Oracle (ORCL) Stock Analysis 2026: The $560B Backlog That Made Larry Ellison the World's Richest Man

Oracle's income statement blends a shrinking legacy licensing business, a steadily growing SaaS applications business, and a rapidly accelerating cloud infrastructure business. The mix matters because each piece is valued very differently by the market.

Remaining performance obligations (RPO) is contracted revenue Oracle hasn't recognized yet — the closest thing to a visible order book that a software company reports. Oracle's RPO more than tripled in a single quarter when it disclosed several multi-billion-dollar, multi-year cloud compute agreements, headlined by a widely reported large commitment from OpenAI, alongside deals with Meta, xAI, and other AI developers needing GPU capacity.

The scale is what makes this remarkable: Oracle's total FY2026 revenue is roughly $66B, meaning the backlog alone represents more than eight years of the company's current total revenue run rate. Converting that backlog into recognized revenue requires Oracle to build data-center capacity at a pace few software companies have ever attempted — which is exactly why the capex numbers below matter as much as the backlog itself.

Oracle Financial Metrics Deep Dive

Oracle's economics are shifting in real time: operating margin is under near-term pressure from the capex buildout, even as the top-line growth rate accelerates on the back of OCI.

It's easy to get swept up in the OCI/AI story and forget that Oracle's original business — the Oracle Database — remains one of the stickiest pieces of enterprise software ever built. Mission-critical systems at banks, telecoms, airlines, and governments run on Oracle Database, and migrating off it is a multi-year, high-risk undertaking most CIOs avoid unless forced.

• Multicloud database agreements with AWS, Azure, and Google Cloud let customers run Oracle databases inside rival clouds — extending Oracle's licensing reach into competitors' data centers rather than losing those workloads entirely
• Autonomous Database and Exadata Cloud@Customer let large enterprises run Oracle's managed database stack inside their own data centers — a hybrid model competitors don't easily replicate
• The license-support annuity (renewed support contracts on existing on-prem licenses) still throws off high-margin cash even as new license sales decline, cushioning the P&L during the capex-heavy cloud transition
• Switching costs are the real moat: ripping out a database underneath decades of custom enterprise applications is a project few companies choose to start voluntarily

Oracle now trades at a premium multiple to most of its historical enterprise-software peer group, reflecting the market's willingness to price in AI-infrastructure-style growth on top of a legacy software base.

Oracle's forward P/E now sits above Microsoft's despite lower operating margins — the market is pricing Oracle less like a mature enterprise software vendor and more like an AI-infrastructure growth story, which raises the bar for what needs to go right.

• $560B+ RPO backlog provides multi-year forward revenue visibility that almost no other software company can match — this isn't a projection, it's contracted demand
• OCI is winning AI-compute customers specifically because it offers a different price/performance profile and multicloud flexibility than the largest hyperscalers — a genuine differentiated wedge, not just excess capacity
• The legacy database business remains a high-margin cash annuity that funds the AI buildout without diluting shareholders as aggressively as some capex-heavy peers
• Multicloud database partnerships with AWS, Azure, and Google Cloud extend Oracle's licensing reach into rivals' infrastructure — a rare case of cooperating with direct competitors to grow revenue
• If even a fraction of the RPO backlog converts to recognized revenue on schedule, current revenue and EPS estimates look conservative for FY2027 and beyond

• Customer concentration risk: a large share of the RPO backlog is tied to a small number of AI-developer customers, some of whom are themselves not yet profitable — if any major contract is renegotiated or delayed, the backlog headline number could prove less durable than it looks
• Capex is running well ahead of current cash flow, funded partly with debt — if backlog conversion slips even a few quarters, the balance-sheet math gets uncomfortable quickly
• Margin compression: heavy depreciation from new data centers is already pressuring operating margin, and that pressure likely intensifies before it eases as more capacity comes online
• Valuation now assumes flawless execution on an unprecedented infrastructure buildout — any AI-demand slowdown from hyperscale customers would hit Oracle's growth narrative disproportionately hard given how much of the recent re-rating depends on it

These are directional, forward-looking scenarios rather than precise targets — the range of outcomes for Oracle over the next two years is unusually wide given how much depends on backlog conversion timing.

What Analysts Say About ORCL in 2026

Wall Street's split on Oracle is less about whether the AI-infrastructure demand is real — most analysts accept that it is — and more about the pace and risk of converting a $560B backlog into recognized, profitable revenue without straining the balance sheet. That's a meaningfully different debate than the one investors were having about Oracle even two years ago.

Bottom Line: Is ORCL a Buy, Hold, or Sell in 2026?

For patient AI-infrastructure investors: Buy, sized appropriately. Oracle's RPO backlog gives it forward revenue visibility that few software companies have ever reported at this scale, layered on top of a database franchise with genuine, decades-proven switching-cost economics. That combination is rare.

The key risk to monitor: Customer concentration in the backlog and the pace of capex-to-revenue conversion. Watch quarterly RPO growth alongside capex guidance — if backlog growth stalls or capex keeps outrunning recognized revenue for several more quarters, the market's patience with the story could shorten quickly.

Our BriMind AI Score for ORCL is 71/100 — reflecting an exceptional backlog and durable database moat, offset by elevated capex, debt, and customer-concentration risk that a mature, lower-growth software company wouldn't normally carry.

The OCI vs AWS/Azure/Google Cloud Question

Oracle Cloud Infrastructure is the smallest of the major hyperscale clouds by revenue, which is precisely why its recent AI-compute wins are notable. OCI has positioned itself less as a general-purpose cloud trying to out-compete AWS on breadth of services, and more as a specialized, cost-competitive venue for large-scale GPU training and inference workloads — the specific thing AI developers need most right now.

• Price/performance positioning: Oracle has publicly emphasized lower networking and compute costs relative to the largest hyperscalers for GPU-dense workloads, an argument that resonates with AI developers burning through capital on training runs.
• Multicloud strategy: Rather than competing head-on for every workload, Oracle has struck partnerships letting customers run Oracle Database and applications inside AWS, Azure, and Google Cloud data centers — extending its reach without needing to win the infrastructure battle outright.
• Capacity, not just customers, is the constraint: The bottleneck for converting RPO into revenue isn't demand — it's how fast Oracle can build and power new data centers. Energy availability and construction timelines are now as important to the Oracle story as the software itself.
• Competitive response: AWS, Azure, and Google Cloud are all expanding their own AI-compute capacity aggressively, meaning Oracle's current price/performance edge is not guaranteed to persist — it needs to keep innovating on cost and networking architecture to defend the wedge that won it these contracts in the first place.

Compare Oracle with Cloud & Enterprise Software Peers

Read Next