A conversation with Satya

Satya Nadella of Microsoft has been in the press quite a bit recently with his comments that we have moved to the age of digital agents and they will cause the end of SaaS and broadly other business applications, which he says essentially only provide skin deep services front-ending databases – create, replace, update, delete (CRUD)

Whenever I hear “end of” or “death of” something, it triggers for me two other thought leaders:

• In 1989, Francis Fukuyama, then deputy director of the State Department's Policy Planning Staff predicted the fall of Communism and the world beyond in an article titled “The end of history”. He was mocked by world leaders like Margaret Thatcher and history has certainly not ended with Hamas and Latin gangs causing chaos and Putin and Xi among others having evolved a new form of capitalistic communism
• In 2003, Nicholas Carr stirred up our industry with his Harvard Business Review article “IT doesn’t matter” and a book similarly titled. He was similarly mocked and the reality is CIOs have become even more strategic with large digital transformations, security challenges, and now with management of hyperscaler clouds and with massive AI budgets.

In their defense, both Fukuyuma and Carr were pointing to inflection points, not “endings”. So I would read Satya with a similar lens – he is pointing to a transition point, not a wall we are about to crash into.

And Satya is only pointing to the obvious. As I wrote in From BOPS to FAANG to the Magnificent Seven enterprise application vendors and their services partners have gradually but precipitously slipped in market leadership over the last 3 decades.

Satya has certainly earned the right to opine on the trajectory of the industry. Watch this fascinating conversation with Bill Gurley and Brad Gerstner. Satya has taken the Azure cloud business from $1 to 66 billion in revenues. He has added $3 trillion to Microsoft’s market cap. His early investment in OpenAI was brilliant. He is humble enough to talk of the “winner’s curse” where competitors can sneak up, duplicate success in a heartbeat. But the really intimidating part for his competitors is towards the end when they talk about model scaling and over 60 data centers across the globe and massive capex spend of nearly $70 billion in 2025. No wonder Satya mostly names his Mag 7 compatriots (he alludes to the Mag 8 to include OpenAI) like Elon, Jensen, Sam and Mark in his comments, not any of the application vendor execs.

I have met Satya a couple of times, and would love the opportunity to take the conversation further with more of an application and business use case perspective : who would train the new world of agents, where would the training data be sourced from, where are we going to provision the humongous energy needed to run power hungry GPUs and data centers etc.

Here are 5 areas I would love to discuss with him given the changes the new US administration is likely to bring to the global economy, and based on research we did for the fiction book we have just released, The AI Analyst

• Tariffs will likely lead to more manufacturing, varied logistics and related robotics and sensors in the US. Immigration changes will lead to more automation opportunities in farming, construction and other sectors. Healthcare is going to see plenty of scrutiny. There is already a debate on H-1B visas and impact on IT outsourcing. Most of the agents today in the markets are back office and IT-centric, and mostly Western economy focused. How are they going to help in operational areas on the shop floor, the operating room, the fields etc.? How are they going to be able to help with the exotic taxes, payment protocols, customs and other compliance requirements being dreamt up by BRICS and other fast growing economies?
• Polestar, the AI/Automation company in our book whose CEO, Barry Roman disappears, is a next-gen tech vendor which combines agentic AI with humanoid robots, drones, UAVs etc. Not just a software focus. Polestar has an expansive definition of its verticals as each of the 800+ occupations the Bureau of Labor Statistics tracks. Their solutions automate limbs, eyesight and other human faculties, not just cognitive skills. They sell solutions including warehouse bots and GPUs “as a service” but are very flexible in allowing customers to scale their use up or down. Is he seeing tech vendors who can come close with the unique domain expertise and faculties needed for each of these occupations? That unique worker productivity is where the buyer ROI is acutely seeking.
• The book also has a next-gen technology buyer in a NYC financial institution, Sheldon Freres which knows how to monetize its mountains of unique vertical industry data. The tensions and shifts in economics between buyers and vendors of technology are brought out quite vividly in the book. As Patrick Brennan, the analyst character summarizes “Given how expensive GPUs and good AI talent is likely to be for the next few years, enterprises will prioritize unique products and market-insight projects, like drug discovery, mineral insights, product design advantages, trading patterns, etc. Smart customers will protect that data for themselves, train their own large language models, or LLMs, and commercialize that data asset. Vendors will continue to generate proposals, job descriptions, demand forecasts, etc. with their AI—clearly useful stuff, but not deserving significant premium pricing. We need to be associated with the first group of customers, the smart ones.” How are vendors going to adjust to a coming marketplace where their customers have the monopoly on deep domain specific training data and want to be compensated for it?
• How are we going to source energy for our massive inference, quantum and other computing while monitoring our emissions? Our utilities have barely been able to keep up with growing EV demand. Even in Germany which has invested trillions over 50 years on wind and solar, renewables barely meet half of their total energy needs. Where does he see next-gen nuclear in the mix? What about carbon capture and hydrogen harvesting? Do we need a new breed of utilities which are as ambitious about capex spend as the Mag 7 have become? At this point our utilities appear to be a major bottleneck and their infrastructure is incredibly dated.
• Technology buyers are also going to be consulting with a new breed of analysts like Patrick. His firm, Oxford Research doesn’t just produce “Golden Circles” (their version of Magic Quadrants) which a copilot called Curmudgeon helps generate. They also have labs which test products and their security vulnerabilities. Their focus is not just IT, but energy and other STEM disciplines and they have a global reach. As Patrick tells his colleagues at an Oxford offsite “COVID, the Ukraine war, climate change, and massive digital transformations have made many vertical edge applications viable—telemedicine and personalized medicine in healthcare, EV battery management and billing in utilities, intelligent returns and reverse logistics around eCommerce, direct-to-consumer and related last-mile, small-lot logistics in consumer sectors, CPQ for industrials to handle complex outcome-based pricing which bundles products’ spare parts, all kinds of monitoring and maintenance services . . . the list is virtually endless.” The conversation around AI is increasingly going to be with much savvier business executives, not just about technology speeds and feeds. As Tom Siebel of C3.ai recently said “Silicon will get commoditized. It always gets commoditized. Infrastructure will get commoditized. It always gets commoditized. What doesn't get commoditized in the long run are the applications and that's where C3 AI plays.” And to my knowledge, even Tom does not have anywhere near the depth of vertical capabilities I describe above.

Would love to get Satya’s perspective on the coming changes in our industry from his perch. I loved his demeanor in the Bill and Brad show. It was an unusually long session but he came across so comfortable, confident and articulate. A true industry leader.