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 actually about to pop, stop watching Nvidia or OpenAI. Look at Oracle.

Yes, the database company. The one that’s been around since the 1970s, the one that sells you enterprise software you probably hate but can’t escape. That Oracle has pivoted hard into AI infrastructure, and it’s doing it in a way that makes me nervous.

Oracle isn’t building foundation models. It’s not competing with Anthropic or OpenAI on the model side. It’s not even really a cloud provider in the traditional sense — though it’s getting into the same bare-metal rental business as CoreWeave. What Oracle is doing is betting that the future of AI runs on specialized, high-density compute clusters that companies will rent by the rack, not by the virtual machine.

And it’s a bet that could either make Larry Ellison look like a genius or sink the company.

Here’s the thing: Oracle is older than almost every AI competitor except Microsoft. Its core database business has been in slow decline for years. The pivot to AI isn’t opportunistic — it’s existential. Ellison has burned the boats. Oracle is no longer a software company that dabbles in AI; it’s an AI infrastructure company that still sells some databases on the side.

That’s a risky position. The hyperscalers — AWS, Google Cloud, Azure — can absorb AI infrastructure costs because they have massive, diversified revenue streams. If AI demand softens, they can reallocate those GPUs to other workloads. Oracle doesn’t have that luxury. Its AI business is essentially a single bet on sustained demand for GPU clusters.

And the signs aren’t great. We’ve already seen some AI startups trimming their infrastructure spend. The cost of renting H100s has dropped. If the demand curve flattens, Oracle’s going to be sitting on a lot of expensive hardware with no one to rent it to.

But Ellison has never been afraid of big bets. He’s been right before — remember when everyone laughed at Oracle buying Sun Microsystems? That turned into a massive Java licensing machine. This time, though, the stakes are higher because the timeline is compressed. AI infrastructure is a winner-take-most market, and Oracle is competing against companies with deeper pockets and more cloud-native expertise.

I’m not saying Oracle will fail. But if you want a single stock that tells you whether the AI boom is real or just another cycle of hype, watch Oracle. If their AI revenue keeps growing, the bubble might have legs. If it stalls, sell everything.

For now, Ellison is balancing a basket of eggs on his head with the OpenAI logo on it. It’s either a circus act or a magic trick. We’ll find out which soon enough.