On Monday, the man who holds half the votes gets to decide what his own stock options cost.

Start with the pitch, since both companies have been selling it for a decade.

The Trade Desk is the champion of the open internet. The objective buyer. The DSP that owns no media, so it has no reason to cook the books, so you can trust it with your money while Google, Amazon and Meta play both sides of the table. Jeff Green has delivered some version of that sermon on every stage that would have him.

Magnite is the world's largest independent sell-side platform. Its own annual report says so, nearly word for word. It runs the pipes that publishers and streamers use when they don't want Google running their auctions.

Two companies. One word: independent. Same antitrust case. Same AI hype cycle. Same macro. Same buyers in a lot of the same rooms.

Here's how the market graded the word in 2026.

The Trade Desk is down 68% this year. Its market cap is about $5.7 billion. In 2024, the same company was worth roughly $69 billion.

Magnite is up 57%.

Opposite year. The lazy explanation is that one company had a bad quarter. Companies have bad quarters. The real explanation is sitting in the SEC filings, and it comes down to three things: what each company owns, who each company answers to, and who gets protected when the stock goes the wrong way.

Here's how 2026 is ending for the two independents. In October, Magnite repriced its loans so it pays less interest. The Trade Desk is repricing its stock options so its people lose less money. And on Monday, October 19, the man who holds 50.2% of the vote on that repricing is the CEO whose own options are on the list.

That's not a metaphor. That's a proxy statement.

On August 6, The Trade Desk reported second-quarter revenue of $715 million, up 3%. A year earlier, the same quarter grew 19%. For a company that compounded revenue at roughly 34% a year over its first decade as a public company, 3% isn't a slowdown. It's a different company.

Green's own words in the release: the quarter "did not meet the standard we set for ourselves."

Then came the guidance. At least $650 million for Q3. Wall Street had been modeling about $804 million. An analyst on the call did the arithmetic out loud: that's roughly a 12% revenue decline year over year. Green didn't argue with the math.

The explanation was pressure from big consumer packaged goods and automotive clients, plus what Green himself called execution shortcomings. Those two categories, he said, are overrepresented on the platform and make up about a quarter of the business.

Read that one again. The champion of the open internet runs a quarter of its business through two categories of Fortune 500 marketers. When Procter and Ford tap the brakes, the open internet gets whiplash. That isn't a movement. That's a client list.

The stock dropped about 22% the next day. It has not come back.

Magnite reported a day earlier, on August 5. Nobody wrote a think piece. They should have.

Revenue up 11% to $192.8 million. Contribution ex-TAC up 17% to $189.6 million. CTV contribution up 36% to $97.1 million. It beat the top of its own guidance on every revenue measure it forecasts, then raised the full-year outlook on four separate lines: contribution ex-TAC growth to 13% to 14%, adjusted EBITDA growth to above 20%, adjusted EBITDA margin to at least 37%, and free cash flow growth to the high 40s.

CTV is now 51% of the business. Its top ten CTV accounts grew in the mid-to-high 40s. And management says no single publisher is more than 5% of the company's contribution ex-TAC. Netflix, Roku, LG, Paramount, Walmart and Warner Bros. are all growing on Magnite, and none of them can sink it by leaving.

Magnite doesn't have a founder giving keynotes about saving journalism. It has the pipes. SpringServe ad serving inside the streamers. Commerce media partners plugged in. And since September 2, a federal judge ordering Google to make its ad exchange work with rivals instead of strangling them.

When Judge Brinkema's remedies ruling landed, Magnite rose 5%, PubMatic rose 6%, and The Trade Desk barely moved. The market read the biggest antitrust win in ad tech history the way it has read everything this year: as a sell-side story.

On September 18, The Trade Desk mailed its shareholders a proxy for a special meeting. There is one real item on the ballot: cut the exercise price on up to 15,582,540 underwater stock options to whatever the stock closes at on October 19.

The options were granted between 2017 and this September. Their strike prices run from $3.69 to $122.59. With the stock at $14.49 the day the proxy's numbers were struck, most of them are worthless as written. The board's argument is retention. Underwater options don't keep anybody.

That argument isn't crazy. Repricing is a real tool, and a company that just lost most of its value has a real problem holding on to engineers. The board hired two consultants, weighed an exchange offer, new grants and waiting it out, and picked the repricing as the cheapest fix. Proxy adviser Glass Lewis generally opposes repricings anyway, on a simple theory: if shareholders eat the downside, insiders shouldn't get a do-over.

Now read the dates.

August 6: The Trade Desk reports the quarter that broke the slogan.

August 13: The board approves the repricing. One week later.

September 4: The company discloses it will cut about 15% of its staff, with $39 million to $51 million in severance charges, partly offset by a $4 million to $5 million reversal of stock compensation. That reversal is the equity the departing employees forfeit.

September 14: The board grants Jeff Green a new performance option on 7 million shares at $14.97. The same day is the record date for the repricing vote.

September 21: The layoffs take effect. The Trade Desk leaves the S&P 500.

October 19: The vote.

Here's who's on the list. Jeff Green holds 1,618,489 options eligible for the repricing, at weighted average strikes of $25.00, $49.24, $61.46 and $74.90. These aren't his giant performance grants, which are carved out. They're ordinary options, the same kind the engineers hold, and they'd reset to Monday's close.

Here's who votes. Green owns 97.6% of the Class B shares, which carry ten votes each. That's 50.2% of the total voting power. The repricing needs a majority of votes cast. Jeff Green can approve the repricing of his own options by himself.

The proxy doesn't hide it. In its own words, the board "recognizes that approval of this proposal may benefit our executive officers and their successors." There's no condition requiring approval from shareholders who aren't Green. There's nothing in the proxy saying he'll sit it out.

And here's who's off the list. To get the new price, an option holder must still work there and must not have received notice of termination by October 19. The roughly 575 people cut in September got their notice before the proxy was even mailed. Unless the board makes an exception, which the proxy allows but doesn't promise, the 2,342 employees who stayed get repriced and the people who were let go keep their options at the old strikes. At $14 a share, that's another way of saying they keep nothing.

The company that told the open internet it was the honest broker is about to give its CEO a cheaper strike price with his own votes, while the people it laid off watch from outside the room.

Now the comparison. Magnite has one share, one vote. Michael Barrett owns about 403,000 shares, a rounding error on the register. If Magnite ever wanted to reprice anything of his, it would need shareholders who aren't him.

Keep reading free with your email: who runs each company and what each CEO is paid, where the insiders put their own money, the audit committee that left in eleven days, the agency audit The Trade Desk says it never failed, and why Apple scares a DSP more than an SSP. ADOTAT+ members also get the 3% growth that wasn't quite 3%, the Nevada move that makes Monday's vote nearly bulletproof, Magnite's own 44% problem, and what each company has to prove next.