Early in my career I was a Novell system administrator. Back then, a big part of the job was just making technology work: installing servers, configuring networks, managing directories, troubleshooting infrastructure, keeping systems up.
I’ve watched several major shifts reshape how businesses operate since then. Novell gave way to Active Directory. Physical servers became virtual machines. Virtualization evolved into hyperconverged infrastructure. Infrastructure moved into the cloud. Now we’re in the middle of another one with AI.
Having lived through all of them, I’ve come to believe the most important outcome of a technology revolution is rarely the technology itself. It’s what the technology lets us stop spending time on.
Every wave removes a constraint
Launching a new application used to start with capacity planning, procurement, hardware delivery, networking, and weeks of coordination before the application team could really begin. Virtualization cut that to minutes. Cloud went further still: we stopped needing to own infrastructure at all, and compute, databases, and dozens of other capabilities became available on demand.
The question changed from how do we build the infrastructure to what can we build with it. AI is moving that same boundary again, and further.
AI is compressing the path from idea to outcome
Businesses have always had more ideas than they could execute. The constraint was never imagination, it was execution: every idea competing for people, budget, engineering time, integration effort, and testing. Leadership had to ration which opportunities were even worth pursuing. That calculus is what’s changing now.
Software development is just the most visible example. The same compression is showing up in customer service, operations, analytics, product development, and research. That’s why I think the AI conversation is too focused on productivity. Productivity is real, but it’s the first-order effect. The bigger shift is that AI makes more ideas economically possible in the first place.
The bottleneck is moving
When execution gets cheaper and faster, the bottleneck moves with it. Technology leaders used to spend most of their time on can we build this, do we have enough engineers, can we integrate it, how long will it take. Those questions haven’t disappeared, but they’re no longer the dominant ones.
The harder questions now are which problem is actually worth solving, which customer experience deserves the investment, which process is worth reinventing, and where the next bet should go. That’s a different kind of leadership challenge: less about protecting scarce technical capacity, more about allocating attention, capital, and talent toward what actually matters.
Cheaper experimentation raises the stakes on judgment
There’s a catch. When experimentation gets cheap, organizations can chase more opportunities, but they can also generate more noise. Speed cuts both ways. A team can ship something fast and still be building blind, without ever pinning down what value they’re actually trying to prove or what field they’re testing it in. A prototype that comes together in a week hasn’t earned anything if nobody checked whether it solves the right problem.
If a team can build almost anything, knowing which thing is worth building becomes the scarce skill, not a nice-to-have. That takes more than technical knowledge; it takes an understanding of customers, economics, operations, and risk.
For technology leaders, that’s the real job now. It isn’t to bring AI into the organization. It’s to understand where AI changes the economics of the business, and where that opens up a better place to invest.
The operating model has to change too
There’s a second implication that’s easy to miss. If execution gets cheaper and faster but the organization around it doesn’t change, the bottleneck just reappears somewhere else. A company can’t capture the value of AI while product, engineering, and business teams still work through slow handoffs and rigid annual planning cycles.
The organizations that get the most out of AI won’t just adopt better tools. They’ll shorten decision cycles, tighten feedback loops, and make it easier to move resources toward what’s actually proving valuable. That’s an operating-model problem, not an AI implementation problem.
Complexity doesn’t disappear, it moves
Every technology wave arrives promising simplification, and reality is always messier than the pitch. Virtualization simplified hardware management but created new operational headaches. Cloud removed the burden of owning infrastructure but introduced new questions about architecture, security, and cost. AI will follow the same pattern: models will get easier to access, agents easier to build, development faster, but the hard questions move upward. How should AI be governed. Which decisions stay with people. How do we measure the business value instead of just the activity. How do we make sure faster experimentation doesn’t just mean faster waste.
The real measure of progress
The technology itself has changed almost beyond recognition across my career, but the direction hasn’t: each generation removes another layer of friction between an idea and its execution, and shortens the distance between concept and realized value. Each one lets technology leaders spend less time on the mechanics of delivery and more time on the decisions that actually shape the business.
That’s the real promise of AI, and it’s the opposite of what a lot of the hype suggests. Technology isn’t becoming the strategy. As it gets easier to access, the quality of our judgment matters more, not less. The question is no longer just how fast we can build something. It’s what’s worth building, and where we place the next bet.