Oracle’s AI gamble is either genius or a disaster — and we’re about to find out

If you want to know whether the AI bubble is bursting, there’s only one publicly traded company that will tell you: Oracle.

Yes, the database company. The one you probably last thought about when you were wrestling with some enterprise CRM implementation back in 2015. But here we are in 2026, and Larry Ellison has burned the boats.

Oracle has pivoted to AI, but not in the way you’d expect. It’s not building foundation models like OpenAI or Anthropic. It’s not quite a neocloud, though it’s entered the same bare-metal business as CoreWeave. It’s a software-as-a-service company that has made an audacious bet on a very specific future version of AI — while its traditional business has been quietly declining.

What makes Oracle interesting as a bellwether is its age. It’s significantly older than any of its AI competitors, save Microsoft. That means it has real legacy revenue, real enterprise customers, and real constraints. When Oracle makes a move, it’s not venture capital play money — it’s a bet on actual market demand.

And the bet is huge. Oracle has been buying up Nvidia GPUs like they’re going out of style, building out data centers, and signing multi-year contracts with AI startups. The company’s stock has been volatile, swinging with every AI earnings report and every new model release.

But here’s the thing nobody’s saying out loud: Oracle’s AI pivot is a bet that the current AI infrastructure buildout will continue at this pace for years. If demand softens, if the model providers consolidate, if the hyperscalers start building their own chips in earnest — Oracle gets squeezed.

The image of Ellison with a basket of eggs on his head, sporting the OpenAI logo, is apt. He’s balancing a lot of fragile things at once. The database business is still printing money, but it’s a declining asset. The cloud business is growing, but it’s competing with AWS, Azure, and GCP. The AI business is the wild card.

I’ve been watching this space for years, and I’ll say this: Oracle’s move is either going to look like genius or a spectacular miscalculation. There’s no middle ground. If the AI infrastructure buildout continues at its current pace, Oracle’s early investments in bare-metal GPU clusters will pay off handsomely. If it doesn’t, they’ll be left with a lot of expensive hardware and a shrinking legacy business.

The next few quarters will tell us a lot. Watch Oracle’s earnings calls. Watch their capital expenditure guidance. Watch their customer churn. That’s where you’ll see the real story of AI’s infrastructure demand play out.

And if you’re looking for a single stock that tells you whether the AI bubble is real or not, Oracle is probably it. Not because they’re the biggest or the most innovative, but because they have the most to lose.