Part IThe Story
The Founder Abstains
On December 12, 2004, the board of directors of PeopleSoft, Inc. met to approve the sale of the company to Oracle Corporation for approximately $10.3 billion, or $26.50 a share. It was the last of five offers Larry Ellison had put on the table over eighteen months, a span that included price cuts, price hikes, a poison pill, a Justice Department antitrust suit, a federal trial, a ruling from the European Commission, the firing of a chief executive, and a public enmity between two software companies that had long since stopped being only about software. The motion carried. The man who had founded PeopleSoft seventeen years earlier did not vote for it. He owned more than 5 percent of its stock, had chaired the company since its incorporation in 1987, and had returned as chief executive barely ten weeks before. He did not vote against it either. David Duffield abstained.
The record does not say where he sat or what he looked at while the others raised their hands. It does preserve what he wrote the next morning, in a letter to some 11,200 employees.
— David Duffield, letter to PeopleSoft employees, December 13, 2004This is a sad day for me, and I'm sure an equally sad day for you.
The letter praised the staff as "hard working and loyal." It reported that he had already begun calling other technology companies to ask them to hire whoever Oracle let go. Then, according to the San Jose Mercury News, he singled out the employees who "may have difficulties finding rewarding job experiences" and offered them "my sincerest apologies for not figuring out a different conclusion to our 18-month saga."
That is an odd sentence for a man about to become very rich. Not figuring out a different conclusion. An engineer's phrasing: the problem had a solution, and he had not found it. The apology assumed a debt. It did not concern shareholders, who were being paid handsomely, or the board, which had voted. It concerned the people in cubicles, whom the founder believed he had failed to protect.
Eight days later, on December 21, Duffield resigned as chief executive, chairman, and director. The SEC filing devoted one sentence to it. A month after the deal closed in January 2005, Oracle cut roughly 6,000 of PeopleSoft's 11,000 employees. By April, Duffield had set up a fund that paid $10,000 grants to former PeopleSoft workers anywhere in the world who had earned less than $150,000 a year and had not found comparable work within three months. The money was for medical bills, relocation, or, as Forbes put it, simply surviving. The magazine's item opened with a line that was half compliment and half eyebrow: the former CEO "has a heart, apparently."
Something else happened that March, between the apology and the checks. A company called North Tahoe Power Tools, Inc. was incorporated. In July it changed its name to Workday.
Duffield was sixty-four. He had lost a company he loved to a man he had spent a decade publicly mocking. Within weeks he was building a new one, aimed at the same customers and the same kind of software, carrying the same conviction he had held since the 1960s: the organization exists for the people inside it, and anyone who says otherwise is a competitor. The rest of his life, before that December and after it, can be read as a long argument with that one abstention, about what a founder owes, to whom, and for how long.
By the Numbers
Six Companies, One Promise
6Business software companies founded over a six-decade career
$10.3BOracle's final price for PeopleSoft, December 2004, after five offers
3%PeopleSoft's annual employee turnover in 1997, one-quarter of the national average
$200MInitial stock gift to Maddie's Fund, then the largest ever to companion-animal welfare
$8.4BWorkday revenue in fiscal 2025, up 16.4% year over year
$371.5MSingle pledge to Cornell in 2026, the largest gift in the university's history
$400M+Personal capital put into Ridgeline, his sixth company, founded at 76
A Slide Rule in Ho-Ho-Kus
The prize was a slide rule. In 1958, Ridgewood High School in Bergen County, New Jersey, gave its top mathematics award for the senior class to a pitcher named Dave Duffield, and the award was the instrument itself, new, the kind of thing a boy who wanted to be an engineer would have used until the markings wore off. That same spring, or near it, he threw the first no-hitter in the school's history. He was co-captain of the baseball team and an honorable mention on New Jersey's All-State list. The alumni association that inducted him in 2022 listed all three distinctions in one sentence, the slide rule first.
