The $690 Billion Paradox: Oracle’s Record Backlog Couldn’t Save Its Stock, and SAP Got Caught in the Crossfire

On June 10 and 11, 2026, a paradox played out in the enterprise software market. Oracle and SAP, two of the world's largest business-software providers, both saw their shares punished by investors—not because of weak sales or waning demand for artificial intelligence, but precisely because that demand is so expensive to meet.

Oracle reported arguably the best top-line numbers in its history: $19.2 billion in quarterly revenue, a 93% surge in cloud infrastructure revenue, and a staggering $638 billion Remaining Performance Obligations (RPO) backlog . Yet within hours, the stock sank 9–10% in extended trading . The following day, SAP—which didn't report earnings at all—fell roughly 4% after Goldman Sachs lowered its gross margin forecast, specifically pointing to rising hardware costs expected through the rest of 2026 .

Together, the moves reveal a market that has stopped rewarding AI ambition and has started auditing the bill.

Oracle’s Q4 2026: Historic Growth, Historic Spending

By the headline numbers, Oracle’s fiscal fourth quarter and full year for 2026 were exceptional :

• Full-year revenue: $67.4 billion, up 17% year over year.
• Q4 cloud infrastructure (OCI) revenue: $5.8 billion, up 93%.
• Remaining Performance Obligations (RPO): $638 billion, up 363% from the prior year and $85 billion sequentially.
• Operating cash flow: A record $32.0 billion.

But buried beneath those figures was a growing strain on the balance sheet that investors found impossible to ignore:

• Capital expenditures hit $55.7 billion for fiscal 2026, well above Oracle’s own original $50 billion projection and roughly 83% of its total annual revenue .
• Free cash flow went deeply negative, reaching negative $23.69 billion for the year—meaning the company burned through tens of billions in cash even as revenue and operating income hit records .
• The company raised its capital-raise target, announcing plans to secure roughly $40 billion in additional debt and equity financing in fiscal 2027. That included a previously disclosed $20 billion at-the-market equity issuance .

The math is stark. Oracle generated a record $32 billion in operating cash flow but reinvested $55.7 billion back into AI data centers—and now intends to spend $70 billion or more in the coming year . To bridge the gap, it is turning to debt and equity markets at an unprecedented scale, diluting existing shareholders and raising concerns about the sustainability of the cycle.

As one analyst note put it, "Fiscal 2026 capex near $55.7B and negative free cash flow near $23.7B turned the earnings beat into a funding story" . Oracle’s backlog proves that customers are lining up for AI cloud capacity. The sell-off proves that investors are no longer willing to fund that capacity at any cost.

SAP’s 4% Drop: The Margin Squeeze Arrives

Hours after Oracle’s earnings landed, SAP shares slid roughly 4% on June 11 . The trigger wasn’t a quarterly report—it was a note from Goldman Sachs. The investment bank cut its gross margin forecast for SAP’s second half from 73.3% to 72.8%, explicitly citing rising hardware costs expected to hit the German software giant in the coming months .

The logic is straightforward. Even companies like SAP—which primarily sell cloud and enterprise resource planning software rather than raw infrastructure—are not immune to the AI capex wave. As hyperscalers pour hundreds of billions into data centers, the cost of compute, networking, and server hardware ripples through the entire technology supply chain. Traditional software firms whose cloud products run on that infrastructure see their margins compress.

Goldman’s adjustment also brought down the firm’s full-year EBIT growth projection for SAP . It was a targeted downgrade, but its timing and rationale placed SAP squarely in the same narrative that was punishing Oracle: the industry is spending so aggressively on AI that even the indirect costs now threaten legacy margins.

The $690 Billion Arms Race

Oracle’s numbers are extreme, but they are not an outlier. They are part of a sector-wide capital sprint that has no modern precedent. The five largest U.S. cloud and AI infrastructure providers—Amazon, Alphabet, Microsoft, Meta, and Oracle—committed a combined $660–$690 billion in capital expenditure for 2026, nearly double the roughly $380 billion spent in 2025 .

Of that total, roughly 75%—over $450 billion—is tied specifically to AI infrastructure: data centers, GPUs, custom silicon, and high-speed networking . To put Oracle’s own position in context :

Company	2026 Capex (Est.)	Capex as % of Revenue
Amazon	$200 billion	~25%
Alphabet	$175–185 billion	—
Microsoft	~$145 billion (annualized)	—
Meta	$115–135 billion	—
Oracle	~$55.7 billion (actual FY26)	~83–86%

Oracle’s capex-to-revenue ratio of roughly 83–86% is by far the highest of the group. To sustain that level of reinvestment, the company is almost entirely dependent on external financing . While Amazon, Alphabet, and Microsoft can fund massive capex programs from enormous free cash flow engines, Oracle has pushed itself into a position where revenue growth and balance-sheet health are pulling in opposite directions.

The Real Signal for Enterprise Software

The twin stock drops on June 11 mark a turning point in how the market values AI-driven growth across the enterprise software sector. Three shifts are now visible:

1. Revenue growth is no longer enough. Oracle’s $638 billion backlog and 21% quarterly revenue growth would have been celebrated in any prior cycle. But investors have now trained their focus on the price of building that backlog and the financing risk required to fulfill it . The market is asking whether these contracts will generate margins that justify the upfront cost.

2. Margin pressure is spreading. SAP’s Goldman downgrade on hardware costs shows that the AI capex shock is not limited to infrastructure providers. Even enterprise application companies face cost inflation as the underlying hardware and cloud-services ecosystem becomes more expensive and constrained .

3. Financing risk is becoming stock risk. Oracle’s plan to raise $40 billion in combined debt and equity represents a direct transfer of project risk to shareholders in the form of dilution and higher leverage . The message from the after-hours sell-off is that investors are no longer willing to provide a blank check for the AI buildout—they want to see proof of returns, not just proof of demand.

The AI arms race has produced staggering top-line numbers and a $690 billion annual capex commitment from Big Tech that is reshaping entire supply chains. But the events of June 10–11, 2026, make clear that the second chapter of this story—where bills come due and margins are tested—has now begun.