Microsoft Supercharges AI Ambitions with OpenAI’s Chip Designs, Says Nadella: “We Get It All”

Microsoft is taking a major leap in its artificial intelligence (AI) hardware journey by joining forces even more closely with OpenAI — not just in software, but at the silicon level. CEO Satya Nadella has confirmed that Microsoft will incorporate OpenAI’s custom AI chip designs into its long-term semiconductor strategy, a move that could redefine how next-generation AI systems are built and scaled.

Speaking on a podcast, Nadella revealed, “We now have access to OpenAI’s chip and hardware research through 2030,” marking a decade-long deepening of their strategic alliance. Under the revised partnership, Microsoft will also continue using OpenAI’s AI models through 2032.

OpenAI, which has been co-developing advanced AI processors and networking hardware with Broadcom, is extending its innovation beyond algorithms into hardware. Microsoft plans to “industrialise” these designs, scaling them for mass production while embedding them into its own intellectual property portfolio. This collaboration is expected to fuel the company’s broader cloud and AI roadmap for the next decade.

A stronger Microsoft–OpenAI partnership

This marks a new chapter in one of tech’s most significant alliances. For Microsoft, the partnership provides faster access to cutting-edge hardware tailored to OpenAI’s model-training demands. For OpenAI, it unlocks Microsoft’s vast infrastructure to bring its innovations to a global scale.

It’s a virtuous cycle — OpenAI designs models that push hardware limits, and Microsoft builds systems capable of running them efficiently. Nadella described this as a “strategic alignment,” one that will accelerate Microsoft’s semiconductor ambitions and strengthen its competitive edge in the AI race.

Fairwater datacentres: Microsoft’s AI engine

At the heart of this effort lies Microsoft’s new Fairwater datacentre architecture — vast, interconnected facilities built for the AI era. Each Fairwater site operates as a node in a massive network designed to train and deploy large-scale AI models.

The Atlanta site, for example, is already operational and features NVIDIA GB200 NVL72 rack-scale systems that can scale to hundreds of thousands of Blackwell GPUs. It also employs advanced liquid cooling technology that consumes almost no water, setting a new sustainability benchmark in data infrastructure.

Scott Guthrie, executive vice president of Cloud + AI at Microsoft, explained, “Leading in AI isn’t just about adding more GPUs, it’s about building the infrastructure that makes them work together as one system.” He added, “Fairwater reflects that end-to-end engineering and is designed to meet growing demand with real-world performance, not just theoretical capacity.”

The broader vision: owning the AI stack

From its early supercomputers built with OpenAI in 2019 to the systems behind GPT-4 and beyond, Microsoft has been evolving every layer of AI infrastructure — chips, networks, and data architecture. The addition of OpenAI’s hardware expertise now positions Microsoft to control the full stack of AI innovation, from silicon to supercomputers to software.

As Nadella put it, this isn’t just about adding computing power — it’s about “owning the full stack of AI innovation.” In the global AI race, Microsoft is no longer just buying GPUs; it’s building the factory that makes them run as one unified system.