He had been born in Cleveland on September 21, 1940, but the family landed in Ho-Ho-Kus, a small borough north of Ridgewood with a name people outside New Jersey tend to repeat twice to be sure they heard it right. His mother, Mary Duffield, taught at the Ho-Ho-Kus elementary school. Her son graduated from it in 1954, in a building where she was on the faculty and his brother, Al Jr., was a student. The usual distance between a schoolboy and his teachers did not apply to him.
His father, Al Duffield, was an engineer without a college degree. That fact appears in Duffield's 2001 oral history for the Computer History Museum, and it is the kind of biographical detail that explains a great deal without any help. A man practices a profession on talent and experience, without the credential. His son goes on to collect two credentials, a bachelor's in electrical engineering from Cornell in 1962 and an MBA from Cornell's business school in 1964, and decades later pays for the buildings in which other people's sons and daughters earn theirs.
At Cornell he pitched for the varsity, joined Beta Theta Pi, and played in a band. The band is easy to skip and shouldn't be, because it comes back. Thirty-odd years later, at a PeopleSoft pep rally in November 1996, the music came from a company band called the Raving Daves.
The brother comes back too. Al Duffield would eventually serve as chief operating officer of PeopleSoft. His retirement in 1999 marked the beginning of the company's long unraveling. The family pattern was set early and held for sixty years. Duffield did not keep work and family in separate rooms. His mother taught at his school. His brother ran operations at his company. In 2006, when a financial-services executive sat down to discuss becoming an early Workday customer, the founder brought his son Mike, then running sales. Mike Duffield later became senior vice president of go-to-market at Ridgeline. A daughter, Amy Zeifang, chairs Maddie's Fund. The Duffields would raise ten children in all, three from his first marriage and seven adopted with his second wife, Cheryl. A man who grew up with that much family inside the schoolhouse seems to have assumed every institution worth building should feel that way.
In December 2022, the Ho-Ho-Kus Board of Education announced a $10 million gift from its former student for a new gymnasium and wellness center. It was the largest gift in the board's history. "I loved my childhood in Ho-Ho-Kus," Duffield said in a written statement. The building was dedicated in September 2024. The local paper identified him not as a billionaire or a founder but as an alumnus of the Class of '54.
Four Rehearsals
Most founders get one company. Duffield's career makes more sense if you count backward from PeopleSoft and treat the first three as drafts.
In 1964, fresh from the MBA, he joined IBM as a marketing representative and systems engineer. By one account he worked out of Rochester, New York, largely serving colleges and universities. The work was physical in its way: installing operating systems on IBM mainframes at customer sites, learning how institutions actually ran their payrolls and personnel files, seeing what the machines could and could not do for the clerks who fed them. On weekends, according to Forbes, he tinkered with software to run corporate personnel and accounting operations, borrowing a program a friend had written to schedule final exams at large universities. Decades later he would say IBM was where he first understood the order of loyalties that defined everything after: "I reflect back on my days at IBM, and it was the same thing. Customers are a close second. Employees are number one."
The weekend tinkering found buyers. In 1968 he left IBM and started Information Associates, a consulting firm selling university software, exam scheduling among it, to clients such as the University of Rochester and Rutgers. In 1972 he founded Integral Systems in New Jersey to commercialize a payroll and human-resources package that had originated at Rutgers. In 1976 Integral won a major contract from the University of California at Berkeley, and Duffield moved the company across the country to Walnut Creek, in the East Bay. Stanford became a client. Integral grew into a leading vendor of HR systems for higher education. In 1979 he started a third venture, Business Software Corporation, which sold real-time HR and payroll applications to commercial enterprises.
The detail worth holding onto is the customer. For his first twenty years in business, Duffield's market was the university: the registrar, the payroll office, the employees of institutions much like the one that had educated him. When Cornell later ran its human resources on Workday, it was buying a product from a lineage that began with a borrowed exam-scheduling program.
