Remember when everyone said software was eating the world? Well, plot twist: AI is now eating the software.
For the past 15 years, Software-as-a-Service (SaaS) was basically a money printer. Build a decent dashboard, slap a monthly fee on it, and watch the cash roll in as companies hired more people who needed more “seats.” It was beautiful in its simplicity.
But here’s the thing nobody saw coming: AI doesn’t need a seat at the table.
Think about it. If one AI agent can do the work of five junior analysts, companies don’t just cut payroll—they cut software licenses too. Why pay $70 per month per employee for a workflow tool when an AI can just… do the workflow?
The Middle Is Getting Squeezed
Most SaaS companies are basically fancy middlemen between humans and databases. They make pretty buttons and colorful charts so people can click around and feel productive. But AI models can now talk directly to databases without needing a shiny interface.
It’s like having a translator who suddenly realizes both parties speak the same language. Awkward.
Companies selling to small and medium businesses are feeling this first. Think LegalZoom (LZ) or Unity Software (U)—their customers are exactly the type to replace human workers with AI. Meanwhile, the big enterprise players like Palantir (PLTR) might survive because they’re so deeply embedded in mission-critical systems.
But that middle layer? It’s getting flattened faster than a pancake.
The Creation Game Is Changing Too
Google’s Project Genie just showed us something wild: you can now generate entire video games from text prompts. If small teams can build production-grade software in days instead of months, why would anyone pay premium prices for legacy tools?
The moat around “hard to build” software just got drained.
So Where’s the Money Going?
While software companies are having an existential crisis, hardware companies are throwing parties. Nvidia (NVDA) and Micron (MU) are printing money because AI is hungry—hungry for chips, memory, and computing power.
The real opportunity? AI memory bottlenecks. These AI models need massive amounts of high-speed memory to function, and we’re running short. Companies like Seagate (STX) have already rallied, but this is just the beginning.
As AI moves from the cloud to your phone, car, and toaster, the demand for specialized hardware will explode. We’re looking at a hardware-first supercycle that could make the dot-com boom look quaint.
The Bottom Line
We might have a 12-24 month window where infrastructure companies surge before the broader economic disruption really kicks in. The question isn’t whether AI will reshape software economics—it’s whether you’ll be positioned with the companies building the machines or the ones being replaced by them.
The Luddites saw the writing on the wall in 1812. The question is: do you?