Part IThe Story
The trades went through over three days, between Monday, March 2, and Wednesday, March 4, 2026, at prices that would have looked like a typo eighteen months earlier: about $23 a share, then $24, then $25. When they were done, Jeff Green owned roughly six million more shares of The Trade Desk than he had on Sunday. He had spent $148.1 million on them. It was his own money, not the company's and not a board-authorized buyback, and it went through the same mechanism he has spent his adult life describing as the only reliable judge of what anything is worth: a market, clearing, in public, at a price.
The week before, the stock had touched a six-year low. Its fifty-two-week high was $91.45. Through 2025 the company's shares had lost about 68 percent of their value, and in January the chief financial officer's chair had been handed to an interim. Analysts' mean price target had fallen from $130.94 to $53.48. The company Green co-founded in 2009 with seven people and $2.5 million had been added to the S&P 500 in July 2025. By the following winter the market was pricing it as though the founding thesis, that the open internet would eventually beat the walled gardens, had been tested and found wanting.
On Thursday he posted to LinkedIn. The post ran three sentences.
Over the last few days I made the biggest purchase of my life. I bought ~150 million dollars of TTD stock. Tomorrow, I'll explain why.
It drew 1,314 reactions and 78 comments. One commenter, Adam Heimlich, wrote: "Me too, Jeff! Mine was a dinner in San Francisco, though." Another commenter offered three words of governance advice: "Hire a CFO!"
The explanation came the next day. Green did not give it to the Wall Street Journal or to Bloomberg. He published it in The Current, a publication owned and operated by The Trade Desk, Inc. That choice is worth holding onto, because Green's story keeps circling one problem: who gets to grade the homework. He built two exchanges on the premise that buyers deserve to see what they are paying for. He walked out of the church of his childhood with a 900-word letter accusing its leaders of not being honest about their finances. For more than a decade he has accused Google of representing the buyer and the seller while also owning the goods. Now, at the lowest point of his public life, he had bought the market's verdict on his company at a steep discount and announced that he would explain his reasons tomorrow, in a magazine he owns.
None of this makes the bet foolish or the convictions false. It shows how conviction looks from the inside, and how it looks from everywhere else.
By the Numbers
The Trade Desk, From Seven Men to the Open Internet
$2.5MFirst financing round, March 2010, for a seven-person company in Ventura
$2.44BRevenue in 2024, with $393M in net income
33Consecutive earnings periods meeting or beating expectations before the Q4 2024 miss
95%Customer retention, sustained for 27 straight quarters as of 2020
20MAd opportunities Kokai analyzes every second, per Green
$148.1MGreen's personal open-market purchase of TTD stock, March 2–4, 2026
90%+Share of his wealth pledged to give away under The Giving Pledge, 2021
A Market Is a Machine for Settling Arguments
The fullest account of where the idea came from sits in two headlines. In August 2018, the month Bloomberg's wealth reporters calculated that his stake had made him a billionaire, the wire ran the story as "Ad Man Inspired by Goldman Becomes a Billionaire." Investor's Business Daily, which described the company as valued at $10 billion at the time, put it more plainly. Trading stocks didn't just make Green richer, the paper said. It gave him the idea. In buying and selling equities he saw what data and analytics had done to one market, and he started asking which other markets had not yet been rebuilt that way.
Advertising was an obvious answer and an unflattering one. In the first decade of the century, an online ad was usually bought the way a used car is bought: through a relationship, at a negotiated price, with little information on the buyer's side and an asymmetry everyone involved preferred not to discuss. Ad networks bundled inventory and marked it up. Publishers seldom knew what their pages were really worth. Advertisers seldom knew where their money went. It had the shape of a market without doing the work of one.
Green's first answer was AdECN, which his company's leadership page describes as "the world's first online advertising exchange." The claim is promotional, but the architecture behind it was real. The pitch was to take the logic of an electronic communications network, the kind of venue that had recently pulled equity trading out of the pits and onto servers, and apply it to ad impressions: open bids, real-time clearing, a price set by many participants instead of a few.
Raising money for it taught him something he still quotes. Years later, after The Trade Desk had built its own venture arm, he recalled pitching Intel Capital for AdECN and finding the firm "wrestling with something i found fascinating." (He writes his LinkedIn posts with a lowercase first person, as though capital letters were a markup he declines to pay.) "Are we here to be the best venture capitalists," the Intel people were asking themselves, "or are we here to invest in companies that make our core vision real?" Many people would have heard that as an investor's private worry. Green heard a question every institution eventually has to answer, which is whether it serves its stated purpose or its own interest. The question returns in nearly every chapter of his life after that.
Microsoft bought AdECN in 2007. The price has never been part of the public record Green has emphasized, and he has emphasized many things. What mattered more than the amount was where it put him. For about two years he ran the exchange as Microsoft's entry into real-time ad trading, inside a company that was also, unavoidably, a seller of its own media. From there he had a close view of the conflict he would spend the next fifteen years calling disqualifying.
Microsoft also introduced him to Dave Pickles. The public record on Pickles is thin. He met Green at Microsoft after the acquisition, became co-founder and chief technology officer of the company they built next, and has shared nearly every institutional honor with Green while giving almost no interviews of his own. In a story where one founder speaks constantly, the other's silence is a fact too. Exchanges need a counterparty.