By 1986, Integral Systems had $55 million in sales. Duffield wanted to take its mainframe HR software and rebuild it for client-server computing, the new arrangement of networked PCs and database servers that was beginning to erode the mainframe's hold on the back office. His board said no. Too risky. Forbes later summarized his view of that board in five words: "too timid for his taste." In 1987 he resigned.
In later years he described this first stretch of his career with unusual severity. He was rich by any ordinary measure, had built a company with tens of millions in revenue, and was respected in his niche. He still graded himself as a failure on the thing he cared about most.
"I aspired to have a harmonious work environment," he said, "and failed the first two or three times, and would have failed the fourth time had I not woken up and took this core values system and beliefs seriously."
A 2019 Forbes profile captured the arc in its headline: twenty years and four startups to develop a corporate culture. Very few founders treat their own successful companies as rehearsals. Duffield seems to have. Product and market had been working all along. What he kept getting wrong, by his own account, was the feeling of the place, and he did not stop until he had it.
The Mortgage
PeopleSoft began with a house as collateral.
Duffield was forty-six in 1987, an age at which most people in the software business were either running something large or leaving the industry. He and Ken Morris, the co-founder whose name appears beside his in the founding record, set up in Walnut Creek with a plan that was almost aggressively unglamorous: take the human-resources management system Integral had sold on mainframes and rebuild it for client-server networks. Duffield mortgaged his home. Seed money came to about $2 million. By one later account he funded the company out of his own pocket for its first two years. The company's only venture backing came from IBM, the employer he had left two decades earlier. George J. Still Jr. of Norwest Venture Partners joined the board.
Board composition was not a side issue for him. He had just walked out of a company because its board would not let him make the bet he believed in. At PeopleSoft he made sure the money came from people who already shared his conviction, starting with himself.
The first version shipped in late 1989. Its historical significance is easy to state and easy to underrate: it was the first fully integrated client-server HRMS application suite, meaning software that let a large company run personnel, payroll, and benefits on networked PCs instead of a central mainframe. Revenue was $1.9 million in 1989 and $6.1 million in 1990. The old mainframe rivals, including Dun & Bradstreet's software arm, Tesseract, and Duffield's own former company, Integral, never caught up. PeopleSoft went public in 1992 and that same year added a financials module covering accounts payable, general ledger, and treasury. Distribution followed in 1994, then manufacturing in 1996 after the acquisition of Red Pepper. Forbes gave a vivid example of what the manufacturing software did: it would tell a lampshade maker how to schedule production and how much raw material to buy before introducing a new model. The company that had started with personnel files was learning to count lampshades.
The architecture kept changing beneath the product. With version 8, the whole suite was rewritten as an n-tier, web-centric design called PeopleSoft Internet Architecture, so that a company's business functions could run inside a browser. Lynn Anderson's Understanding PeopleSoft 8, a technical manual from 2001, documents that rebuild and the proprietary toolset, PeopleTools, that made it possible. It is the kind of book only implementation consultants read, and it records a company willing to tear down its own foundations, once from mainframe to client-server and again from client-server to web, rather than let a competitor do it.
Around the time of the founding, during what Duffield later called the stormy period, he and Cheryl brought home a miniature schnauzer. They had first seen her at ten days old. "She melted our hearts from the first second we saw her," he said many years later, "with her sweet ways, her stubbornness, her independence, her intelligence, her spirit and her devotion." They named her Maddie. In the company's official telling, and in nearly every interview he gave about her, she was the "lighthouse" of those years: the thing in a mortgaged house that did not care about client-server architecture or the burn rate.
Hold onto her. She comes back at the end.
Have Fun, Kill the Enemy
By January 1997, PeopleSoft had 2,300 employees, and its founder worked in a cubicle. He answered his own phone and opened his own mail. At a recent all-hands meeting, Forbes reported, the fifty-six-year-old chief executive had danced the macarena in front of 500 cheering employees.
— David Duffield, Forbes, January 27, 1997If you're not having fun, go work for a competitor.
