Ellison Files a $7.5 Billion Oracle Sale Plan. Then Immediately Cancels It.

Larry Ellison filed to sell $7.5 billion of Oracle stock on Friday. By Saturday morning, he had cancelled the plan. Not on Monday. Not after some weeks of deliberation. Before the markets opened the next trading day.

No shares were sold. No money changed hands. But the sequence itself is a message.

Here is how it works, and what the plumbing says about Oracle's leadership and about you as a reader trying to figure out whether this stock is recovering or just stalling.

The mechanism

Oracle announced on Friday, September 11, that Larry Ellison — executive chair and chief technology officer — had adopted a Rule 10b5-1 trading plan authorizing him to sell up to 50 million shares of Oracle common stock, worth roughly $7.5 billion at the current price. The plan was adopted on June 22 and was set to run through October 24.

That plan is a standard securities-law device. Under normal SEC rules, insiders can't trade while they hold material non-public information — which is basically always, if you run the company. The workaround is a 10b5-1 plan: you set up an instructions-ahead-of-time framework, commit to a schedule, and then the trades execute mechanically, insulating you from insider-trading accusations because you agreed to the plan before you knew the news.

But here's the thing that makes a 10b5-1 plan interesting as a communication tool: you can cancel it at any time. Before any trade executes, you can tear it up. The cancellation carries no penalty, requires no explanation, and is not itself a trade.

So Ellison had a plan on the books that could have sold $7.5 billion. And then he announced he was ripping it up.

On Saturday morning, Oracle issued a press release: "Larry Ellison has cancelled his 10b5-1 Plan to sell Oracle stock". No Oracle stock was sold under that plan, he said, and he has no other plans to sell any of his Oracle stock.

Friday: I'm selling $7.5 billion. Saturday: Actually, never mind. I'm not selling anything.

That was fast.

The timing isn't accidental

A 10b5-1 plan doesn't execute on a schedule the insider controls in real time — it's supposed to be mechanical, precisely to avoid the appearance of timing. So the plan was adopted June 22, three and a half months before Oracle announced it publicly. During those months, Oracle's stock fell to around $150 as the company burned through billions on AI data centers. The market got nervous about the debt, the concentration in OpenAI, the sheer scale of capital expenditures.

Then on September 10, Oracle reported fiscal Q1 2027 results. Revenue hit $19.35 billion — 30% growth year over year, beating expectations. Non-GAAP earnings per share came in at $1.92, above the $1.74 consensus. Management raised full-year revenue guidance to $90 billion.

On September 11, Oracle announced Ellison's existing plan to sell 50 million shares.

On September 12, he cancelled it.

The basic point is this: the plan was filed in June — before earnings — which means the SEC doesn't need to question whether Ellison was trading on inside information. Then Oracle waited until after the strong earnings to announce the plan existed, then immediately cancelled it. The sequence reads like a signal: "We considered a massive sell-off. The results changed our mind. We are confident."

It's a way of broadcasting confidence without actually selling a single share. You file the plan months in advance, you announce it when the news is good, you cancel it to remove the overhang — and you haven't technically traded on inside information because the plan itself was established before you knew the results.

I don't know how deliberate the timing was. The plan was genuinely adopted in June, which is the hard fact the SEC looks at. But the public disclosure and the cancellation were both within 24 hours of each other, and both immediately followed a quarter that blew past expectations. The sequence works too well for you to ignore it.

What it isn't

This isn't a signal that Oracle is out of the woods. And it's not Ellison saying, "the stock is going to $280 again."

What it is, is the most powerful person at Oracle telling the market: I don't need $7.5 billion of cash right now.

That matters because the opposite signal would have been devastating. If Ellison actually started selling — even under a legitimate pre-arranged plan — investors would ask what he knows that they don't. The company is spending aggressively on AI infrastructure, has taken on significant debt to fund it, and its stock has fallen roughly 51% over the trailing twelve months — about 54% from its 52-week high of $329.50 — to today's levels around $150. In that environment, a $7.5 billion insider sell program is the kind of overhang that changes how people price the stock. It doesn't matter if Ellison has "plenty of shares left." What matters is that the guy who sees every internal number decided to dump a chunk.

So removing the overhang is, by itself, a useful piece of information. It's not the same as a buy signal. It's more like: "one risk you were worried about isn't materializing."

The real story the cancellation sits on top of

The cancellation is useful. It doesn't answer the harder questions about Oracle.

The company's fiscal Q1 2027 results were genuinely strong — 30% revenue growth, a raised full-year guidance of $90 billion, and what management calls remaining performance obligations (their backlog) that stunned Wall Street. Oracle's cloud infrastructure business, OCI, is growing fast and the company has become one of the primary compute providers for OpenAI. That's the growth story.

But Oracle's stock is still roughly 50% below its high. The trailing P/E sits around 23x with forward earnings at 35x. The enterprise value — which includes debt — is $521 billion. Oracle has borrowed aggressively to fund its AI buildout, and capex surged 162% in the prior fiscal year. The company is spending enormous sums on data centers that won't generate full returns for years, and OpenAI is a large enough concentration to matter if anything goes wrong with the partnership.

The earnings beat and Ellison's cancelled plan are both pieces of good news. Neither one resolves the fundamental question of whether Oracle's debt-fueled AI bet pays off in a way that justifies the price investors are paying. The earnings showed the top-line is growing. The cancellation showed the insider isn't fleeing. But the stock's valuation still asks you to believe that the spending will keep converting to profit and that the market isn't going to reprice the debt risk.

What you take from this

The cancellation of a $7.5 billion insider sale plan is a positive data point — a removed risk. Ellison, who controls roughly 40% of Oracle, just said publicly he has no plans to sell. That's meaningful information about one of the biggest potential sellers in the market.

But it's one sentence, not a thesis. The company is growing fast, spending faster, and carrying more debt than it did a year ago. The earnings beat is solid. The cancelled plan is reassuring. Neither of them guarantees the spending trajectory produces returns that exceed the cost of capital — or that the market won't reprice the leverage if growth slows.

The cancellation tells you what Ellison isn't doing. Whether that's enough to change what you think Oracle is worth depends on everything else you know about the company — and how much you believe the growth justifies the debt.