Microsoft wants to ditch NVIDIA and AMD chips for in-house custom silicon

by

Recently, Microsoft has been working hard to build its own strength in the rapidly growing field of artificial intelligence. While the company has invested $13 billion in OpenAI – the creators of ChatGPT – and uses their technology in many of its products, Microsoft is also focused on developing its own AI capabilities.

Satya Nadella, CEO of Microsoft, has signaled a shift away from Bill Gates’ original focus on just software. The company is now expanding into areas like artificial intelligence, integrating different technologies, and developing intelligent solutions.

Earlier this year, OpenAI announced its ambitious $500 billion Stargate project, aiming to build data centers throughout the U.S. to support its cutting-edge AI technology. As a result, Microsoft is no longer OpenAI’s exclusive cloud provider, although Microsoft still has the first opportunity to be considered for future cloud services.

Salesforce’s CEO, Marc Benioff, believes Microsoft will eventually stop relying on technology from OpenAI. This idea gained more weight recently when Microsoft’s AI chief, Mustafa Suleyman, revealed the company is building its own AI models. However, Suleyman acknowledged these models will initially lag behind OpenAI’s by three to six months. He explained that Microsoft’s approach is to be a close second to OpenAI in AI development.

Microsoft is planning to use its own AI chips in its data centers, according to Chief Technology Officer Kevin Scott. This could reduce the company’s dependence on chips made by AMD and NVIDIA, as reported by CNBC.

According to Scott:

We don’t limit ourselves to specific chip brands. For a long time, Nvidia has consistently offered the best combination of price and performance. We’re open to considering any option to make sure we have enough processing power to handle the current demand.

All the big tech companies are rushing to incorporate artificial intelligence, driving up the need for the specialized chips that power it. This increased demand is a major reason why NVIDIA, a leading chipmaker, has seen its value soar, recently reaching $4 trillion.

I was really excited to hear that Microsoft is planning to power most of its data centers with chips they design themselves! Even cooler, they’re not just *planning* for the future – they’re actually using their own chips right now. It’s a big move and shows they’re serious about controlling their hardware destiny.

Microsoft’s AI chief, Mustafa Suleyman, recently announced the company is working on creating its own AI chip cluster. This move aims to make Microsoft less reliant on others for its AI technology.

We need to be able to create our own cutting-edge AI models, ranging in size, but we should also be practical and use existing models when they fit our needs. For a company like ours, with so many different businesses, it’s essential that we can handle our AI requirements independently if we choose to.

Microsoft AI CEO, Mustafa Suleyman

Microsoft introduced the Azure Maia AI Accelerator in 2023 to improve performance for artificial intelligence tasks. Now, reports suggest they are designing their own advanced chips. This will allow Microsoft to create complete, customized systems for their data centers, tailored to specific requirements.

According to Scott:

This is about designing the whole computing system – including the network and cooling – and having the flexibility to adjust things to get the best performance for whatever you’re working on.

Recently, the relationship between Microsoft and OpenAI appears to be weakening. Reports suggest Microsoft canceled two large data center projects because they didn’t want to help further train ChatGPT. However, OpenAI’s CEO, Sam Altman, has stated that they are no longer limited by computing power.

Stay up-to-date with the latest news, insights, and features from Windows Central by following us on Google News!

Read More

2025-10-03 14:43