Seven Men in Ventura
On March 23, 2010, Peter Kafka of AllThingsD, a Wall Street Journal property, ran a short item headlined "Another Ad Exchange Player." The word another deserves attention. "Last fall Jeff Green left his job running AdECN," Kafka wrote. "He didn't go far." Green was building a start-up "designed to help marketers buy advertising from exchanges like the one he left." He had raised about $2.5 million. Founder Collective and Roger Ehrenberg's IA Ventures led the round. The angels were Josh Stylman, a co-founder of Interpublic's Reprise Media, and Jerry Neumann, a veteran of Omnicom.
These first investors were not tourists. Ehrenberg's firm specialized in data, and its whole thesis was that information plus computation would remake industries that still ran on relationships. Stylman had built a search-marketing agency inside one of the big holding companies and sold it there, so he knew the agency business from its plumbing up. Neumann had come out of Omnicom and would later become one of the more rigorous thinkers on venture economics. More than a decade on, when the podcast host Patrick O'Shaughnessy had his first conversation about pure technology investing, Neumann chose The Trade Desk as the first case study. The money behind the company was unusually well informed about the problem it was trying to solve.
Kafka was skeptical in the way trade reporters of that era had to be. "It's still unclear if ad buyers and publishers will embrace the exchanges," he wrote. "But investors love the notion, as well as the idea that the big ad holding companies, or perhaps even Google, will want to buy a DSP or three. So the market for this stuff is beginning to get a bit frothy." Then came the detail that matters more than anything else in the piece: "That's one of the reasons Green is reluctant to use the word DSP to describe his seven-man company."
A demand-side platform was the new category, software for advertisers to bid across the exchanges. Everyone was building one, and everyone was building one in order to be acquired. Green was a man who had just been acquired, had seen what acquisition did to an independent venue, and did not want to be filed under the year's fashionable acronym. He would not describe the product until it worked. "Given that his team is still building the product and that he doesn't want to talk about it until it's up and running," Kafka noted, "we'll have to use that term for now."
The company was based in Ventura, California, on the coast about 70 miles northwest of downtown Los Angeles and more than 350 miles south of the Sand Hill Road offices whose money he had partly declined to court. It is a beach town with oil derricks in the hills and a mission at its center. Nearly a decade later, in July 2019, the Ventura County Star ran a story with a headline that read like a homecoming: "Ventura's The Trade Desk returns to where it all started." In November 2024, when the company announced its most ambitious and most controversial product, a streaming-television operating system meant to sit inside the screen itself, Green named it Ventura.
Choosing a Side
The choice that defines The Trade Desk was made early, and every product launch since has been an argument for it. Green would represent the buyer and only the buyer.
When O'Shaughnessy introduced him on Invest Like the Best, he framed the conversation around "the parallels between The Trade Desk and an equity exchange" and "why Jeff chose to align with ad buyers and not sellers." Green was careful about the analogy. The Trade Desk is not the exchange; he had already built one of those and sold it. It is closer to the institutional broker's trading terminal, the tool a fund manager uses to reach every venue at once and get the best execution. Its value depends on its neutrality. If the terminal's maker also owned a stock exchange, a fund, and half the listed companies, no manager would trust its routing.
That was his indictment of Google, repeated for fifteen years in many forms. A later compilation of his remarks renders it as: "You cannot effectively represent the buyer, represent the seller, and own the largest media properties all at the same time." In his essays for The Current, the same argument becomes a self-description. The Trade Desk is sustainable, he wrote around the launch of OpenPath, because it represents the buy side objectively, does not own media, and is aligned with advertisers rather than trying to serve every side of the market at once.
It sounds like a virtue. Commercially, it is a constraint. Google earns money on every side of a transaction. Amazon owns the store, the shopper data, and the shelf space for ads. Meta owns the feed. The Trade Desk owns nothing except the software and the trust of the people paying for it. It takes a fee on what advertisers spend through the platform, so its incentives run toward advertisers spending more, spending well, and continuing to come back. In 2020 the company reported customer retention above 95 percent for 27 consecutive quarters, the kind of number that follows from alignment.
The constraint also forced inventiveness. A company that cannot own content has to persuade the companies that do. In 2012 Facebook made The Trade Desk one of the early partners in Facebook Exchange, its real-time bidding platform. In 2017 The Trade Desk built connected-TV buying and measurement into the platform at a time when "programmatic television" still sounded like a contradiction, and it bought the marketing-insights firm Adbrain. In 2018 it went into China, giving marketers programmatic access to Alibaba, Tencent, and Baidu, a deal Wall Street loved and few Western ad-tech companies attempted. In 2019 Amazon opened Fire TV inventory to it. In March 2024 NBCUniversal made its Paris Olympics broadcast available to automated buyers through The Trade Desk, which one trade publication called "the crème de la crème" of programmatic. In May 2024 Netflix, which had spent most of its existence without ads, named The Trade Desk among the first partners outside its original ad-tech arrangement.
Each of these was the same move: a neutral buyer's terminal persuading a content owner that the market would pay more than the walled garden would.
Trust, Rebuilt at Eighteen Dollars
On September 21, 2016, The Trade Desk went public at an initial price target of $18. The Wall Street Journal had reported two weeks earlier that the plan valued the company at $573 million, which was small enough to be notable. Ad tech had become something close to a dirty word on Wall Street. The previous wave of public companies in the sector had disappointed, and the category had a reputation for opacity, arbitrage, and revenue that disappeared on inspection. Green said at the time that he hoped the offering would rebuild "trust" between ad-tech companies and investors. It was an unusual thing for a founder to say on IPO day. Most founders talk about opportunity. Green talked about credibility, as if he were taking an industry public along with his company.
