Oracle, once considered a stalwart of enterprise software, now finds itself at the epicenter of the artificial intelligence (AI) infrastructure boom—and the financial risks that come with it. As the company prepares to report its third-quarter earnings after market close on March 10, 2026, investors and analysts are watching closely. The results will serve as an unofficial litmus test for the entire AI trade, revealing whether the massive bets placed on cloud and AI infrastructure can deliver real, near-term returns—or if they’re simply building a mountain of promises on shaky ground.
According to CNBC, Oracle’s Remaining Performance Obligations (RPO)—the total value of contracts signed but not yet fulfilled—reached a staggering $523 billion in Q2 FY2026. That’s a 438% jump year-over-year, representing a near-quadrupling of committed but undelivered work. Yet, while this backlog signals enormous potential, the market has become laser-focused on what Oracle is actually collecting today. The company’s quarterly revenue for the same period was $16.06 billion, missing analyst estimates by about 5%. The result? Oracle stock tumbled 13% on earnings day, capping off a brutal six-month stretch that saw shares plunge by 54% despite a 3.4% revenue increase and a 20% rise in net income margin, according to Trefis.
This disconnect between bookings and revenue sits at the heart of Oracle’s current dilemma. Bookings are promises for the future, but revenue is cash in hand. Right now, Oracle is spending enormous sums—capital expenditures in just the first half of FY2026 reached $20.54 billion, up 250% year-over-year—to build data centers and infrastructure for AI workloads. Free cash flow turned negative in Q1, falling to -$362 million. This is the so-called “J-curve” effect, a concept borrowed from private equity and infrastructure investing: you spend heavily upfront, cash flow dips sharply, and then—if the thesis holds—revenue eventually catches up and surpasses the investment.
But the risk is real: what if that revenue catch-up takes longer than expected, costs more, or never materializes? For Oracle, the pressure is coming from multiple directions. The company now carries more than $100 billion in debt and has reportedly planned cuts of 20,000 to 30,000 jobs to free up cash. According to TD Cowen, such a reduction in workforce could drive $8-10 billion of incremental free cash flow. Other cost-cutting measures, including divestitures and vendor financing deals, are also being considered.
Adding to the uncertainty, Oracle announced a $50 billion financing plan in early February 2026, combining debt and equity to fund its ambitious AI infrastructure buildout. The latest debt raise included a $5 billion convertible preferred offering and about $25 billion in senior notes at different maturities, a deal that was oversubscribed—suggesting strong demand, at least for now. However, Oracle’s 5-year credit default swaps have widened, a clear sign that bond investors are becoming skeptical about the company’s ability to maintain its investment-grade credit rating, which currently sits just two notches above junk status, according to CNBC.
One major source of optimism—and anxiety—is Oracle’s $300 billion deal with OpenAI. The market has become hypersensitive to any news tied to this partnership. Last September, news of the deal sent Oracle’s stock soaring by 35%, its biggest intraday gain since 1992. But the euphoria was short-lived. In late fall, Oracle’s decision to raise significant debt fueled concerns about its balance sheet. More recently, talks to expand Oracle’s deal with OpenAI in Abilene, Texas, fell through. However, a source familiar with the matter told CNBC that Oracle’s commitment to deliver eight sites to OpenAI remains on track and on schedule. OpenAI’s executive Sachin Katti confirmed on X that while the company considered expanding in Abilene, it chose to put additional capacity in other locations, including a new site in Wisconsin where the first steel beams went up just this week.
Oracle’s position as a central player in the AI infrastructure race is reinforced by direct commitments from Meta, NVIDIA, and others. Yet, the company’s massive capital expenditure guidance—now raised to approximately $50 billion for FY2026—has created what Trefis calls an “AI Capex Overhang.” The market is now pricing in the risk of multi-year negative free cash flow and margin compression, overshadowing operational growth in the cloud business. Meanwhile, Oracle’s legacy software revenues contracted by 3% in Q2, signaling that gains in cloud are being partially offset by declines in traditional on-premise software.
The company’s financing partner, Blue Owl Capital, has also shown signs of strain. In February 2026, Blue Owl halted redemptions at a private credit fund aimed at retail investors. Its stock has fallen roughly 34% year-to-date and 47% over the past year. When the entity financing the buildout starts restricting liquidity, it’s a meaningful signal about the sustainability of these mega-deals.
The big question facing Oracle investors is whether they have the time horizon and risk tolerance to wait for the J-curve to resolve in Oracle’s favor. The analyst consensus price target of $257.29 is premised entirely on the expectation that Oracle will eventually convert its $523 billion backlog into profitable revenue. CEO Safra Catz has projected Oracle Cloud Infrastructure (OCI) revenue scaling from $18 billion in FY2026 to $32 billion, $73 billion, $114 billion, and $144 billion over the following four fiscal years, with much of that revenue already locked in the RPO. If actual IaaS revenue is tracking toward that $18 billion target, the thesis remains intact. But if it lags, the gap between backlog and reality will only widen.
It’s important to note that not all companies in the AI stack are in the same boat. NVIDIA, for example, is the supplier getting paid now. The chipmaker reported Q4 FY2026 revenue of $68.13 billion, up 73% year-over-year, with data center revenue of $62.31 billion driving most of that growth. Oracle, on the other hand, is the infrastructure builder spending now and hoping to get paid later. For investors, understanding which companies are at which stage of the J-curve is more useful than simply betting on the AI theme as a whole.
For those holding AI-themed funds or ETFs, the “canary in the coal mine” framing is a prompt to look under the hood. A fund that holds both NVIDIA and Oracle is exposed to both sides of the risk curve—one company printing cash, another burning it. As borrowing costs remain elevated (the 10-year Treasury yield sits at 4.13%), the cash flow question for Oracle grows more urgent.
Ultimately, the practical takeaway is to apply the bookings-versus-revenue test to any AI infrastructure company under consideration. When a company announces a large contract or backlog, investors should ask: When does this convert to revenue? What must be spent before that happens? And what’s the financing plan if cash flow goes negative in the interim?
With Oracle’s next earnings report imminent, the market’s verdict on the company’s J-curve experiment—and, by extension, the sustainability of the AI infrastructure boom—may soon become much clearer. For now, the tension between promise and payoff defines Oracle’s journey, and the outcome will shape not just its own future, but the broader narrative of AI’s economic reality.