Oracle: the AI cloud nobody is talking about enough

From databases to AI infrastructure – a $664 billion backlog, the Tencent deal and the question that decides everything: can Oracle turn demand into cash?

It didn't start with AI. It started with databases.

For decades, Oracle built the infrastructure behind some of the world's largest companies. Banks. Airlines. Governments. Telecom. Global enterprises.

But something has changed. Oracle is no longer just building the systems that store the world's data. It is building the infrastructure that could power the next generation of AI.

And the deeper you look, the more interesting the story becomes.

$664 billion

This is the number that made us look at Oracle differently. At the end of its latest quarter, Oracle had:

That's contracted future business that has not yet been recognized as revenue. A year earlier, the number was $455 billion. Now? $664 billion.

Oracle also booked more than $30 billion in new AI cloud contracts during the quarter. And Cloud Infrastructure revenue jumped:

This isn't an old software company trying to catch the AI wave. Oracle is building the infrastructure underneath it.

Then came Trump and Xi

In late September, Donald Trump hosted Chinese President Xi Jinping in Washington as the world's two largest economies continued to navigate trade, technology and strategic competition.

Behind the headlines sits another battle: AI. The United States wants to protect its technological lead. China wants to build its own AI capabilities. And both sides need the same thing:

COMPUTE.

Advanced AI chips have become one of the most strategically important resources in the global economy. And U.S. export restrictions make access to the most advanced AI hardware increasingly complicated for Chinese companies.

Then something very interesting happened. Just days after the Trump–Xi meeting, reports emerged that Tencent had agreed to lease access to approximately:

The reported deal is estimated at around $7 billion over five years. Reuters, citing the Financial Times, reported the agreement on September 30. We have not seen the terms confirmed by Oracle or Tencent.

But if the numbers are accurate, it would represent Tencent's largest overseas AI infrastructure lease to date. And suddenly, the Oracle story gets much bigger.

Think about what that means

China needs compute. The U.S. controls much of the world's most advanced AI hardware. AI companies need data centers. And Oracle has built a global cloud infrastructure where that compute can be accessed.

Oracle doesn't necessarily need to manufacture the chip. It can sell access to the compute. That's a very different business.

NVIDIA sells the accelerator. Oracle sells the infrastructure around it.

And as AI models become larger, compute becomes one of the world's most valuable resources.

OpenAI

Then there is OpenAI. Oracle is a major infrastructure partner in the Stargate buildout. Oracle and OpenAI have agreed to develop up to 4.5 GW of additional AI data-center capacity.

This isn't "let's build another data center." This is industrial-scale AI infrastructure. And that's where Oracle starts to look very different from the company many investors still think they own.

The Oracle AI machine

Oracle is also guiding for at least $90 billion in total revenue for FY2027. That is a very different growth profile from the old Oracle story.

And then there is the real moat

This is the part of the story that AI hype can make investors overlook. Oracle has something many new AI companies don't:

DATA.

The companies using Oracle already have some of their most valuable business data inside Oracle databases. And AI needs access to that data. That creates a potentially powerful chain:

Oracle isn't just trying to sell compute. It already sits on top of decades of enterprise data and relationships. That could become incredibly valuable in the AI era.

The multicloud bet

And Oracle is making another interesting move. It isn't necessarily asking companies to leave AWS, Azure or Google Cloud. It wants Oracle to exist alongside them.

• AWS
• Microsoft Azure
• Google Cloud
• Oracle Cloud

That matters. Large enterprises don't want to rebuild their entire technology stack every time the cloud market changes. Oracle can potentially become the bridge between different cloud environments. And as companies use more AI models, more AI services and more types of compute, that role could become even more important.

But here's the catch

This isn't a risk-free AI rocket. Building AI infrastructure is extremely expensive. Oracle spent approximately:

Free cash flow was approximately −$5 billion. Oracle is spending enormous amounts of money today to build infrastructure that it expects to monetize for years to come. That's the biggest question in the entire story.

