Same Night, Opposite AI Bets: Oracle and Adobe Close Out Earnings Season

Two giant software names report after the close on September 10, and they could not be trading in more different places. Oracle is up more than 5% over the past five sessions going into its print. Adobe slid 6.7% in a single session this week after announcing its CEO succession. Same calendar date, opposite sentiment — and each stock is a bet on a different answer to the same question: does AI enrich established software companies, or replace them?

Oracle is the easiest trade to read, and the hardest to own through the noise. It guided fiscal-first-quarter revenue up 27% to 29% year over year, with cloud revenue up 58% to 64%. That is the market's AI-infrastructure buildout thesis in one line: spend on data centers now, convert the backlog to revenue, fix free cash flow later.

The backlog is real and enormous. Oracle finished its last fiscal year with a record $638 billion in remaining performance obligations, up $85 billion in a single quarter. A year ago its first-quarter report, built on multi-billion-dollar OpenAI-era contracts, set off a record single-day rally and briefly made Larry Ellison the richest man in the world. Within days, analysts were warning that much of the coming growth was concentrated in a single client: OpenAI.

That is the commercial flywheel working in theory. The problem is the cost of buying it. Oracle spent $55.7 billion on data centers in fiscal 2026, overshooting its own $50 billion guidance, and has said it will raise roughly another $40 billion. Free cash flow over the trailing twelve months sits at roughly negative $24 billion, against about $218 billion in total debt and close to $100 billion in net debt. This is the purest "build first, monetize later" balance sheet in the AI trade.

So the scoreboard going into Thursday is a 90-point hurdle. To beat it, Oracle needs cloud growth at or above the 58%–64% guide, plus evidence the $638 billion backlog is converting to reported revenue on schedule — not just sitting on the books. A print that merely matches won't re-rate a stock already down roughly a third over the past year and trading near 37 times forward earnings.

There is also a real narrative crack. CNBC reported in March that OpenAI is walking away from expanding its flagship Stargate site with Oracle because it wants next-generation Nvidia chips at new sites instead. If the customer that drove the 2025 rally is shopping its compute elsewhere, "the backlog is the moat" gets weaker, no matter what the number says.

Adobe is the mirror image: a profitable incumbent trading like it already lost. The numbers underneath the fear are strong. Adobe guided fiscal third-quarter revenue to $6.67–6.72 billion, and the business just grew revenue about 11% year over year with an operating margin near 36%, a free-cash-flow margin near 40%, about $10 billion in trailing free cash flow, and almost no net debt. That is a different balance sheet language entirely — cash generated, not cash burned.

The market has decided none of it matters yet. Adobe is down about a quarter over the past year and still sits roughly a quarter below its $370 peak, trading near 15 times forward earnings and roughly 4 times sales. The reduction to a single-digit-exit growth multiple is the market pre-pricing a world where AI-native rivals — Canva, Figma, generative tools — eat the Creative Cloud subscription base that generates the cash.

The bear case has fuel. Morgan Stanley downgraded Adobe to Underweight in July, cutting its target to $240, and cited AI competition to Creative Cloud recurring revenue. This week the company named Anil Chakravarthy to succeed Shantanu Narayen as CEO in December, added a couple of executive departures to a run that already included the June exit of the CFO, and the stock dropped on it. Leadership churn at a moment of strategic uncertainty is exactly what a cheap but unloved stock does not need.

But here is what the market's fear still has to prove: that AI is destroying digital-media revenue, not adding to it. Adobe grew revenue 11% last quarter and still prints a 40% free-cash-flow margin. The $638 billion question for Oracle and the $10 billion cash-flow answer for Adobe are the same AI cycle measured at opposite stages.

The deeper rewrite. Surface read: two earnings reports to watch. Underlying read: an elimination-round test of whether the AI cycle pays the people who spend, or the people who already monetize. Oracle argues the backlog converts; Adobe argues the incumbent survives. One of them gets a new valuation anchor on Friday morning.

The checkpoints are concrete. For Oracle, watch the reported cloud growth against the 58%–64% guide and whether the debt-funded buildout keeps free cash flow from getting worse than already guided — and any word on the OpenAI relationship. What invalidates the thesis: cloud growth slipping below the guide on a balance sheet that still cannot self-fund. For Adobe, watch Digital Media annual recurring revenue — whether new AI features like Firefly are adding net-new paid seats or quietly becoming free tiers — and how the new management team steadies the company. What invalidates the thesis: ARR deceleration or evidence the free-tier shift is pulling paying subscribers away.

The market has already chosen its favorite. On Thursday night, one of these two stories gets tested against the company's own forecast — and the loser on the scoreboard tells you which half of the AI trade can actually convert.