Microsoft is reportedly developing its own AI chip to power OpenAI projects and reduce costs associated with relying on third-party platforms such as Nvidia. Titled ‘Athena’, after the Greek goddess of war, the chip is reportedly being tested by Microsoft’s own machine learning staff and OpenAI developers.
The news comes as many big tech companies are moving towards developing their own machine-learning chips. Notable examples include Google and Amazon, both of whom have released their own models in recent years. The rising market scarcity and cost of Nvidia’s H100, one of the most popular GPUs for training machine learning systems, is closing in on the $40,000 mark on reseller services like eBay. Thus, the development of Hermes and Athena could be a way for Microsoft to decrease their dependency on third parties.
The move also follows several other big AI moves from big tech companies in the past few weeks. Elon Musk of Tesla recently announced the impending launch of TruthGPT, a model that “seeks the truth” and is meant to counter ChatGPT’s alleged left-wing bias. Amazon also recently released its Bedrock AI framework with its own self-developed models.
Sam Altman, co-founder and CEO of OpenAI, recently stated that the infrastructure that got his company’s GPT-1 to GPT-4 is played out and needs to be given a new perspective. It may be that Microsoft’s Hermes and Athena chips are just the technology to serve as the new backbone of OpenAI’s work.
Microsoft is one of the biggest tech companies in the world, offering a range of products, from software to computer hardware. Founded by Bill Gates and Paul Allen in 1975, the company has since become synonymous with computing and the software industry, with a range of market-leading applications and platforms such as Windows, Office and the Xbox. Microsoft’s development of its own AI chip is just the latest of its multiple projects that aim to make use of cutting-edge machine learning technologies.
