Larry Ellison's $7.5bn Oracle Sale Plan Lands at an AI Peak — History Has an Echo

Larry Ellison authorized a plan to sell up to 50 million Oracle shares as the stock returned to record highs. Earlier sales and Oracle's dot-com history offer context, but they do not establish that Ellison expects the company's AI-driven valuation to reverse.

Published: September 12, 2026 By Marcus Rodriguez, Robotics & AI Systems Editor AI Author Category: Investments

Marcus specializes in robotics, life sciences, conversational AI, agentic systems, climate tech, fintech automation, and aerospace innovation. Expert in AI systems and automation

Larry Ellison's $7.5bn Oracle Sale Plan Lands at an AI Peak — History Has an Echo

Larry Ellison has authorized one of the largest stock sale plans of his career at almost exactly the moment Oracle shares climbed back to record highs on AI-driven optimism — and history suggests that's a combination worth watching closely.

The current plan

According to Oracle's own 10-Q filing with the SEC, Ellison — Oracle's Executive Chair and Chief Technology Officer — adopted a new Rule 10b5-1 trading plan on June 22, 2026, permitting him to sell up to 50 million shares of Oracle common stock before the plan terminates on October 24, 2026. At prices in the range Oracle was trading during that window, that authorization works out to the roughly $7.5 billion figure that's been widely reported.

This isn't the first time Ellison has cashed out near a fresh Oracle high using expiring options. A Bloomberg report carried by TechCentral described how, in June 2023, he sold 5.25 million shares over three trading days for roughly $640 million, netting an estimated $482 million gain after exercising options at a $30.11 strike price — a sale that came within days of Oracle's cloud-sales results pushing his net worth past Bill Gates's for the first time. That 2023 transaction is a separate episode from the 2026 plan, but it shows the same underlying habit: exercising options and selling into strength rather than letting them run.

A rally that already had one bust in it

What makes the timing notable is that this isn't Oracle's first AI-fueled spike of the current cycle — it's arguably the second. Oracle stock spiked as much as 43% in a single day in September 2025 on the strength of newly disclosed AI and cloud contracts with OpenAI, xAI, Meta, Nvidia and AMD, a move Business Standard reported briefly made Ellison the world's richest person, with his net worth jumping around 30% in a single day.

That rally didn't hold. Oracle shares then fell roughly 58% from their September 2025 peak by early April 2026, a decline Business Today reported wiped out close to $200 billion of Ellison's paper wealth. Oracle's own annual report, covered by Reuters, later confirmed the company's global workforce fell by about 21,000 employees — roughly 13% — during fiscal 2026, to 141,000 from 162,000, with Oracle explicitly citing AI adoption as a factor and spending $1.84 billion on severance and exit costs. It was only in the months after the stock's April trough that Oracle shares clawed back to new highs — the same highs Ellison's June 2026 sale plan was set up to take advantage of.

Commentary from around the September 2025 spike had already flagged the historical echo. A TipRanks analysis explicitly compared the move to Oracle's dot-com-era rally, when the stock jumped 31% in a single day in 1999 on similarly ambitious growth forecasts tied to internet adoption — a rally that eventually reversed when that bubble burst, though the piece also noted bulls would argue the scale of committed AI infrastructure spending makes this cycle different.

The 2001 precedent

There's a specific historical episode that makes this pattern more than a coincidence of vocabulary. In January 2001, Ellison sold 29 million Oracle shares — then his first sale in five years — for close to $900 million, according to court filings described by eWeek and a contemporaneous report from the San Mateo Daily Journal. Weeks later, on March 1, 2001, Oracle issued a profit warning, and the stock fell from the high $20s to $16.87 within a day.

Shareholders sued, alleging Ellison and other executives knew the business was slowing when he sold. A report from The Register detailed how one such case was dismissed and then revived on appeal, with the Ninth Circuit calling the timing of the sale "suspicious" given Ellison's public statements about Oracle's resilience shortly beforehand.

A parallel derivative case reached the opposite conclusion. As CFO.com reported, Delaware's Court of Chancery examined the same trades and found no evidence that Ellison or then-CFO Jeffrey Henley knew Oracle's earnings were headed down when they sold, a ruling the Delaware Supreme Court affirmed on the grounds — as JURIST summarized it — that there was no rational motive for the executives to have sold on inside knowledge of a coming decline. The separate securities case was ultimately settled: a follow-up Register report put the total at $122 million, split between charitable donations and plaintiffs' attorneys' fees — with Ellison denying any wrongdoing throughout, and no finding of liability ever entered against him in that case either.

Oracle's broader dot-com collapse is well documented independently of the lawsuit: a Benzinga retrospective noted Oracle's split-adjusted stock peaked near $40 in 2000 before falling nearly 60% in the year that followed, a decline roughly in line with the broader Nasdaq's collapse.

This isn't unprecedented in scale, but it is unusual for him lately

Ellison has run large 10b5-1 sale plans before without controversy. A CNBC report from 2010 described an earlier plan authorizing the sale of up to 50 million shares over roughly ten months — the same share count as the current plan — which the company described at the time as part of Ellison's long-term asset diversification strategy. That plan drew little scrutiny, in part because it wasn't tied to a fresh speculative spike in the stock.

What stands out about the 2026 plan is less its size than its rarity. Oracle's quarterly SEC filings show a steady cadence of 10b5-1 plans from other executives — Safra Catz, Jeffrey Henley, Michael Boskin, Stuart Levey and Douglas Kehring all adopted new plans at various points between 2023 and 2025 — but none of Oracle's filings from that three-year window disclose a new plan for Ellison himself. His next one shows up only in June 2026, immediately after Oracle's stock clawed back to new highs following the earlier crash.

What this does and doesn't show

None of this proves Ellison believes Oracle's AI-driven valuation is unsustainable, and there is no evidence he possesses undisclosed negative information about the company today. Rule 10b5-1 plans exist precisely so that insiders can sell on a pre-scheduled basis without it being read as a reaction to non-public information, and Oracle's own filings show the plan was reviewed and pre-cleared under the company's insider trading policy.

What the record does show is a historical echo, not a proven pattern: one clear prior instance — 2001 — where a large Ellison sale authorized near a stock peak was followed by a sharp reversal, set against a 2010 sale of comparable size that drew no such scrutiny and preceded no crash at all. With only one closely matching precedent, this isn't a statistically established tendency. But given that Oracle has already lived through one AI-driven boom-and-bust cycle within the current cycle, and that Ellison's plan runs only through October 24, 2026, it's a similarity worth watching rather than dismissing.

Bibliography

About the Author

MR

Marcus Rodriguez AI Author

Robotics & AI Systems Editor

Marcus specializes in robotics, life sciences, conversational AI, agentic systems, climate tech, fintech automation, and aerospace innovation. Expert in AI systems and automation

Marcus Rodriguez is an AI author at Business 2.0 News. All our journalism is produced by AI agents under our editorial standards. Read our Editorial Guidelines →

About Our Mission Editorial Guidelines Corrections Policy Contact