Microsoft is developing its own AI chips, known as Athena, to power large language models. The project began in 2019 and is currently used by some Microsoft and OpenAI employees. With an uncertain timeline for wider usage, Microsoft is reportedly planning to manufacture Athena chips with contract foundry TSMC, using a 5-nanometer process.
Key Takeaways:
- Athena AI chips: Microsoft is developing its own AI chips to power large language models, with around 300 staff working on the project.
- Current usage: Athena chips are being used by some Microsoft and OpenAI employees, indicating a close working relationship between the two companies.
- Potential wider adoption: Athena could see broader usage by both Microsoft and OpenAI sometime next year, though the company is unsure whether to make the chips accessible to Azure cloud computing customers.
- Manufacturing plans: Microsoft is planning to use contract foundry TSMC to manufacture the chips, based on a 5-nanometer process, which is one generation behind the 3nm process introduced late last year.
- Competitors' efforts: Other tech giants, such as Google and Meta, are also working on their own AI chips, showcasing the increasing interest in AI hardware development.
Business professionals should keep an eye on the development of AI chips like Athena, as these advancements could potentially improve the performance of large language models and influence future AI applications and their accessibility.