AI today feels like the dot-com boom on steroids.
The pattern is almost identical, only faster, louder, and with more zeroes.
In 1999, we had people registering random domain names and raising millions because they had a “web business plan.” No product, no revenue, just a buzzword and a slide deck. Sound familiar?
Now it’s AI. Every company suddenly “powered by AI.” Every VC deck dripping with references to “transformative machine learning.” Every incumbent rushing to sprinkle “AI-enhanced” across their website like it’s SEO in 2007.
The parallels run deep.
The late ’90s were about potential. The internet could change everything, but the pipes weren’t ready. Bandwidth was terrible, infrastructure didn’t exist, and most consumers were still on dial-up. The big vision was right, but the stack was wrong.
Today’s equivalent is data. Everyone’s building on shaky data foundations. You can’t run great AI on garbage data any more than you could stream Netflix over a 56k modem.
And just like back then, there’s a quiet second wave building.......one that will outlast the hype.
The first wave is crowded with consumer gimmicks: AI-generated headshots, résumé writers, chatbots that can tell jokes. Cute, but forgettable. The second wave, the one that matters, will be about infrastructure, tooling, and vertical integration.
The winners won’t look like tech companies.
They’ll look like insurance firms that use AI to price risk with real-time accuracy.
Manufacturers that use AI to predict downtime before a machine breaks.
Retailers that automate pricing decisions based on millions of micro-signals.
Hospitals that integrate AI diagnostics directly into clinical workflow.
And critically, they won’t sell “AI.” They’ll sell outcomes.
Think of Amazon in 2002. Everyone thought they were a bookstore. They were actually building the world’s most advanced logistics and data infrastructure. Or Salesforce in 2004..........while everyone else was building on-premise software, they built the SaaS rails that every company now runs on.
That’s what the next decade of AI will look like. Quiet infrastructure plays that make everything else possible.
Most of today’s “AI startups” will vanish......not because AI failed, but because they mistook the technology for the product. The eventual giants will use AI the way we use electricity: invisibly, everywhere, and never as the selling point.
Right now, we’re still in 1998. The noise is deafening, the valuations are absurd, and the real builders are heads-down, laying the pipes for the next 20 years.
History doesn’t repeat, but in tech, it rhymes so loudly you can hear it through the servers.