The company had raised about $120 million in venture capital by then, and its S-1 showed triple-digit revenue growth and, rarer still in ad tech, profits. On opening day the market treated it as a vote of confidence.
Then he did what almost no growth company does. He met or beat expectations thirty-three consecutive times.
Revenue rose 52 percent to $308 million in 2017, then reached $477 million in 2018 and $836 million in 2020. In 2022 advertisers spent $7.8 billion through the platform, and revenue passed $1.5 billion. By 2021 the market capitalization had reached $40 billion. Forbes had named it one of America's most promising companies in 2015. Fortune began putting it on lists of the fastest-growing companies, the best workplaces, and the Future 50. Green and Pickles collected Ernst & Young's Entrepreneur of the Year honors. A newsletter for an angel syndicate, writing after the first-quarter results in May 2022, called Green "the greatest ad tech operator this industry has ever known."
The streak shaped the company and the man in ways that became visible only when it ended. Thirty-three quarters is more than eight years. Over that span Green turned from an executive into a figure, someone conference audiences treat less as a speaker than as an oracle. At ATS London in 2024 he went on stage for what the organizers billed as "one of the most highly-anticipated sessions of the year" and said: "The best of the internet is in the open internet." It is closer to a creed than a sentence of analysis, and the room received it that way.
In the fourth quarter of 2024 the company fell short of analyst expectations by about two percent. Quarterly revenue was $741 million, and full-year revenue had grown 26 percent to $2.4 billion. By the standards of nearly any company, those were very good numbers. In February 2025 the stock fell more than 32 percent.
That kind of drop comes from a story breaking, not from a two-percent miss. The market had been paying for invulnerability, and invulnerability, once disproven, is worth nothing.
Remembering What You Told Me
Cookies are the oldest and least elegant answer to a basic question of the commercial internet: who is this? A cookie is a small file a browser keeps so that a site, or a party the site invites in, can recognize the same visitor later. The whole advertising economy of the open web rested on that file, on the ability to say this is the same person who saw the car ad yesterday, without which frequency capping, measurement, and targeting stop working.
When Google announced, years ago, that Chrome would stop supporting third-party cookies, the entire independent ad-tech sector faced the possibility that the ground under it would disappear. Green's response was to build a replacement and then give it away.
The first version, Unified ID, came out in 2018 alongside an AI forecasting engine called Koa. The second, Unified ID 2.0, launched in mid-2020. It replaced the cookie with an encrypted, hashed identifier based on the one piece of personal data almost everyone already hands to the services they use, their email address. Then the company open-sourced it. Disney adopted it, and so did Paramount. Procter & Gamble and Unilever, two of the biggest advertisers in the world, signed on. Warner Bros. Discovery followed. Salesforce and Snowflake integrated it. A European version, EUID, was built on the same open-source base with data walled off so nothing could pass between the United States and Europe, which Green predicted "will prove to be the most GDPR-compliant technology in market."
At Cannes in June 2023, Ronan Shields of Digiday asked him whether an email address was really as benign as he kept saying. Green was impatient, the way he gets with questions he considers settled.
"I don't think there is that much debate about it," he said. "I think sometimes, there's a desire to inject controversy, but I don't think that that's very controversial." Apple personalizes through the Apple ID, which is an email. Google personalizes through its sign-on. Facebook personalizes through its login and the social graph. "None of them would be using an email address if they felt it wasn't a privacy-safe methodology."
Then he said the most revealing thing he has said about what his company actually sells:
— Jeff Green, to Digiday at Cannes Lions, June 2023The truth is an average consumer needs something to anchor their settings on. So I need some way to remember what you told me, including the opt-out, and if I can't use a cookie — and cookies are not the best way to do it — there's a much more persistent way to remember somebody's settings.
Some way to remember what you told me. Strip away the auctions and the AI, and the industry Green helped build is a machine for remembering who people are. The headline Digiday ran that week was characteristically blunt: "I think it's a strategic mistake for Google to get rid of cookies in Q1 2024." Green did not believe Google's deprecation was about privacy. Elsewhere he has said it always seemed driven by antitrust fears. He went further and predicted that "you can expect Google to exit the open internet at some point, simply because strategically, it's good for them and reduces their regulatory risk."
Identity is the whole game. It is also a problem whose difficulty a reporter learns the moment the reporting begins.
The Other Jeff Greens
Type the name into an encyclopedia and the first thing returned is not a person but a disambiguation page, the internet admitting it cannot tell who you mean. "Jeff or Jeffrey Green may refer to," it begins, and then lists them.
There is a Canadian multimedia producer born in 1956. There is a video-game journalist born in 1961, the last editor-in-chief of Games for Windows: The Official Magazine, who later hosted a podcast called Branching Narratives and remembered his early career as rolling into the office at 10:30 to play '90s PC games. There is an English comedian, born in Chester in 1964, who lives in Melbourne. There is a Kentucky state senator who died in 1997, a Conservative council leader from the Wirral, and a stock-car driver. There is a historian. And near the bottom, between the basketball player and the historian, there is "Jeff Green (businessman) (born 1977), American businessman, co-founder of Trade Desk Inc."
