您现在的位置是：Microsoft to launch its new AI chip Athena next month >>正文

Microsoft to launch its new AI chip Athena next month

上海品茶网 - 夜上海最新论坛社区 - 上海千花论坛5人已围观

简介By subscribing, you agree to our Terms of Use and Policies You may unsubscribe at any time.Microsoft...

By subscribing, you agree to our Terms of Use and Policies You may unsubscribe at any time.

Microsoft has been developing a computer chip under a project codenamed ‘Athena’ for some time. Now, as per a recent report, it is believed that the company is ready to unveil the AI chip at its annual event ‘Ignite’ next month.

Microsoft to launch its new AI chip Athena next month

This is a significant development because it would mean that Microsoft would be able to reduce its reliance on GPUs produced by NVIDIA, the uncrowned king of the semiconductor industry, reported Interesting Engineeringin April. Athena has been designed to power Microsoft’s data center servers that are used to train and run large language models.

Semiconductor race heats up

Everybody wants an NVIDIA chip to power their AI systems, but unfortunately, they are in short supply and come with a hefty price tag. The H100 chip, which is currently the US-based chipmaker’s most powerful AI offering, is priced at $40,000. Microsoft also uses NVIDIA chips to power its large language models, which require high computing power. Producing in-house GPUs would save cost – and let's not forget that having its own chip would mean the company can push Azure ahead harder.

See Also Related
  • Codename Athena: Microsoft developing secret AI chips to challenge Nvidia's dominance 
  • OpenAI may produce the hottest commodity in tech: AI chips 

IEhad also reported earlier that Microsoft has been secretly developing these chips since 2019. Incidentally, that is the same year when Microsoft invested $1 billion in OpenAI, the father of ChatGPT. Over the years, Microsoft has invested close to $13 billion in the AI company. One might say that apart from giving support to its own AI goals, Microsoft’s chip could also assist OpenAI in its own GPU needs.

OpenAI developing its own AI chip

But it seems OpenAI may have taken the matter of its GPU needs into its own hands. IEreported last week that the Sam Altman-led company is also interested in developing its own chip. Either that or acquire a chipmaker that could make customized chips for OpenAI.

This could bode well for the company because, as per a Reutersreport, if ChatGPT were to grow its queries to a 10th of the scale of Google search, OpenAI would be required to cough up roughly $48.1 billion to buy the necessary GPUs and then about $16 billion worth of chips a year to stay operational. Altman has also previously complained about issues resulting in the functioning of his products due to a GPU shortage.

Till now, ChatGPT has been running on 10,000 NVIDIA GPUs, which have been integrated into a Microsoft supercomputer. As ChatGPT shifts from being a free service to categorizing itself as a commercial service, it would require over 30,000 NVIDIA A100 GPUs.

The global chip supply shortage has spurred the prices of chips produced by NVIDIA, which announced that it would launch a new chip next year called the GH200, which will have the same GPU as the H100 but with triple the memory capacity. The company said systems running on GH200 will start in the second quarter of 2024.

‘Ignite’ is the Microsoft annual event where developers and IT professionals converge as the company announces important updates for its products. This year’s event is set to take place in Seattle from November 14 to 17.

Tags：

上一篇：Cryptos on the decline, Wall Street under pressure

下一篇：Startup exhibits wearable that displays captions in wearer's view

相关文章



热门文章

最新文章

友情链接