Can Oracle Finally Silence the AI Doubters?

Oracle heads into Tuesday’s earnings report as one of the most important tests of the AI infrastructure trade, not because investors doubt demand, but because they are still wrestling with how that demand will be financed and monetized. Oracle has increasingly become ground zero for the market’s concern that the AI buildout may require enormous upfront capital, years of cash burn, and a level of execution that leaves very little room for mistakes. That is why this report matters so much. Oracle is no longer being judged simply as a legacy software company with a fast-growing cloud business. It is being judged as a capital-intensive AI infrastructure provider, and investors want proof that the backlog is real, the revenue is arriving, and the financing plan is enough.

Wall Street is looking for fiscal third-quarter revenue of about $16.9 billion and adjusted earnings per share around $1.69 to $1.70. Oracle’s own guidance called for total revenue growth of 16% to 18% in constant currency, or 19% to 21% in U.S. dollars, with non-GAAP EPS of $1.70 to $1.74 in reported terms. The company also guided cloud revenue growth of 40% to 44%. Those are strong numbers on paper, but this quarter is much more about the composition of growth than the headline results.

The key number to watch is Oracle Cloud Infrastructure. In the prior quarter, OCI revenue grew 66% year over year to $4.1 billion, while total cloud revenue rose 33% to $8.0 billion. GPU-related revenue jumped 177%, multicloud consumption surged 817%, dedicated region and Alloy consumption grew 69%, and Oracle delivered 50% more GPU capacity than in the previous quarter. Remaining performance obligations ended last quarter at an eye-popping $523.3 billion, up 433% year over year, while short-term RPO accelerated to 40% growth from 25% in the prior quarter. That RPO figure is the headline bulls point to when they argue Oracle has one of the clearest demand runways in the entire AI infrastructure ecosystem.

This quarter, analysts appear to be circling OCI growth in the mid-80% range, with several firms flagging roughly 86% as an important bogey. Investors will also be watching total cloud growth versus the company’s 40% to 44% guide, net RPO additions, and any update on when the backlog actually converts into recognized revenue. In the last report, management said the vast majority of new bookings related to opportunities where near-term capacity was available and that the added backlog should contribute an incremental $4 billion of revenue in fiscal 2027. If management can show that this backlog is moving forward faster than feared, that would go a long way toward calming nerves.

But that is where the debate gets tricky. Oracle’s backlog is enormous, yet bears argue that fulfilling it may require even more capex than the market is comfortable with. Last quarter, capex hit $12 billion, free cash flow was deeply negative, and Oracle raised its fiscal 2026 capex plan by $15 billion to around $50 billion. Since then, the company has completed a major bond issuance, announced plans for up to $20 billion of common equity, and outlined a broader $45 billion to $50 billion fundraising plan. The buyback announcement matters too, because it signaled management still wants to project confidence in long-term value even while asking investors to stomach a much more capital-hungry business model. Still, the market has not exactly thrown confetti. Investors remain concerned about whether this is just the first round of financing rather than the last.

That is why the call will matter even more than the print. Investors need clarity on capex needs, margin timing, and the path to acceptable returns on all this spending. Oracle has argued that it can reduce borrowing needs through alternative structures, including customer-supplied chips and vendor leasing arrangements. Management has also said it expects to need less, if not substantially less, than the $100 billion some analysts have modeled for the full buildout. The problem is that many investors still want to see it before they believe it. The recent widening in Oracle’s CDS spread became a shorthand for that anxiety, reflecting market unease around leverage and execution risk even after the successful bond sale.

OpenAI remains another central issue. Some analysts believe the market is too focused on OpenAI concentration risk and missing the broader OCI opportunity beyond that single customer. Others think it is impossible to ignore given how much of the backlog may be tied to OpenAI-related demand. Reports that existing plans remain on track have helped somewhat, but headlines around collapsed talks on a planned expansion site in Texas reinforced the idea that Oracle’s AI story still depends on massive, complex projects coming together without major delays. Investors will want updates on the Abilene site, additional go-lives, the pace of capacity delivery, and whether the company is seeing real revenue contribution from previously announced contracts.

Margins are another crucial piece of the puzzle. Barclays and others have warned that gross margins and EPS may face headwinds as lease expense timing and upfront investments hit before the full revenue benefit is recognized. Management has previously discussed a 30% to 40% gross margin profile over the life of AI contracts, but investors need reassurance that Oracle can get there without endless dilution or more debt. Any commentary suggesting that OCI margins are stabilizing, or that the period of negative timing effects is relatively short, could help the stock.

Shares have been crushed, falling roughly $200 from their September high near $350, and investors are clearly hunting for a bottom. Relief could come from a few places. First, OCI growth that clearly beats expectations, especially if it lands closer to the high end of the Street’s bullish range. Second, strong net RPO additions alongside evidence that short-term RPO is still accelerating. Third, commentary that financing needs for calendar 2026 are now largely addressed. Fourth, any sign that margins are holding up better than feared despite the capex ramp. And finally, confidence that Oracle’s cloud targets are achievable even without heroic assumptions around OpenAI.

The bottom line is that Oracle’s earnings are not just a quarterly report. They are a referendum on whether the market will keep funding one of the largest AI infrastructure buildouts in corporate America. Demand is clearly there. The question now is whether Oracle can prove that the backlog is becoming revenue, the revenue is becoming profits, and the financing plan is not a moving target. If management can deliver that message cleanly, the stock may finally get the relief rally investors have been waiting for. If not, the market may conclude that Oracle is still building first and explaining later, which is rarely a popular strategy on Wall Street.