The basketball player keeps coming up. The Boston Celtics drafted him fifth overall on June 28, 2007, the same year a different Jeff Green sold the first ad exchange to Microsoft, and traded his rights the same night to Seattle in the deal that brought Ray Allen to Boston. He was a trade asset within hours of being drafted. He went on to play sixteen seasons for twelve franchises, one of which no longer exists. A routine physical in 2011 found an aortic root aneurysm, and in January 2012 Dr. Lars Svensson opened his chest at the Cleveland Clinic. Green later described the nine-inch scar, and how, recovering in Waltham, Massachusetts, he was the only patient under sixty and was lapped on the treadmill by people who never stopped smiling. In June 2023 he won a championship with the Denver Nuggets. By then he had played with 240 different teammates, and Rob Mahoney of the Ringer worked out that the entire history of the NBA, back to the league's first basket in 1946, could be traced within six degrees of Jeff Green. The journeyman had become "a nexus point for the game itself," a node through which everything connected.
There is a Bloomberg News reporter named Jeff Green who, on February 4, 2026, marked forty years as a paid journalist. He had decided to be a reporter in third grade. He started on a typewriter with a jar of rubber cement. Looking around his industry, he wrote that "the headlines on publications ceasing operation are staccato, and so many more die in silence." His near-namesake in Ventura has argued for years that without an efficient advertising market, independent journalism cannot survive.
There is a grandfather in Glenview, Illinois, who on June 6, 2022, submitted a public comment to the Securities and Exchange Commission supporting a rule that would require corporations to disclose climate risk. "This rule would level the playing field," he wrote, "by eliminating any competitive advantage created by a lack of climate transparency." The sentence could have come from an essay in The Current. It was signed Jeff Green.
One letter away there is Jeff Greene, the Los Angeles real-estate man who in 2006 talked with John Paulson about the housing bubble, bought credit default swaps, and, by his account to Forbes, was up 1,400 percent in eighteen months, a fortune made on the conviction that a market had mispriced itself. An Amazon author page under the name sells Pluto: The Evolutionary Journey of the Soul, which is a different kind of guidance altogether.
None of these men is our subject, and that is exactly the point. The machine Green built exists to tell them apart: to know which Jeff Green opened the email, which one watched the streaming show, which one opted out and should be left alone. Ad tech's central problem is the disambiguation page. Green's proposed answer was to anchor each person to the most mundane thing they own, an inbox, and remember what they said. A reporter with nothing more than a search bar learns quickly how hard that problem is.
Resignation
On Monday, December 20, 2021, the man the Salt Lake Tribune called the wealthiest native of Utah sent a letter of about 900 words to Russell M. Nelson, president of the Church of Jesus Christ of Latter-day Saints.
Green had been raised in the faith. He served a mission, the period of proselytizing that young Latter-day Saint men are expected to give the church, and he graduated from Brigham Young University. He had stopped "believing, attending, or practicing," he wrote, more than a decade earlier. The letter made it formal. He asked for his records to be removed. He was 44 years old, divorced, the father of three children, and, by Forbes's estimate, worth $4.9 billion.
"Although I have deep love for many Mormons and gratitude for many things that have come into my life through Mormonism," he wrote, "I have not considered myself a member for many years, and I'd like to make clear to you and others that I am not a member." Then came the indictment:
— Jeff T. Green, letter to LDS Church President Russell M. Nelson, December 20, 2021The church leadership is not honest about its history, its finances, and its advocacy. I believe the Mormon church has hindered global progress in women's rights, civil rights and racial equality, and LGBTQ+ rights.
He pointed to the church's reported "more than $100 billion in assets." He wrote that its members, "often poor," give "expecting the blessings of heaven," and that "the church has exploited its members and their need for hope to build temples, build shopping malls, and cattle ranches… rather than alleviating human suffering in or out of the church." Most members, he allowed, are "good people trying to do right." The institution, he concluded, "is actively and currently doing harm in the world."
Read the letter beside his decade of essays on ad tech and the vocabulary carries over almost unchanged. An institution that holds its members' money and will not show its books. A body that grades its own homework. Opacity presented as stewardship. The man who would later write that "walled gardens ask advertisers to trust their internal metrics without allowing independent verification" had turned the same lens on the institution he grew up in. It may be that the faith gave him the habit of demanding an accounting. It may be that the habit is what ended the faith. The record does not say which came first.
He did not leave alone. Eleven members of his family and a friend resigned with him. His sister, Jennifer Gaerte, told the Tribune she had once had "that picture-perfect Mormon family," until a death in her husband's family pushed her "into survival mode," her husband stopped attending, and other children threw rocks at hers. When she asked a church leader to release her from her calling in the Young Women organization, he warned that she would become inactive. "If you won't release me," she said, "I'll release myself." A cousin, Doug Whittemore, had refused to serve a mission at all. "Something was not clicking for me intuitively," he said.
Green attached a check to the letter. His family foundation's giving arm, which he named Dataphilanthropy, would make its first major gift: $600,000 to Equality Utah, with almost half going to a new scholarship program for LGBTQ students in the state, including any who "may need or want to leave" BYU, his alma mater. "We made this investment sizable and publicly," he wrote, "to send a message that Equality Utah isn't going anywhere."
Sizable and publicly. The month before, he had signed the Giving Pledge and committed to give away at least 90 percent of his wealth. The foundation was built to measure itself. At California State University Channel Islands, near Ventura, it had since 2020 made gifts and pledges of more than $3 million for peer-mentorship programs and scholarships, and Green described the work the way he describes an ad campaign. "Not only have we seen significant improvements in terms of lowering drop-out rates," he said in 2022, "but we also get really positive feedback from embedded student peer mentors." His company had hired 27 of the university's graduates. On May 21, 2022, the school gave him an honorary Doctorate of Humane Letters. "I was floored," he said. What he admired in its graduates, he said, was "a highly developed sense of curiosity," the need to know "how things work and why things are organized the way they are."
