2 AI Infrastructure Stocks That are Looking Too Cheap

AI and AI infrastructure stocks have been punished, and while a bottom might not yet be in sight, I still think now might be as good a time as any to start doing some buying. As is the case with many sell-offs, oversold stocks have the potential to bounce quite sharply once the bottom is finally hit.

Of course, timing is hard on the way down and buying the dips could remain a painful thing to do. Either way, though, the following pair of names, I think, is getting cheap. And that alone makes them worth pursuing, even if negative momentum in AI accelerates.

Oracle

Oracle (NYSE:ORCL | ORCL Price Prediction) has been beyond out of favor in recent quarters. If it’s not the heavy spending on advancing its AI infrastructure efforts and the pile of debt, or the bloodbath in software stocks, perhaps it’s Michael Burry’s bearish bets on the name. Undoubtedly, Burry doesn’t seem to be pulling punches with Oracle.

With much buzz about the latest and greatest Nvidia (NASDAQ:NVDA) chips on the Vera Rubin platform, which make massive leaps over the last generation, perhaps the last-gen chips have a shorter shelf life than expected. Undoubtedly, there’s rapid obsolescence in the semiconductor scene, and if we’re talking about Nvidia, it seems to be a master at unlocking shocking year-over-year performance gains.

While there’s a fear of outdated data center builds and the potential for big write-offs to come, I’m not sure how much of the fear of the underutilized data centers is already in the stock. Burry has probably already done well with his puts. And the big upside, I think, could lie in what happens if older chips are put to work on less-intensive inference tasks for far longer.

If Vera Rubin is short in supply versus demand, perhaps the only option is to settle for Blackwell. All the while, Oracle will be buying the latest and greatest, as data centers undergo more of a steady transition to the next generation.

In any case, Oracle is moving fast in data centers, and it seems to be doing the same with agents. Cantor Fitzgerald thinks the firm is “taking a more disruptive approach to deploying AI agents.” Perhaps Larry Ellison is right in that the SaaS-pocalypse is more of a concern for other software companies. Any way you look at it, I think Oracle has heavily underestimated upside, especially at 27.7 times trailing price-to-earnings (P/E).

Amazon

Amazon (NASDAQ:AMZN), the behemoth behind AWS, might be an obvious AI infrastructure pick, but it’s one that’s worth sticking with, especially as the company spends a boatload of cash to advance its efforts. The e-commerce giant will reportedly spend a massive fortune with Nvidia on chips and more through 2027.

Combined with its upper hand in physical AI (warehouse robots, its Zoox autonomous vehicle, and its more recent Rivr last-mile robot acquisition), and I think Amazon stock is just another oversold, undervalued AI infrastructure winner that’s not appropriately priced in this moment of market-wide panic. The stock trades at 29.3 times trailing P/E. That’s close to the lowest it’s ever been. And while AWS might not be as growthy as it once was, I do think that another AI-driven second wind is on the horizon.

It’s spending big money on AI, and, in due time, I think such spending won’t seem as horrific once we get a better gauge of the returns to be had. Whether Amazon is the biggest automation play out there remains the big question. Either way, AWS continues to be the AI infrastructure giant to beat. And until Amazon pulls back on the spend, there might be no catching it.