Oracle (NYSE: ORCL) is closing out 2025 with the kind of volatility normally reserved for pure-play AI startups, not a 47-year-old enterprise software heavyweight. Shares are down about 30% so far this quarter, putting the stock on pace for its steepest quarterly decline since 2001, according to market reporting that tracked the move against Oracle’s early-2000s drawdowns. [1]
The selloff has been fueled by a single, recurring question: Can Oracle finance—and execute—an AI data-center buildout big enough to meet demand without crushing its cash flow and balance sheet? That concern has only intensified as Oracle’s ambitions expand beyond the cloud. The company is now connected, directly or indirectly, to:
- A massive OpenAI-related cloud backlog that helped power a mid-2025 rally (and later sparked anxiety about OpenAI’s ability to pay at scale). [2]
- A high-profile TikTok U.S. restructuring, in which Oracle is positioned as a “trusted security partner” and cloud custodian for U.S. user data. [3]
- A headline-grabbing Hollywood takeover fight, where Oracle co-founder Larry Ellison has personally guaranteed tens of billions of dollars to back Paramount Skydance’s bid for Warner Bros. Discovery. [4]
This is the story behind Oracle’s year-end slump—and the emerging debate over whether the selloff is the start of something darker, or the setup for a surprising 2026 rebound.
From AI darling to AI “bubble” bellwether: how Oracle’s 2025 turned topsy-turvy
Just months ago, Oracle was being treated like a late entrant that had finally cracked the AI infrastructure code. Then sentiment flipped.
By late December, the stock had swung dramatically within its 52-week range—from roughly $118 to $345—before settling near $197. [5] Those extremes matter because they mirror the market’s changing belief in Oracle’s AI thesis: first, that Oracle could monetize AI demand faster than the hyperscalers; later, that Oracle might be spending too aggressively to keep up.
Business Insider framed Oracle as a “poster child” for a broader set of AI excess fears—an emblem of what can happen when investor excitement meets the reality of giant capital outlays. [6]
Barron’s likewise described the company’s year as volatile and warned that the stresses of Oracle’s cloud shift—pressure on margins, rising debt—could bleed into 2026. [7]
The core worry: AI capex shock, cash burn, and rising dependence on OpenAI
Oracle’s latest earnings didn’t show a company in decline. But they did show a company in transition—and transitions can be expensive.
In its fiscal 2026 second-quarter results (reported Dec. 10), Oracle said:
- Revenue rose to $16.1 billion (up 14% in USD).
- Cloud revenue reached $8.0 billion (up 34% in USD).
- Remaining Performance Obligations (RPO) jumped to $523 billion, up 438% year over year—an enormous backlog signal that Oracle argues reflects multi-year demand. [8]
Oracle CFO Doug Kehring emphasized that the RPO surge was “highlighted by new commitments from Meta, NVIDIA, and others.” [9]
So why are investors nervous?
Because the market is focusing less on demand—and more on who pays upfront
Oracle’s AI strategy requires an unprecedented buildout of capacity. And Wall Street has been questioning whether the near-term spending curve is getting too steep relative to near-term revenue recognition.
Reuters summarized the mood bluntly after Oracle’s forecasts and spending plans disappointed: Oracle’s massive spending and uneven AI payoffs are feeding doubts about how quickly these bets translate into profits. [10]
On Oracle’s earnings call, management acknowledged the unusual nature of the moment and stressed discipline—saying it would pursue further expansion only when it meets profitability requirements and capital is available on favorable terms. [11]
The OpenAI question is now inseparable from the Oracle question
Oracle’s AI narrative is increasingly tied to OpenAI, not just as a customer but as a driver of the capacity buildout itself.
Reuters reported that investors have scrutinized Oracle’s AI infrastructure buildout as debt rises and “its fortunes become increasingly tied to OpenAI,” noting that OpenAI is loss-making and has not fully detailed how it would finance extremely large infrastructure ambitions. [12]
Meanwhile, Barron’s highlighted that Oracle’s OpenAI contract—reported as $300 billion—was a pivotal catalyst that initially excited investors and then became a source of concern about counterparty strength. [13]
A financing flare-up: Blue Owl steps away from a $10B Michigan data center plan
Those concerns got sharper in mid-December when a major Oracle data-center funding plan hit turbulence.