It is a good description of a missionary who stopped believing. It is also a good description of a trader.
Media for Humankind
In September 2019 The Trade Desk ran an advertising campaign for itself. That is unusual for business-to-business software, which tends to market through sales teams and trade-show booths. The campaign was called "Media for Humankind." It cast the company as the transparent alternative to Google and Facebook, and the trade press noticed which ribs it was aimed at. "The Trade Desk's new ad campaign pokes Google and Facebook in the ribs," one headline read. Another said the company was focusing "on people, not programmatic."
A former missionary knows how to give a campaign a name. Humankind is a word from sermons, not from software. That summer the company also argued publicly with Google over a transparency initiative that would let buyers see every party selling or reselling a bid request, which is the ad-tech equivalent of following the money through each intermediary's hands. The Trade Desk had already built that visibility into its own platform in 2017.
In 2022 it launched OpenPath, a direct route from buyers to publishers' inventory that skipped the resellers. Condé Nast, Reuters, and the Washington Post signed up. In 2025 it introduced OpenSincera, a free API built on data from Sincera that measured the health of the digital ad supply chain, and Deal Desk, a tool for managing direct deals that one trade headline described as an attempt "to make deal IDs not terrible." In January 2026 it announced OpenAds, a transparent auction framework, with the Guardian and Hearst among its supporters. The word Open kept appearing in product names the way Saint appears in the names of mission towns.
The evangelism worked, up to a point. The industry treated him with reverence, and the reverence eventually drew its own critics. Pesach Lattin, who writes the trade newsletter ADOTAT, put it with relish: "The adtech industry doesn't interview Jeff Green. It lotions him up and asks if he'd like a warm towel." Green, he wrote, walks into rooms where "grown adults in lanyards lose basic motor function." Lattin called him "Moses of the open internet," which is mockery, though the trade's affection for him was real enough to make the joke work.
Green's own version of the gospel had sharpened by 2026. "The open internet was never Google's to lose," he wrote in a piece for Business Insider, assisted by his chief marketing officer, Sarah Gavin. "Let me explain..." He had been "pretty heads down," he admitted, and hadn't been able to write as much as he wanted.
There is always more to explain.
The Middle
The open internet, as Green describes it, depends on a crowd of intermediaries: agencies that plan and buy, verification firms, data providers, resellers, exchanges, the DSP itself. His whole career has been an effort to squeeze the bad middlemen out of that chain while keeping the good ones. The trouble is that the people he needs most are the ones being squeezed.
Agencies are The Trade Desk's largest customers in practice. They hold the clients' budgets and operate the platform's seats. They have also spent fifteen years watching their margins shrink as advertisers paid them less and as Google, Meta, and Amazon offered to do the work for free inside their own walls. Green saw this, and in a 2025 LinkedIn post he defended them in terms that read almost like a rebuke to the advertisers paying the bills:
"Agencies have to develop better business models and brands have to pay for quality. The tendency for brands and marketers to pay agencies as little as possible is costing them way more than they save. Value and incentives are bigger vectors than price alone."
Agencies did not necessarily return the favor. In November 2025 AdExchanger reported that Omnicom had allegedly moved part of its third-quarter spending from The Trade Desk to Amazon, which runs exactly the kind of closed model Green has spent his career opposing, one where the ad platform, the store, and the shopper data belong to the same company. A buyer's terminal depends on buyers choosing it. They can also choose against it.
There was trouble inside the platform as well. Kokai, launched in June 2023 with heavy investment in distributed AI and deep learning, was supposed to be the next stage of the company. By the February 2026 earnings call Green was acknowledging the platform's complexity problems. In December 2025, a year after its last major reorganization, the company laid off staff again. Headcount had reached about 3,900. In January 2026 Tahnil Davis was appointed interim CFO, and the stock fell 7 percent to $34 that day. "Tahnil is an exceptionally strong operator and leader who understands our business inside and out," Green said, which is what CEOs say.
Then, at Possible in Miami Beach, Green sat across from Michael Kassan, a dealmaker who for decades has made a living standing between buyers and sellers of media, and said the thing about the middle that he had been circling for years:
"You can't keep taking money out of the middle, providing a higher level of service in a more complicated ecosystem, and do that for less and less."
He was talking about the agencies. He could have been talking about his own company, which charges a percentage to stand between advertisers and the open web and was being asked by a market convinced that "software is dead" to justify that percentage again.
Tomorrow
The essay appeared on March 6, 2026, under the headline "Why I spent $150M on my own company???" The three question marks suggest a man who had already heard the question many times.
"In short," he wrote, "I'm putting my money where my mouth is."
He gave three reasons. The first was AI. The company had been building machine-learning tools "for more than a decade — long before the recent AI hype phase," and Kokai could analyze "20 million ad opportunities every second, each with thousands of variables." He wrote out an equation: AI + objectivity + 16 years of learning + massively undervalued data + complexity = a bigger TAM and bigger share for the objective leader. The word objectivity appeared three times in two paragraphs. The second was inventory. Chatbot placements and sponsored shopping listings, the new search, would fragment the way connected TV had, and fragmentation was where The Trade Desk did best. The third was a heading in bold:
— Jeff Green, The Current, March 6, 2026Wall Street is wrong. Not about everything, but about one thing.
