Recently, I’ve been looking at the US stock AI infrastructure sector and found that Oracle (ORCL) has a particularly interesting logic.

Many people’s understanding of Oracle still stays at “an old company that makes databases,” but that’s just the surface. Oracle’s core is one thing: helping enterprises manage data. Database software, cloud infrastructure, cloud applications—simply put, “store data, manage data, use data.” Once this system is in place, it costs a huge amount for companies to switch providers, so customer stickiness is extremely high.

But real change happened after the AI wave arrived. Oracle started shifting from traditional software licensing models to cloud subscriptions, with customers moving from outright purchases to monthly or yearly payments. More importantly, AI requires massive data processing and efficient computing power, which is exactly Oracle’s strength. It is transforming from a “database company” into an “AI infrastructure provider.”

The latest financial data directly refutes market skepticism. Third-quarter revenue was $17.2 billion, up 22% year-over-year, hitting a 15-year high. Cloud revenue grew 44% annually to $8.9 billion, accounting for over half of total revenue for the first time. Even more impressive is OCI (Oracle Cloud Infrastructure), which grew 84% year-over-year to $4.9 billion, clearly reflecting exploding demand for AI computing power.

The most shocking figure is RPO (Remaining Performance Obligation), which surged to $553 billion, up 325% year-over-year. This means Oracle has locked in high-growth revenue for the coming years, with extremely high visibility. Most of these orders come from tech giants like OpenAI and Meta, with customers even providing their own GPUs; Oracle only needs to handle deployment and operation.

The stock performance also confirms this shift. In September last year, after news of OpenAI orders, the stock soared 36%. Although it later pulled back significantly, it recently rebounded strongly again. Analyst consensus ratings are “Strong Buy,” with an average target price over $245.

But it’s important to note that Oracle also faces significant challenges. To support these huge orders, capital expenditure has skyrocketed to $50 billion, leading to consecutive quarters of negative free cash flow. Operating profit margins might decline from 72% to 54%. Additionally, over-reliance on a few large clients like OpenAI poses risks. In the cloud market, Oracle’s market share is only 3%, far behind AWS and Azure.

Honestly, investing in Oracle isn’t just about whether it’s a good company, but about what the market is willing to assign to it. If the market views it as an AI infrastructure company, its valuation will significantly increase. If it’s still seen as a traditional software company, the upside is limited. That’s the real core of the trade.

Taiwan’s supply chain is also benefiting from this wave. Foxconn is a major supplier of Oracle AI servers, with Wistron, Quanta, and Mitac also entering the supply chain. Lianjun directly supplies optical communication modules, seen as a hidden winner. TSMC is also indirectly benefiting from increased chip demand.

For different types of investors, strategies vary. Short-term traders can focus on financial report volatility and AI collaboration news, as Oracle reacts clearly to these events. Swing traders should track cloud revenue growth and RPO conversion progress. Long-term investors need to evaluate whether Oracle can successfully transform into an AI infrastructure provider.

My suggestion is not to rush into buying yet. You can observe with a demo account for two weeks, watching Oracle’s stock reaction after earnings, OCI revenue growth, and market sentiment toward AI infrastructure. Once you get a feel for the stock’s volatility rhythm, then try with small capital. The opportunity for Oracle concept stocks definitely exists, but rhythm is more important than anything else.