← Back to Library

Satya nadella – how Microsoft thinks about agi

The conversation begins with Nadella reflecting on why he believes artificial general intelligence represents the most significant technological shift since the industrial revolution, while also acknowledging we're still in early stages. He draws on his experience as a Turing Award winner to frame AI fundamentally as either a cognitive amplifier or guardian angel—a tool that extends human capability rather than replaces it.

The Scale of Infrastructure

Nadella and Scott Guthrie gave Dwarkesh Patel a tour of Microsoft's newest Fairwater 2 data center, currently the most powerful in the world. The facility aims to increase training capacity tenfold every 18-24 months—effectively representing a ten-times jump from what powered GPT-5. The network optics in this single building nearly match all of Azure's total capacity from two and a half years ago.

Satya nadella – how Microsoft thinks about agi

The design space for these massive computing installations requires balancing decisions around model architecture, physical planning optimized for specific hardware like GB200s and NVIL, and preparing for future chips that will fundamentally change cooling requirements and power density. The goal is aggregating flops across regions for large training jobs while maintaining flexibility to adapt as technology evolves rather than locking into one specification.

Economic Transformation

The shift from software licenses to subscriptions transformed Microsoft's business, but the next transition involves costs of goods sold becoming far more significant. Nadella argues that AI will mirror what cloud computing did—expanding markets dramatically by enabling fractionally cheaper access. When coding tools like GitHub Copilot grew to near a billion in revenue within one year without obvious competitors initially, it demonstrated how AI expands market opportunity rather than just capturing existing demand.

The pricing models will remain similar: subscriptions for entitlements, consumption-based pricing, advertising units, and device gross margins. Microsoft is positioned across all these categories at the portfolio level, though time will reveal which models make sense in specific categories.

The AGI Question

Nadella frames his optimism carefully—excited about AI's potential while grounded in the reality that this remains early innings despite useful tools emerging and scaling laws continuing to work. He's optimistic they'll continue working but acknowledges real science breakthroughs are needed alongside engineering progress. The economic growth question centers on what happens when machine-generated tokens become valuable: if one million tokens of 'Satia' carry significant worth, where do margins flow and what level of value capture becomes Microsoft's role?

The key insight is that even as technology diffuses rapidly, true economic growth requires changing work artifacts and workflows—a transformation corporations must undergo that shouldn't be discounted. The compression of what took 150 years during the industrial revolution into a 20-year period represents the fundamental shift underway.

Bottom Line

Nadella articulates a vision where AI serves as human augmentation rather than replacement—the 'cognitive amplifier' framing offers both strategic optimism and philosophical grounding. His infrastructure bet is clear: build for flexibility across generations of hardware rather than optimizing for one model. The vulnerability lies in whether consumption-based pricing can sustain margins when AI costs fundamentally change the economics of software subscriptions.

Deep Dives

Explore these related deep dives:

Sources

Satya nadella – how Microsoft thinks about agi

by Dwarkesh Patel · Dwarkesh Patel · Watch video

Maybe after the industrial revolution, this is the biggest thing. But at the same time, I'm a little grounded in the fact that this is still early innings. If you're a model company, you may have a winner's curse. You may have done all the hard work, done unbelievable innovation, except it's kind of like one copy away from that being commoditized.

We didn't want to just be a host for one company and have just a massive book of business with one customer. That's not a business. You can't build an infrastructure that's optimized for one model. If you did that, you're one tweak away.

Some like breakthrough that happens and your entire network topology goes out of the window. Then that's a scary thing. Our business, which today is an enduser tools business, will become essentially an infrastructure business in support of agents doing work. The thing that you have to think through is not what you do in the next 5 years, but what do you do for the next 50.

>> Today we are interviewing Satya Nadella. We being me and Dylan Patel who is founder of semi analysis. Satya, welcome. >> Thank you.

It's great. Thanks for coming over at Atlanta. >> Yeah, thank you for giving us a tour of the new facility. It's been really cool to see.

>> Absolutely. >> Satya and Scott Guthrie, Microsoft's EVP of cloud and AI, give us a tour of their brand new Fairwater 2 data center, the current most powerful in the world. >> We try to 10x the training capacity every 18 to 24 months. And so this would be effectively a 10x increase 10x from what GPD5 was trained with.

And so to put in perspective, the number of optics, the network optics in this building is almost as much as all of Azure across all our data centers 2 and a half years ago. >> It's kind of what 5 million network connections. >> You've got all this bandwidth between different sites in a region and between the two regions. So is this like a big bet on scaling in the future that you anticipate in the future there's going to be some huge model that needs to require two whole different regions to train >> the goal is to be able to kind of aggregate these flops for a large training job ...