The bear thesis, he wrote, was that tools like Claude would let anyone rebuild Salesforce or The Trade Desk from scratch. "AI changed a lot, but it didn't evaporate integrations, moats, trust, customers and tools — let alone people."
For a few days the market agreed. Reports surfaced late that Wednesday that The Trade Desk was in early talks to manage advertising sales for OpenAI. Combined with the Form 4 filing, the news produced an 18 percent rally to $29.79. One analytics service, describing the setup, noted that the stock was still in a "downtrend," below its declining fifty- and two-hundred-day averages, and that the risk-reward ratio favored the downside. Lattin published his column on March 12: "A meditation on faith, stock prices, and the surprisingly thin line between founder conviction and a hostage statement." Green had blocked him on LinkedIn after Lattin tagged him with questions. "The prophet of the open internet," Lattin wrote, "has a very closed LinkedIn."
Both readings can be true. A man can be right about markets and still want to choose his own venue. He can demand that Google open its books and still decline a newsletter writer's questions. He can describe himself as the objective leader and still publish his strongest case in a magazine his company owns, with a disclosure line at the bottom. Conviction is not objectivity. Every exchange Green has built depends on the difference, and he knows it better than his critics do. That is why he bought the stock rather than just writing about it. A purchase is a statement backed by money.
On the Briefcase podcast, around then, he talked about an interview in which someone suggested to Mark Zuckerberg that Facebook's success had been obvious, and about how much he liked that Zuckerberg got irritated.
— Jeff Green, on the Briefcase podcast, 2026It is not obvious. You can fail. There will be times when no one believes in it except for you. And if in that moment you're like, I can carry this, that is what it takes to win.
Months later, the encyclopedia entry for his company lists The Trade Desk under its ticker as a component not of the S&P 500 but of the S&P 600. The LinkedIn post is still up, with its 1,314 reactions. Further down the comments, a man named Richard wrote that he had bought at $46 and was down more than 60 percent. "Jeff, we are down 20% from your purchase price," he wrote. "Why did you give us hope."
Above that comment, the post still ends where it always did, on a promise that had been kept once and that Green keeps making: Tomorrow, I'll explain why.
Part IIThe Playbook
Green's career gives an unusually clean test of one idea: that in a market full of conflicted intermediaries, the participant who refuses conflict can win, provided it can survive long enough for buyers to notice. The principles below are drawn from how he built two exchanges, chose a side, gave away standards, left institutions, and paid for his convictions with his own money. Some are tactics. Others are dispositions, with costs that are just as instructive as their payoffs.
The Principles
- 1.Learn the market from the venue before you build for the buyer.
- 2.Pick a side and make refusing the other side your product.
- 3.Own nothing you might be tempted to favor.
- 4.Decline the fashionable label.
- 5.Build where the herd isn't.
- 6.Give away the standard to win the market.
- 7.Make your rival's conflict the subject of your marketing.
- 8.Defend the middlemen you depend on, out loud.
- 9.Treat a streak as a liability you are accruing.
- 10.Make your exits sizable and public.
- 11.Measure generosity the way you measure campaigns.
- 12.Back conviction with capital, and watch where you publish it.
T
The Decisions That Defined the Arc
Two exchanges, one side, and the costs of neutrality.
2007
Microsoft acquires AdECN, Green's real-time ad exchange. He runs it inside a company that also sells media.
2009
Green leaves Microsoft in the fall and co-founds The Trade Desk with Dave Pickles in Ventura.
2010
$2.5M first round led by Founder Collective and IA Ventures. Seven employees. Green declines the "DSP" label.
2016
IPO on September 21 at an $18 target. Green frames it as rebuilding trust between ad tech and Wall Street.
2019
"Media for Humankind" campaign positions TTD against Google and Facebook.
2020
Unified ID 2.0 launches and is open-sourced; Disney, Paramount, P&G, and Unilever later adopt it.
2021
Signs the Giving Pledge. Resigns from the LDS Church with a public letter and a $600,000 gift to Equality Utah.
2025
Q4 2024 miss ends a 33-period streak; stock falls more than 32% in February and about 68% for the year.
2026
Buys $148.1M of TTD stock at roughly $23 to $25 a share, March 2–4.
Principle 1
Learn the market from the venue before you build for the buyer
The Trade Desk was Green's second exchange-adjacent company, and the order mattered. AdECN taught him how a venue works: how bids clear, where latency matters, where the people running it face temptation. Running it inside Microsoft taught him how that temptation looks once the venue is owned by a seller of media. When he built a buyer's tool afterward, he knew exactly what a buyer needed protection from, because he had sat on the other side of the table.
Founders usually go straight to the customer they want to serve. Green came to the customer by way of the infrastructure, and that gave him a map of the market's conflicts that a buyer-only founder would have needed years to draw. Neumann's choice of The Trade Desk as his first case study in technology investing reflects that. It was a company founded by someone who had already seen how the system could be gamed.
Tactic: Before building a tool for one side of a market, spend time operating, or closely studying, the venue where that side transacts, so you know where its interests are most likely to be betrayed.
Principle 2
Pick a side and make refusing the other side your product
The usual platform strategy is to serve everyone and capture value from every direction. Green did the opposite. He chose the buy side and has turned down the sell side for fifteen years. As he framed it on Invest Like the Best, the decision was about trust: a buyer's terminal is worth only as much as its neutrality. The 95 percent retention over 27 quarters is what that alignment produced. Customers who trust their agent tend to stay.
This is harder than it sounds, because the other side is always profitable. Every year of refusal is a year of revenue left on the table and a year of competitors who will take it. The refusal holds up only if the founder keeps explaining it, which Green has done, sometimes to the point of exhausting his audience.