Few did. Annual employee turnover was 3 percent, one-quarter of the national average. Sebastian Grady, then thirty-two and vice president for customer services, gave Forbes the line that would follow the company for the rest of its existence: "If Dave gave us Kool-Aid, we'd drink it." He was a young executive in a company that kept making young executives. Before he was thirty-five he had seen the founder clown on a stage and had decided the clowning was sincere. The quote was meant as praise. It also carried an edge Grady may not have heard, the edge every cult-of-founder story eventually has to face.
The details from those years read today like a list of things a modern HR department would flag, and like the reasons people stayed. Duffield gave every employee a laptop. Twice he invited the entire company to spend a weekend at his house on Lake Tahoe. Later accounts describe Hawaiian shirts, beer bashes, and dogs in the office. An internal memo instructed staff to "treat fellow employees, customers . . . and the bagel delivery people with the same respect and courtesy you'd offer the late Jerry Garcia." Another directive said flame mails and offensive internet behavior were "not classy" and did not fit a culture "where we respect the gender, race, opinions, ear-lobe apparel and choice of clothing of all employees." "It's mandatory we have fun here," he said, and the joke worked because he meant it as policy.
The policy was backed by money. Employees who won outstanding-service awards could take $500 in cash or 100 stock options. Forbes reported that they invariably took the options and had not been disappointed: PeopleSoft stock had returned 115 percent a year since the 1992 IPO. Sales, profits, and headcount had roughly doubled every year since the founding. The 1996 results were expected to show $56 million in earnings on $430 million in revenue, both up 89 percent. PeopleSoft sold $500,000 software packages into a $6 billion market contested by Oracle, Germany's SAP, and the Netherlands' Baan. It held 30 percent of the human-resources segment, trailing only SAP, and Forbes judged it the company to watch.
Then comes the passage that complicates the warmth. At the November 1996 pep rally, Duffield read aloud, with obvious pleasure, a series of nasty press clippings about Larry Ellison. The employees cheered. The beer and wine came out, and the Raving Daves played. Forbes titled the piece with the slogan the scene implied: Have fun, kill the enemy.
The two halves of that sentence are not in tension, though people kept assuming they were. Duffield said so himself: "Being nice and competitive are two different things." The niceness pointed inward, toward the bagel delivery people. The competitiveness pointed outward, across the bay toward Redwood City, where Oracle was known, in the restrained phrase of one Bay Area newspaper, for its "competitive culture." A family needs someone to be against, and Duffield picked Ellison. He would come to regret how personal the rivalry became, but not before it had shaped his hiring decisions, his board's strategy, and eventually the sale price of his company.
There is a darker note in the record from this period, and leaving it out would be dishonest. In 1996, an employee left work after drinking and was in a car accident that killed two people. The company's famous parties were toned down afterward. No source records Duffield's response in detail, and this piece will not invent one. The fact belongs here because it is the cost that celebrations of the beer-bash era usually omit, and because a founder who apologized in writing to employees he could not save would have understood it as a cost.
A Man Hired for His Disdain
By 1999, he was tired.
The analysts said it before he did. "It's been obvious for a year or more that David wanted to get out," Jim Shepherd of AMR Research told CNET. "He's tired. He's done it." PeopleSoft shares had fallen 45 percent in a year. The enterprise-software market was saturated, and corporate buyers were deferring new purchases while they fixed Year 2000 problems. Duffield laid off 430 employees, a first for a company that had run on growth and loyalty. His brother Al retired as chief operating officer. Several other senior executives followed him out.
When PeopleSoft stepped down from its founder's direct leadership that year, it had nearly 7,000 employees and $1.4 billion in revenue. Duffield stayed on as chairman, with the board saying he would focus on customers and employees, the two constituencies he had always cared about most. The chief executive's job went to Craig Conway, forty-four, previously CEO of a San Jose company called OneTouch Software and before that an Oracle executive. In 2000, Duffield gave an explanation of the hire that is remarkable for its frankness. He had picked Conway partly because Conway "had a similar disdain for Larry Ellison."
