Larry Ellison Rarely Sells Oracle. Now He's Set Up to Sell $7.5 Billion of It.

Title: Larry Ellison Rarely Sells Oracle. Now He's Set Up to Sell $7.5 Billion of It.

Larry Ellison has, since the start of the century, never sold more than 25,000 Oracle shares in a single transaction. That is not a quirk. For a man whose fortune is essentially one stock, refusing to sell it is a standing statement that the stock is worth holding. So when the market learned this month that Ellison had quietly arranged to sell up to 50 million shares, about $7.5 billion at current prices — it registered as the loudest divestment signal of his career. Then you notice the details, and the story gets stranger.

The news is not that Oracle's founder flipped bearish. It's that he set up a machine to make the question of whether he's bearish almost beside the point.

What the plan actually is

The mechanism is a Rule 10b5-1 trading plan, adopted on June 22 and running through October 24. In plain English, it is a pre-arranged schedule of sales that an executive sets up in advance and then lets run automatically. A 10b5-1 plan does two jobs. It gives the insider an affirmative defense if anyone ever accuses him of trading on material nonpublic information — the trades happen on a schedule, not around news. And it lets a person holding billions of dollars of one stock bleed it out gradually, which matters when selling all at once would crush the price and tip off everyone.

The part that changes how to read it is the words "up to." A plan like this is permission to sell, not a promise. It authorizes a maximum, and the trades are typically gated by price and by prevailing market volume. If the stock sits near its lows and the plan's floors aren't met, very little may get sold at all. So "$7.5 billion plan" is a ceiling, not a forecast. The share count tells you the same thing: even if every share under the plan were sold, Ellison would still own roughly 1.1 billion Oracle shares and keep control of well over 40% of the company.

That is the odd part worth sitting with. This is a man arranging to sell a rounding error of his stake, and the market reads it as a signal anyway — because for someone this loyal to a single stock, even the existence of a sale plan is news.

Why a man who rarely sells picked now

Timing makes the move stranger. Ellison set the plan up on June 22, and the very same week Oracle's stock took its worst beating in 25 years, shedding about 19% as investors balked at the bill for AI. The broader numbers are stark: Oracle has loaded up on roughly $130 billion of debt to build the data centers behind its AI push, capital spending surged 162% last fiscal year to nearly $56 billion, and free cash flow ran about $24 billion negative. The stock has lost more than half its value from its September 2025 peak.

Selling into a stock that has been cut in half is a strange thing for Oracle's biggest bull to do unless the sale is not really about the stock. And the most concrete reported reason points somewhere else entirely: in July it emerged that Ellison had personally guaranteed $40.4 billion of debt behind his son David's bid to buy Warner Bros. Discovery.

This is where the plumbing matters. A fortune that is one stock is not a liquid fortune. To actually honor a $40 billion personal guarantee, you need cash, and the practical way for a tech billionaire to convert a single-stock fortune into cash is to sell a slice of it. Seventh and a half billion dollars is trivial next to an ownership stake worth roughly $170 billion, but it is real money when it is backing a personal balance sheet. The seller's motive lives outside Oracle — on his own guarantee — which is both reassuring and a little unsettling.

What it means if you own the stock

There is a temptation to read a founder's sale plan as a vote of no confidence. The evidence here runs the other way on the important point: Ellison is keeping the asset, not exiting it. People sell for all sorts of reasons, and the visible reason here is a personal guarantee that requires liquidity, not a freshly formed belief that Oracle is overvalued. Selling 3% of a stake to fund a guarantee looks more like financial plumbing than a bearish thesis.

But the optics are still a real overhang. A founder arranging to sell into a collapse, layered on top of a company that is borrowing tens of billions to finance AI infrastructure that has not yet produced the returns investors were promised, is a reminder of how this whole story is being paid for. And the market's reaction will do some of the work: the plan surfaces mid-crash, investors ask whether the founder's cash needs signal pressure, and the stock trades on the question.

The honest synthesis is this: the founder who rarely sells is selling a sliver, at a terrible price, and the reason has more to do with a $40 billion guarantee than with Oracle's business. That is, in a sense, good news — the person closest to the company is still holding nearly all of it. It is also a useful reminder that a fortune concentrated in one stock converts to cash only by selling into whatever price the market happens to offer, which right now is a halved one. When the founder who rarely sells finally does, the story is rarely about the company at all.