Reuters reported that Oracle’s $10 billion data center project in Saline Township, Michigan—a 1+ gigawatt site tied to the “Stargate” infrastructure push with OpenAI—ran into trouble when Blue Owl Capital was no longer part of the equity talks. Oracle said negotiations remain on track and that Blue Owl was not included, while a source said Blue Owl backed away because lease/debt terms were less favorable than in prior deals. [14]
DataCenterDynamics added key industry details: the planned campus is expected to reach about 1GW IT capacity on a 250-acre site and includes multiple large buildings, with construction targeted for early 2026; it also noted that Blackstone had been discussed as a potential replacement partner. [15]
The broader implication isn’t simply “one partner walked.” It’s that investors are testing the weakest link in the AI infrastructure chain: the economics of financing data centers at enormous scale.
Off-balance-sheet AI debt is under scrutiny—and Oracle is part of the trend
Oracle’s financing debate is also happening amid a wider Wall Street shift: using special purpose vehicles (SPVs) and private credit structures to build AI data centers without loading every dollar directly onto corporate balance sheets.
The Financial Times reported that major tech companies have shifted more than $120 billion of AI data-center debt off-balance-sheet via SPVs, and that Oracle has secured over $60 billion through similar structures for data centers leased to OpenAI—while experts warn this can obscure true risk and create vulnerabilities if AI demand cools. [16]
That matters for Oracle investors because it reframes the question from “How much debt does Oracle have today?” to “How much economic obligation is Oracle taking on across leases, partnerships, and financing vehicles—and what happens if utilization or pricing disappoints?”
The bull case Oracle keeps making: “chip neutrality,” huge backlog, accelerating cloud growth
Oracle’s argument is straightforward: demand is real, bookings are real, and the buildout is the price of admission.
A central theme in Oracle’s earnings release was “chip neutrality.” Larry Ellison said Oracle sold Ampere because it no longer sees designing its own chips as strategic and instead plans to deploy whatever chips customers want—while continuing to buy the latest GPUs from Nvidia. [17]
That positioning is important because it lets Oracle pitch itself as the “Switzerland” of AI infrastructure: it can host workloads from many model builders without forcing a proprietary chip stack.
And Oracle can point to numbers that are hard to ignore:
- RPO at $523 billion, a staggering backlog figure. [18]
- Cloud revenue up 34%, suggesting the cloud transition is accelerating. [19]
- Guidance on the earnings call projecting strong cloud revenue growth in the next quarter. [20]
This is why some market commentators argue the selloff may be overcorrecting. The Motley Fool, for instance, has made the case that Oracle could be a “surprising winner” of 2026 if it converts backlog into revenue and growth accelerates—arguing that valuation has become more reasonable after the drawdown. [21]
New leadership under pressure: Oracle’s co-CEO era begins in the middle of an AI arms race
The timing of Oracle’s market stress is especially notable because the company is undergoing a leadership transition aimed directly at AI execution.
Oracle announced in September that it promoted Clay Magouyrk and Mike Sicilia to serve as CEOs, with longtime CEO Safra Catz becoming executive vice chair. [22] Reuters described the move as a surprise joint-leadership structure designed to reinforce Oracle’s AI and cloud focus. [23]
That leadership change is central to the investor narrative captured in market coverage: with the AI buildout accelerating, the market wants proof the new CEO team can deliver capacity, financing, and profitability—without turning Oracle into a leveraged data-center developer.
TikTok’s U.S. deal puts Oracle in the middle of national security, algorithms, and cloud custody
One reason Oracle keeps showing up in headlines unrelated to databases is TikTok.
Axios reported that TikTok signed a deal to divest U.S. operations into a joint venture led by American investors, with a target closing date of January 22, 2026, aiming to resolve longstanding national security concerns. [24]
Reuters added crucial detail: Oracle is expected to serve as the “trusted security partner,” responsible for auditing and validating compliance, including safeguarding sensitive U.S. user data that would be stored in an Oracle-run cloud environment in the United States. [25]
Barron’s has framed Oracle’s TikTok involvement as part of a broader shift that’s unusual for a traditionally enterprise-focused tech investor base—bringing Oracle closer to the political and cultural center of gravity that TikTok represents. [26]
For Oracle bulls, the upside is clear: TikTok-scale workloads and security responsibilities could reinforce Oracle Cloud Infrastructure’s relevance and credibility. For skeptics, it’s another reminder that Oracle’s future returns may be shaped not just by technology, but by regulation, geopolitics, and public trust.
Hollywood enters the Oracle orbit: Ellison’s $40.4B guarantee in the Warner bidding war
Then there’s the subplot that feels ripped from a streaming drama: Oracle’s founder financially backing a major media acquisition fight.
