division’s attention.
For now, Microsoft and AMD have chosen to remain silent about this alliance. The intention of those from Redmond would be to start massively deploying the chips for ChatGPT and other AI applications from 2024.