Microsoft, a tech giant known for its innovative solutions, has recently announced the development of its own custom-designed computing chips for artificial intelligence (AI).

The Genesis of Athena
The AI chip, internally referred to as “Athena”, has been in the works since 2019. The initial version of the chip is planned to use Taiwan Semiconductor’s (TSMC) 5-nanometer process. The Athena chips are reportedly designed to train large-language models (LLMs), including inferring from data the models acquire in training.

A Strategic Move Amidst Chip Shortages
The decision comes amidst chip shortages that have driven up the prices of highly sought-after Nvidia AI GPUs. Other companies, including Amazon, OpenAI, IBM, and AMD, have also been considering designing their own AI accelerator chips.

The Potential Impact
The introduction of Microsoft’s AI chips could potentially reduce the cost per chip by a third when compared with Nvidia’s offerings. This could result in significant savings for Microsoft, especially as it continues to add AI capabilities beyond Bing Chat to its Office products and GitHub.

The Future of AI at Microsoft
Microsoft’s move to develop its own AI chips underscores its commitment to pushing the boundaries of AI innovation. With its multi-billion investment in OpenAI, Microsoft is poised to make significant strides in the realm of AI.

Insiders View

In coclusion Microsoft’s Athena, debuting at Ignite 2023, marks a milestone in their AI commitment, though not a broad Nvidia replacement.