Larry Ellison Cancels Plan to Sell Up to $7.5 Billion in Oracle Stock

Larry Ellison will not sell 50 million Oracle shares after all.The company said Saturday that its co-founder and executive chairman canceled a Rule 10b5-1 trading plan that would have let him dispose of as many as 50 million shares through October 24. At Friday’s close near $150, that block was worth about $7.5 billion. When the plan was set on June 22, the same shares were worth closer to $8.75 billion.“No Oracle stock was sold under that plan, and he has no other plans to sell any of his Oracle stock,” Oracle said. It gave no reason.The plan appeared in a Friday regulatory filing, a day after quarterly results. On Saturday the plan was dead. Markets read founder sales as a weather report. Canceling one is the opposite report — or at least the removal of a storm warning.What a 10b5-1 Plan IsU.S. executives use these plans to sell stock on a pre-set schedule so they are less exposed to insider-trading claims. The calendar is locked when the person is not supposed to be trading on hidden news. Ellison adopted his on June 22. It was due to expire October 24. Nothing printed under it.Ellison, 82, left the CEO job in 2014 and remains executive chairman and chief technology officer. He is still the largest shareholder, with roughly 38 to 40 percent of the company depending on the count — on the order of 1.16 billion shares. Fifty million shares would have been about 4 percent of his stake, not a controlling-block exit. It would still have been one of the year’s largest scheduled insider sales.A person familiar with Ellison’s thinking told reporters he considers the stock undervalued. That person did not say he would buy more.Why the Stock Needed a SignalOracle shares are down about 22 to 23 percent in 2026. From the last close before the June plan, they were more than 18 percent lower. Investors have spent the year arguing about capital expenditure: data centers for AI customers, including a huge OpenAI-related commitment that sent the stock soaring last September and then gave back a large part of the gain as margins and cash burn came into focus.This past week’s quarter beat estimates and showed a smaller cash burn than feared. The stock still slipped. The company also raised the estimated cost of job cuts to about $2.8 billion, $700 million more than an earlier figure. AI infrastructure is expensive. Labor reductions are how Oracle is trying to show discipline while it builds.A scheduled 50-million-share sale into that tape would have looked like the founder taking chips off a table he designed. Canceling it, after the price had already fallen, looks like he does not want to lock in the drop — and does not want the tape to assume he does.Oracle has also become a major owner and security partner for TikTok’s U.S. operations. Ellison family money has backed David Ellison’s Warner Bros. pursuit, now in court. Those stories sit next to the stock plan. They do not explain the Saturday statement. They explain why every Ellison filing is treated as more than a 10-Q footnote.The Overhang ProblemWhen a filing says the chairman may sell $7.5 billion of stock by late October, every dip can be blamed on that supply. Funds that model float start assuming those shares will appear. Canceling the plan removes the overhang. It does not add a buyer. It subtracts a known seller.That is useful after a week in which Oracle tried to prove the AI build is producing revenue without setting the cash pile on fire. Beating estimates is one message. The founder keeping every share is another. Together they are aimed at people who have been selling the “capex forever, equity never” story.It is not a guarantee. Ellison can adopt a new plan later. “No other plans” means none disclosed today. Founders change their minds. This founder just did, in public, within 24 hours of the first disclosure.What It Does Not SettleIt does not settle whether Oracle’s AI contracts will earn an adequate return. It does not settle the job-cut bill. It does not settle how much debt the data-center program still needs. It does not make a $150 stock a bargain by itself.It does tell holders that the person with the most Oracle equity is not using this window to lighten up. In a year when many tech founders have sold into strength or diversified, Ellison is doing the opposite after weakness.Compare that with the June decision. Adopting a plan after a slide already underway can look like estate planning, taxes, or liquidity for other bets. Killing the plan after another slide can look like a valuation call. Outsiders only have the statement and the price chart.How Investors Will Use ItShort-term, the news is a relief bid ingredient, not a thesis. One canceled sale does not refinance a cloud build. Medium-term, analysts will watch whether Ellison files a new 10b5-1, whether he adds shares, and whether Oracle’s next quarter still shows capex rising faster than free cash flow.If the stock recovers toward the June levels, the canceled sale will be remembered as timing. If it keeps falling, critics will say he should have sold. Either way, the float will not include those 50 million shares this October.Ellison built Oracle by holding. The company is now a leveraged bet on other people’s AI models needing his iron and power. Holding through that bet is consistent with the man. Disclosing a sale and retracting it in a weekend is consistent with a market that punishes any hint the insider wants out.No shares sold. No new sale plan. That is the entire official story. The unofficial story is that $7.5 billion of paper supply just left the calendar, and the largest owner would rather own Oracle at $150 than own more cash.