Few founders would admit that a shared enemy was one of their criteria for a successor. It reads as a joke at first, then as a confession, and finally as a prophecy. Conway set about building what analysts called a more organized corporate culture. PeopleSoft had been, in CNET's words, "a casual, employee-friendly place." Conway pushed more salespeople into the field and laid out an internet strategy. The company recovered. By 2003 it had 11,500 employees and $2.3 billion in revenue. That year it agreed to a friendly merger with its smaller rival J.D. Edwards, whose software served mid-sized companies PeopleSoft's products were too large for.
Days after that merger was announced, in June 2003, Oracle launched a hostile bid for PeopleSoft.
The ensuing fight has been told many times as a business-school case in takeover defense, and the tactics deserve attention. PeopleSoft's board rejected bid after bid. Its most inventive defense was a customer guarantee, widely called a poison pill, that promised customers refunds of two to five times what they had paid if the company were acquired and their software orphaned. The provision turned PeopleSoft's customer base into a liability on Oracle's books. In February 2004 the U.S. Department of Justice sued to block the deal on antitrust grounds. In May, Oracle cut its offer to $7.7 billion. In September, U.S. District Judge Vaughn Walker ruled that the government had not proved its case. In October the European Commission reached the same conclusion.
Conway, by most accounts, was waging the war as a personal crusade. That was arguably exactly what he had been hired to do. On Thursday night, September 30, 2004, the five independent members of the PeopleSoft board voted unanimously to remove him, with Conway in the room. The next day the company announced the firing, and in the same news cycle the government said it would not appeal Judge Walker's ruling. Director Skip Battle said the board had lost "confidence in Craig's ability to lead the company." Later reporting tied the firing to statements the board concluded had misled investors. Analysts mostly believed it had something to do with Oracle as well. PeopleSoft shares rose $2.98, or 15 percent, to close at $22.83.
The board's replacement was the founder. On a conference call that Friday, October 1, Duffield, then sixty-four, was asked how long he intended to stay. "I'm here for the long term," he said.
Wall Street split on what that meant. Tad Piper at U.S. Bancorp Piper Jaffray said Duffield had "a long-term vision to look beyond the current turmoil" and would calm the employees. Steve Mader, an executive recruiter at Christian & Timbers, was blunter: Duffield "is the navigator to a new spot. And that's all." In other words, the board had brought back the beloved founder so the inevitable would go down easier. USA Today framed the question in the terms that matter for this piece: Duffield "may be PeopleSoft's savior — or the person called to preside over the company's demise."
He believed, then and afterward, that he might have saved it.
— David DuffieldI had come out of retirement into a very difficult situation with the PeopleSoft takeover, got through it, and was having a good time, frankly. We just ran out of runway at PeopleSoft. Had we had another year, maybe two years, I think we would have made it.
"Having a good time, frankly." At sixty-four, in the final weeks of a hostile takeover, with the company's independence collapsing and its founder recalled to oversee either a rescue or a wake, he was enjoying himself. That is the quietly startling phrase in the quote, more than the counterfactual. Weeks earlier the company had reported third-quarter license revenue above $150 million, $30 million to $50 million more than analysts expected. The business was working. Only the calendar had run out.
In November, Oracle raised its offer to $9.4 billion. In December it reached $10.3 billion. Then came the board meeting and the abstention.
Later Duffield described the scene that followed, when the decision was explained to employees in a conference room. "There were people crying," he recalled, "men crying — in that room." He called it the worst moment of his life. In 2010, asked by Fortune whether Oracle had treated his product well, he answered with a generosity that cost him something: "They've done a nice job. Frankly it feels good. Having spent 17 years with the company, you don't want the product to go away, or for the customers not to enjoy the fruits of your labor." More than two decades after the takeover, Oracle was still extending PeopleSoft support through at least 2037. The product outlived its independence by at least a generation.