CAN ORACLE TURN ALL THIS DEMAND INTO CASH?

This is where I think the story gets interesting

I don't think Oracle needs to become the next NVIDIA. It doesn't have to. NVIDIA owns the accelerator layer. Microsoft has Azure. Amazon has AWS. Google has its AI ecosystem.

Oracle needs to own something different: the infrastructure layer connecting enterprise data, cloud and AI compute. And if AI demand continues growing at anything close to today's pace… the need for compute will be enormous.

Think five years ahead

AI agents working around the clock. Companies automating thousands of processes. AI models requiring enormous amounts of inference. China needing compute. The U.S. needing compute. Europe needing compute. OpenAI needing compute. And enterprises needing to connect their data to AI.

Who benefits? Not only the company making the chip. Not only the company building the AI model. But the company providing the infrastructure between them.

THAT IS ORACLE'S BET.

The bull case

The bullish Oracle thesis is straightforward:

• AI demand keeps accelerating
• OCI keeps growing at triple-digit rates
• Oracle continues signing massive AI contracts
• RPO keeps expanding
• Data-center capacity keeps coming online
• OpenAI and other major customers keep scaling
• Multicloud becomes increasingly important

And eventually:

If that happens, the market may stop valuing Oracle as an old enterprise software company. It could start valuing Oracle as an AI infrastructure compounder.

The bear case

But the risks are real. AI infrastructure requires enormous capital. Free cash flow is currently negative. Oracle needs continued access to financing.

Competition is brutal: AWS, Microsoft Azure, Google Cloud, CoreWeave, NVIDIA, AMD — and specialized AI infrastructure companies such as Cerebras.

There is also customer concentration. A meaningful portion of Oracle's enormous backlog is tied to a relatively small number of very large customers. If AI spending slows, Oracle could feel it. The infrastructure still has to be paid for.

The number I will watch

Forget the hype for a moment. Watch:

FREE CASH FLOW.

Revenue can explode. RPO can hit hundreds of billions. AI contracts can create spectacular headlines. But eventually, infrastructure has to produce cash. The Oracle thesis gets much stronger if we see:

• OCI growth staying strong
• RPO continuing to expand
• RPO converting into revenue
• Data-center utilization increasing
• CapEx becoming more efficient
• Free cash flow turning positive

That's when the story changes.

Oracle vs the hyperscalers

Oracle doesn't need to replace AWS. It doesn't need to beat Azure. It doesn't need to become Google. It may simply need to become infrastructure that enterprises cannot ignore.

The bigger AI story

There is an even bigger trend underneath all of this. AI is moving from:

Oracle sits directly in the middle of several of these layers. NVIDIA supplies compute. AMD challenges the accelerator market. Energy companies supply the power. Data-center companies build the physical infrastructure. And Oracle is trying to become one of the companies connecting it all through the cloud.

Grabit take

ORACLE IS NOT THE NEXT NVIDIA. AND THAT'S EXACTLY WHY THE STORY IS INTERESTING.

NVIDIA dominates the accelerator layer. Oracle is attacking a different layer: the AI cloud. Oracle already has enterprise customers, databases, OCI, multicloud, AI contracts, OpenAI exposure — and now massive AI infrastructure investments.

At the same time, the global race for AI compute is becoming increasingly geopolitical. The U.S. China. Trump. Xi. NVIDIA. AMD. OpenAI. Tencent. And right in the middle: Oracle.

That's why I don't think investors should look at Oracle as "an old database company." That's yesterday's Oracle. The new question is:

CAN ORACLE BECOME ONE OF THE MOST IMPORTANT INFRASTRUCTURE COMPANIES OF THE AI ERA?

That's what makes ORCL interesting to me. Not what Oracle was. But what Oracle is becoming.

THE OLD GIANT IS BUILDING THE AI CLOUD.