Tactic: Write down which party you will never represent, and treat every new revenue opportunity that would require representing them as a threat to the product, not an addition to it.
Principle 3
Own nothing you might be tempted to favor
The Trade Desk owns no media. That has been both a commercial weakness and the basis of its credibility. Green's case against Google, that you cannot represent the buyer and the seller while owning the largest properties, works only because his company owns none. The moment it bought a publisher, that argument would stop working.
The principle applies far beyond ad tech. Any advisor, broker, or platform that holds inventory has an incentive to steer clients toward it. Green made the absence of inventory a feature he could point to, and then built a series of products, including OpenPath, OpenSincera, Deal Desk, and OpenAds, that make the supply chain visible in a way an inventory owner could never afford to.
⚖
The Conflicted Model vs. Green's Model
How Green defines objectivity against the walled gardens.
| Dimension | Walled garden | The Trade Desk | Status |
|---|---|---|---|
| Who it represents | Buyer, seller, and itself | Buyer only | Core |
| Owns media | Yes: search, feeds, retail shelves | No | Core |
| Measurement | Internal, self-graded | Independent verification, supply-path visibility | Product |
| Identity | Proprietary login | Open-sourced UID2 | Product |
| Inventory access | Guaranteed by ownership | Must be negotiated, deal by deal | Risk |
| Agency relationship | Disintermediates | Depends on, and defends | Tension |
Tactic: Audit your balance sheet for assets that give you a reason to favor one outcome over your client's, and either divest them or disclose them before a competitor does.
Principle 4
Decline the fashionable label
In March 2010 Green was "reluctant to use the word DSP to describe his seven-man company." The market was "frothy," and investors were betting that holding companies or Google would buy "a DSP or three." Accepting the label meant accepting its exit path. Green had just been acquired and had seen what acquisition does to an independent venue. Refusing the label was a way of refusing a fate.
He also refused to describe the product until it worked. That kind of discipline is rare in a funding environment that rewards narrative. It also kept the company out of a category whose shakeout, when it came, claimed most of its members.
Tactic: When your category becomes the hottest acronym in your industry, describe your company by the problem it solves instead, so investors and acquirers cannot price you alongside the herd.
Principle 5
Build where the herd isn't
The Trade Desk is headquartered in Ventura, more than 350 miles from Sand Hill Road and about 70 miles from Los Angeles. The company has returned to the town's name repeatedly, most ambitiously with the Ventura streaming operating system in 2024. It has developed a recruiting pipeline from nearby CSU Channel Islands, hiring 27 alumni and praising their "grit and drive," which Green said is "forged early in life."
Distance from the center of an industry offers two things. One is protection from groupthink: the neutral buyer's agent was an unfashionable idea, and fewer people nearby were insisting it would fail. The other is a talent pool that big competitors overlook. The cost is isolation, and Green has paid it by becoming the company's most visible traveler.
Tactic: Put your headquarters, or at least your core team, where your competitors' recruiters aren't looking, and invest seriously in the local institutions that train the people you hire.
Principle 6
Give away the standard to win the market
When Google threatened to remove the cookie, the foundation of every independent ad-tech business, Green built a replacement and open-sourced it. UID2 is anchored to an encrypted email and built to carry a consumer's preferences, "including the opt-out," across devices. It became an industry utility that Disney, Paramount, Warner Bros. Discovery, Procter & Gamble, Unilever, Salesforce, and Snowflake adopted.
A proprietary identifier would have made The Trade Desk a gatekeeper, and gatekeepers are exactly what Green spends his career attacking. Giving the standard away kept his story consistent and made his company the natural home for a standard it no longer formally controlled. Its value as the most experienced operator of the protocol exceeded whatever it might have charged in licensing.
Tactic: When an infrastructure change threatens your whole category, build the replacement standard, release it openly, and compete on being its best user instead of its owner.
Principle 7
Make your rival's conflict the subject of your marketing
"Media for Humankind" was a business-to-business company running a consumer-style campaign with a sermon's vocabulary, aimed squarely at Google and Facebook. It worked because it did not claim The Trade Desk was better at targeting. It claimed the company was structurally different: transparent where its rivals were opaque, aligned where they were conflicted.
Pointing to a competitor's structural conflict holds up better than pointing to a feature gap. Features get copied, and structures rarely change. Google cannot stop owning search to win an argument with Green. The approach has a cost, though. A company that markets itself on objectivity will be held to that standard in everything it does, including where its CEO publishes his op-eds.
Tactic: Find the conflict your largest competitor cannot resolve without dismantling its business model, and build your positioning around it rather than around anything they could copy.
Principle 8
Defend the middlemen you depend on, out loud
Green's mission is to remove bad intermediaries from the supply chain, yet his largest customers are agencies, intermediaries under heavy margin pressure. He has defended them in public: "brands have to pay for quality," and "You can't keep taking money out of the middle, providing a higher level of service in a more complicated ecosystem, and do that for less and less."
Omnicom's reported move of spending to Amazon in late 2025 shows why this matters. An ally under financial pressure becomes a churn risk. Defending agencies' economics is not charity. It protects the channel through which The Trade Desk reaches most of its revenue, and it separates the middlemen Green wants to keep from the resellers he wants gone.
Tactic: Identify which intermediaries add real value in your value chain and argue publicly for their economics, before a competitor that benefits from their decline gets them to switch.