Reuters reported that Larry Ellison personally guaranteed $40.4 billion to support Paramount Skydance’s bid for Warner Bros. Discovery, an effort aimed at outmatching a competing proposal from Netflix, with deal mechanics including a higher breakup fee and timeline extensions into January 2026. [27]
Paramount’s own press release distributed via PR Newswire spells out the guarantee and other terms, including that Ellison agreed to provide an “irrevocable personal guarantee” of $40.4 billion of the equity financing, and referenced the Ellison family trust’s Oracle holdings. [28]
Why this matters to Oracle shareholders: whether or not Oracle is directly involved, markets pay attention when a founder’s personal financial commitments become large enough that they could influence perceptions of stability, focus, or even the potential need to raise liquidity.
Barron’s argued that Oracle’s growing intersection with entertainment and social media is increasingly difficult for tech-only investors to ignore—especially as cable and legacy media economics weaken. [29]
Is Oracle a 2026 comeback story—or a warning label for AI infrastructure economics?
The market is trying to decide whether Oracle’s late-2025 plunge is a temporary confidence crisis—or an early sign that AI infrastructure returns may be slower, more volatile, and more financing-dependent than investors hoped.
The long-term optimism case
Industry research often supports the idea that AI spend will remain durable. IDC has projected enormous macroeconomic impact from AI adoption, estimating that by 2030 every new dollar spent on business-related AI solutions and services could generate $4.60 into the global economy through indirect and induced effects. [30]
If you accept that view, then Oracle’s backlog and cloud acceleration look like the foundation of a multi-year growth cycle—even if the near-term capex is painful.
The near-term realism case
At the same time, the “AI ROI is obvious” narrative remains contested inside many companies. A CFO.com report highlighted survey results suggesting only 14% of U.S. finance chiefs surveyed said they’ve seen a clear, measurable impact from AI investments so far (while many still expect impact within two years). [31]
That tension—big promised value vs. uneven realized value—helps explain why the market is demanding more proof that Oracle’s AI spend will translate into durable margins and free cash flow, not just bookings.
What to watch next for Oracle stock and Oracle’s AI buildout
As Oracle heads into 2026, several measurable checkpoints are likely to determine whether the company stabilizes—or continues to be a volatility magnet:
- Data-center financing clarity
Investors will watch whether Oracle and partners finalize funding structures for projects like Michigan—and whether terms tighten across private credit and lease markets. [32] - Backlog conversion (RPO to revenue)
Oracle’s $523B RPO figure is headline-grabbing, but markets will demand evidence of steady conversion into recognized revenue. [33] - Cloud growth durability vs. margin pressure
Cloud is growing fast, but the cloud transition can compress margins and push costs forward. [34] - Execution under the new CEO structure
Magouyrk and Sicilia were elevated specifically to lead the “AI era” strategy. 2026 is the year the market will grade them. [35] - TikTok U.S. deal milestones
If the TikTok JV closes as planned in January 2026, Oracle’s role as a trusted cloud and security provider could become more visible—and more politically sensitive. [36] - The Ellison media deal ripple effects
The Paramount–Warner battle is not Oracle’s operating business, but it may influence investor psychology around Oracle’s broader “orbit” of risk. [37]
Oracle’s late-2025 swoon is ultimately a referendum on a single bet: that Oracle can transform from a software-and-database stalwart into a scaled AI infrastructure provider—without the financing and execution risks overwhelming the story first.
References
1. stockanalysis.com, 2. www.barrons.com, 3. www.reuters.com, 4. www.reuters.com, 5. www.barrons.com, 6. www.businessinsider.com, 7. www.barrons.com, 8. www.oracle.com, 9. www.oracle.com, 10. www.reuters.com, 11. www.fool.com, 12. www.reuters.com, 13. www.barrons.com, 14. www.reuters.com, 15. www.datacenterdynamics.com, 16. www.ft.com, 17. www.oracle.com, 18. www.oracle.com, 19. www.oracle.com, 20. www.fool.com, 21. www.fool.com, 22. www.oracle.com, 23. www.reuters.com, 24. www.axios.com, 25. www.reuters.com, 26. www.barrons.com, 27. www.reuters.com, 28. www.prnewswire.com, 29. www.barrons.com, 30. my.idc.com, 31. www.cfo.com, 32. www.reuters.com, 33. www.oracle.com, 34. www.barrons.com, 35. www.oracle.com, 36. www.axios.com, 37. www.reuters.com
