Honestly, if you told someone five years ago that Oracle would be the hottest ticket in the AI revolution, they probably would’ve laughed. People used to think of Oracle as that "old school" database company—the one with the aggressive sales reps and the legacy software your IT department complained about.
But look at the charts lately. Oracle stock is up because they’ve pulled off one of the most unlikely pivots in tech history. They aren't just surviving; they are basically the backbone for the biggest AI models on the planet right now.
The $523 Billion Elephant in the Room
The real reason everyone is talking about Oracle right now comes down to one specific, massive number: $523 billion. That is their "Remaining Performance Obligation" (RPO) as of late 2025.
Think of RPO as a giant pile of IOUs. It’s money customers have legally committed to spend but haven’t paid yet because the services haven't been delivered. When Oracle reported their fiscal 2026 second-quarter results, that backlog had jumped by an insane 438% compared to the year before.
Why such a huge spike? Because names like Meta and NVIDIA are signing multi-billion dollar deals to use Oracle Cloud Infrastructure (OCI). While Amazon and Google were busy building general-purpose clouds, Oracle built OCI specifically for high-performance workloads. It turns out, if you want to train a massive AI model like Llama 4, you need the kind of raw, "bare metal" power that Oracle specializes in.
Why AI Startups Are Flocking to OCI
It’s not just the giants. Startups are moving there too.
Basically, OCI uses a different networking architecture than AWS or Azure. They use something called RoCE (RDMA over Converged Ethernet), which sounds nerdy, but what it actually means is that their clusters of GPUs can talk to each other faster.
In the AI world, speed isn't just a luxury; it’s the difference between a model costing $10 million to train or $50 million.
- Meta is using OCI to bolster its Llama family of AI solutions.
- OpenAI has been tapping into Oracle’s capacity to supplement what they get from Microsoft.
- NVIDIA itself is a major customer, using OCI to run its own internal AI research.
The "If You Can't Beat 'Em, Join 'Em" Strategy
For years, Oracle tried to fight the other cloud providers. It didn't work. So, Larry Ellison—Oracle’s legendary founder—did something nobody expected. He started making deals with the "enemy."
📖 Related: How Many Millions Are in 1 Trillion? The Math Behind These Massive Numbers
Now, you can find Oracle’s database services sitting directly inside AWS, Google Cloud, and Microsoft Azure. They call it "Multicloud." It’s been a game-changer. This segment grew at a staggering 817% in the last reported quarter.
Imagine you’re a big bank. You’ve used Oracle databases for 30 years, but your developers love AWS. Previously, you had a headache trying to make them talk to each other. Now, you just run the Oracle database inside the AWS console. It’s seamless. This move removed the biggest barrier for customers who wanted to stay with Oracle but move to a modern cloud environment.
The Nvidia Blackwell Factor
Oracle was also one of the first to get their hands on the newest chips. They’ve already started deploying over 96,000 NVIDIA Grace Blackwell GB200 units.
Being first to market with the fastest hardware means they get the biggest customers. It’s a simple cycle. They spend billions on chips, AI companies rent those chips, and the stock price moves up because the revenue potential is finally becoming reality.
What Most People Get Wrong About the Debt
If you look at the bears—the people betting against the stock—they always point to the debt. Oracle is spending money like crazy. Their capital expenditure (capex) for fiscal 2026 is projected to hit $50 billion.
That’s a lot of zeros.
Some analysts, like those at Scotiabank, have warned that this massive spending might hurt short-term profits. But here’s the thing: most of that equipment is already spoken for. Oracle isn't building "spec" houses; they’re building custom mansions for customers who have already signed the lease.
As Safra Catz, Oracle's CEO, put it during a recent earnings call, the demand is so high they literally can't build data centers fast enough. That’s a "problem" most companies would kill for.
Is Oracle Actually Overvalued?
This is where it gets tricky.
If you look at the Price-to-Earnings (P/E) ratio, Oracle sits around 37x. That’s higher than the software industry average of about 32x. Some models, like the Discounted Cash Flow (DCF) analysis from Simply Wall St, suggest the stock might be trading at a premium of about 20% over its "intrinsic" value.
But valuation in a bull market—especially an AI bull market—is more about momentum and future earnings. Analysts at Deutsche Bank and Mizuho are still bullish because they see that $523 billion backlog as a guaranteed revenue stream for the next five years.
📖 Related: When Will Elizabeth Holmes Be Released? What Most People Get Wrong
Recent Wins and Dividends
- Oklo and Meta Deal: A recent energy deal between Oklo and Meta benefited Oracle stock because it ensures Meta has the power to keep running its AI on Oracle's servers.
- Dividend Payouts: Oracle recently declared a quarterly dividend of $0.50 per share, showing they still have enough cash flow to keep investors happy while they build out their AI empire.
What Really Happened With the "Ampere" Sale
One detail that slipped under the radar for many was Oracle selling its stake in Ampere, the chip company. They booked a $2.7 billion pre-tax gain from it.
Larry Ellison was blunt about why: they want to be "chip neutral."
By not being tied to one specific chip designer, Oracle can pivot to whoever has the best tech—whether it’s NVIDIA, AMD, or someone new. This agility is a huge part of why Oracle stock is up. They aren't betting on one horse; they're building the track.
The Road Ahead: What to Watch For
If you're looking to play this move, don't just watch the stock price. Watch the "Cloud Infrastructure" revenue growth. That's the engine. Last quarter, it was up 68%. If that stays above 50%, the stock likely has more room to run.
Also, keep an eye on the "Sovereign AI" trend. Governments in places like Abu Dhabi and the EU are starting to want their own clouds that don't send data back to the US. Oracle is the only major player offering a "Dedicated Region" where they basically build a miniature version of their cloud inside a government's own data center. It's a niche, but it's a high-margin one.
Actionable Insights for Investors
If you're wondering how to handle the current Oracle surge, here are a few practical steps to consider:
- Check the Backlog-to-Revenue Ratio: Don't just look at the earnings. Look at the RPO. If the backlog is growing faster than the revenue, it means there’s a "bottleneck" in building data centers. That's actually a good sign of future demand.
- Monitor Capital Expenditure: If Oracle suddenly stops spending $50 billion a year, that might be a sign that the AI gold rush is cooling off. For now, high capex is a sign of confidence.
- Diversify Within Cloud: Oracle is a great "pick and shovel" play for AI, but it's still volatile. If you're heavy on Oracle, make sure you're balanced with the hyperscalers like Amazon or Google, who provide the consumer-facing AI apps.
- Watch Interest Rates: Since Oracle carries significant debt to fund this expansion, higher-for-longer interest rates can eat into their margins more than they would for a cash-rich company like Microsoft.
The "old" Oracle is gone. The new Oracle is essentially a massive, high-speed AI factory. As long as the world keeps demanding more tokens and more training data, the wind is firmly at their back.