Principle 9
Treat a streak as a liability you are accruing
Thirty-three consecutive earnings periods of meeting or beating expectations is one of the great streaks in growth investing. When it ended with a miss of about two percent, on 26 percent annual growth, the stock dropped more than 32 percent. The market had been paying for the streak itself. Once it broke, the stock was repriced as a company without one.
The lesson is about expectations management. Every quarter of a streak raises the cost of eventually breaking it. A founder with a long run can deliberately lower guidance, widen the range, or explain in advance what an ordinary bad quarter would look like, all while the streak still buys credibility.
!
The Cost of Invulnerability
How the market repriced the streak.
| Moment | What happened | Market reaction |
|---|---|---|
| 2016–2024 | 33 periods meeting or beating expectations | Premium multiple |
| Q4 2024 | ~2% miss; $741M quarter; 26% annual growth | -32% in February 2025 |
| 2025 | Kokai complexity, layoffs, agency spend shifts | ~-68% for the year |
| March 2026 | $148.1M founder purchase; OpenAI talks reported | +18% rally, trend still down |
Tactic: In the middle of a winning streak, publicly define what a normal disappointing quarter looks like, so that when one arrives the market reads it as information, not as a broken promise.
Principle 10
Make your exits sizable and public
Green's resignation from the LDS Church was not quiet. It was a 900-word letter to the church's president, sent together with eleven family members and a friend, accompanied by a $600,000 gift and a scholarship explicitly offered to students who might want to leave his alma mater. "We made this investment sizable and publicly," he wrote, "to send a message."
For someone with public standing, a quiet exit leaves the story to others. Green had stopped practicing more than a decade earlier. Making the exit formal and attaching a check turned a private change of belief into a public, verifiable commitment. That is the same thing he asks of the companies he criticizes: put the claim where it can be checked.
Tactic: When you leave an institution whose values you no longer share, attach a concrete, measurable commitment to the announcement, so the exit stands for what you are building, not only what you are rejecting.
Principle 11
Measure generosity the way you measure campaigns
The foundation is named Dataphilanthropy, and the name is accurate. At CSU Channel Islands, the more than $3 million in gifts and pledges since 2020 went to peer mentorship and scholarships, and Green reports results in drop-out rates and feedback from the student mentors embedded in the program. He has said he wants to scale what works to other campuses. Philanthropy is treated like a campaign: target, measure, optimize, then expand.
The Giving Pledge commits him to giving away more than 90 percent of his wealth. At that scale, the difference between giving and giving effectively is measured in large sums. Green treats it as an optimization problem, and he keeps the work close to home, near Ventura, where results can be observed.
Tactic: Apply your operating discipline to your giving: choose one measurable outcome, fund a pilot near enough to watch, and scale only what the data supports.
Principle 12
Back conviction with capital, and watch where you publish it
Buying $148.1 million of his own stock at a six-year low was the clearest signal Green could send. It was personal money, disclosed on a Form 4, at a price anyone could check. It made the argument more forcefully than any essay. The rally that followed, helped by reports of OpenAI talks, showed that the market still listens when founders put money behind their claims.
Then he published the explanation in The Current, which his company owns. ADOTAT's criticism stung because it applied Green's own standard to him: the man who attacks self-grading published his case in a venue he controls and blocked a critic who asked questions. The purchase was costly and therefore credible. The choice of venue gave critics an opening. The lesson is not to avoid either one, but to recognize that a founder known for objectivity gets judged on both.
Tactic: When you put personal capital behind your thesis, explain it somewhere you don't control and take hard questions there, because the credibility of the money depends on the credibility of the explanation.
Free playbook
Get The Business Model Playbook
58 business models, one visual page each: how the money flows, the metrics that matter, and who runs it. Free when you join the Faster Than Normal email.
Free. No spam. Unsubscribe anytime.
Part IIIQuotes and Maxims
In their words
— Jeff Green, on pitching Intel Capital for AdECN, LinkedIn, 2026Are we here to be the best venture capitalists or are we here to invest in companies that make our core vision real?
— Jeff Green, LinkedIn, 2025The tendency for brands and marketers to pay agencies as little as possible is costing them way more than they save. Value and incentives are bigger vectors than price alone.
— Jeff Green, quoted by Lou Paskalis, 2025Transactions are done at an impression level because that's the most granular level we can transact on. That does not mean that the outcome has to be based in the impression level.
— Jeff Green, on CSU Channel Islands graduates, 2022It's so important to be curious about how things work and why things are organized the way they are.
— Jeff Green, The Current, March 6, 2026AI changed a lot, but it didn't evaporate integrations, moats, trust, customers and tools — let alone people.
Maxims
- Represent one side. A broker who serves the buyer, the seller, and itself serves no one well.
- Own nothing you could favor. Owning no inventory is what makes an objectivity claim believable.
- Refuse the hot acronym. Accepting a fashionable label means accepting the exit that comes with it.
- Give the standard away. The best operator of an open protocol often captures more value than the owner of a closed one.
- Attack structure, not features. Competitors can copy a feature; they cannot copy freedom from their own conflicts.
- Remember what they told you. Identity, done right, exists to keep a person's preferences, including the opt-out.
- Price the streak. Every quarter of perfection adds to the cost of the first imperfect one.
- Exit in public. A departure tied to a measurable commitment says what you stand for.
- Measure the gift. Generosity at scale is an optimization problem, and you should treat it like one.
- Pay for your conviction, then submit it to questions. A purchase proves belief; letting critics challenge your explanation proves objectivity.
