Right now, there’s a lot of hate for the Cloud Czars and their leaders. There are now four such Czars, the masters of the Cloud. Apple (AAPL) has chosen not to play. (The illustration of Greek Gods lost in a modern city is from Leonardo.AI.)
The hate is directed at the Decibillionaire controlling shareholders. We hate Jeff Bezos of Amazon (AMZN), not CEO Andy Jassy. Or we hate Sergey Brin of Google (GOOG), not CEO Sunjay Pichai. We go after Bill Gates of Microsoft (MSFT), not Satya Nadella. OK, Mark Zuckerberg (Meta). You got me there.
The strategies of the Cloud Czars aren’t driven by their owners, but by their managements. They all saw OpenAI, its early success with ChatGPT, and decided that Large Language Models (LLMs) were the new “Great Game.”
The Czars are playing the new AI era as they did the Cloud Era where they were born. The idea is to pour operating cash flow into ever-larger data centers, to control what happens. The AI Bust I predicted in January was the result.
It’s not greed that’s driving these moves. It’s fear. Fear Of Missing Out is a real thing in technology, and the leaders in one era often fall apart in the next. Apple had a near-death experience in the 1990s, and Microsoft did in the succeeding decade. All the other Czars were born after the Web was spun. They’re products of that time who got the cloud right.
But we’re in a different era now and the strategies of the last era won’t work. LLMs offer no competitive moat, despite the Czars’ enormous investments. They wanted to control the future but they wound up becoming AT&T.
Where’s My Moat?
The key to success in the AI Era is software. It’s not about creating a super-human intelligence. This is not like the Internet Era.
It’s more the PC era.
To get the cash to be AT&T, the Czars dumped precisely what you need to succeed in software. People. You need people who understand database technology and who can explain what you’re doing to other people.
What we have learned with LLMs can be easily applied to smaller, specialized databases. But each one of those databases will have to be carefully taught. Even a telephone operator can’t be replicated easily. Extensive training is needed after the software is developed, not just to clear bugs but to create value. That training must have human back-up, just as a newly trained operator needs a supervisor. The training jobs are highly skilled, career-making jobs.
I’m reminded of AthenaHealth, profiled by Michael Lewis in his podcast Against the Rules. The company you see today was built by finding an expert in medical billing codes and having them train the developers. They took a women six levels down on the organization chart, lifted her up, and created billions of dollars in value.
For every skill set AI is developing, in hopes of finding productivity, there are experts deep within corporate organization charts who have that skill. Turns out the way to build great AI companiesis through finding talented people who can explain what they do in plain English, teaching the developers what they need to develop.
The Czars fired the wrong people. The entrepreneurs who find the right people, who learn from them, and use them to train software, along with developers who can understand English, will create AI moats and build the future.