North Tahoe Power Tools
Aneel Bhusri came to PeopleSoft in 1993 with an unusual set of credentials. He held an electrical engineering degree from Brown, had spent time in corporate finance at Morgan Stanley, and had a Stanford MBA. An engineer who had studied business, he resembled the founder more than either may have noticed at first. Bhusri rose to vice chairman and head of product strategy. When the end came, he took it as hard as anyone. "It felt very personal back then," he told Fortune, "because we were the first one for Oracle or SAP. Neither had really done big acquisitions."
Fortune reported that it took only a couple of months before Duffield and his "old buddy" Bhusri began laying the groundwork for the next venture. The company was incorporated in March 2005 under a placeholder name that sounded like a hardware store near the lake: North Tahoe Power Tools, Inc. By July it was Workday.
The idea, as Fortune summarized it, was "the software you need to manage people in a big company, but delivered through a web browser. Think of it as Salesforce.com for employee records." Workday was built from the start to live in the cloud as a single shared system its vendor ran and updated, rather than as software sold, installed, and customized on each customer's servers. Workday's own history calls it "the first true cloud application provider for HR and finance." It was the third time Duffield had bet his career on an architectural shift his incumbents dismissed. First came mainframe to client-server, which Integral's board had refused. Then came client-server to web, inside PeopleSoft. Now it was on-premises to cloud, and the incumbent he was betting against was Oracle, which now owned his old product.
Workday Human Capital Management launched in 2006. Workday Financial Management followed in 2008, along with the acquisition of an Irish integration company, Cape Clear Software. The founders interviewed the first 500 employees themselves, a practice they described as smarter than their hiring at PeopleSoft. The company stated six core values, and their order matters: employees, customers, innovation, integrity, fun, and profitability. Profitability came last, and that was not a pose. Workday lost $49.9 million in 2009, $56.2 million in 2010, and $79.6 million in 2011, even as revenue rose from $25.2 million to $68 million to $134.4 million. When it filed to go public in 2012, it reported an accumulated deficit of $329.5 million, 325 customers, 1,450 employees, and a product that shipped a new update three times a year.
One line in the S-1 explains more about the founders than any mission statement. The co-founders, who were also the co-CEOs, controlled a majority of the voting stock. Greylock held 11 percent and NEA 10.1 percent, but voting control stayed with the two men who had sat through PeopleSoft's eighteen months. One startup blogger, reading the filing, summed it up plainly: Workday's equity was "set up such that it would not be subject to any hostile takeovers in the future." Duffield had built one company that could be bought from under him. He would not build a second.
In April 2010, Fortune found him back at Workday's Pleasanton headquarters at nearly seventy, a billionaire reporting each morning to his desk. "Yes, it's an actual cubicle," the writer added, as if expecting disbelief. Asked why a company that had already raised $165 million would bother with an IPO when public markets had become such a hassle, Duffield's answer showed the old order of loyalties intact.
"It's a rite of passage," he said. "It's a special thing to be a public company." He gave a practical reason: Fortune 500 buyers trust a vendor whose balance sheet they can read. Then he gave the real one. "It's definitely positive for the employees. I'd love for people to be able to make some money here, and do good work with it, and help their families out."
Workday listed on the New York Stock Exchange in 2012. Cornell later counted Duffield as one of only two people in American history to take two companies they founded public at valuations of a billion dollars or more.
In May 2013, the New York Times asked Duffield what a company fundamentally is. His answer stated a thesis about the economy that doubled as a description of himself. "We're targeting business services," he said. "Most companies are going to services, away from manufacturing. They are becoming more people-based." Oracle and NetSuite, he allowed, "do a good job with manufacturing and inventory. They're good things, but they are less relevant in the global economy." Zachary Nelson, NetSuite's chief executive and, like Marc Benioff and NetSuite founder Evan Goldberg, a veteran of what the Times called the Oracle tribe, shot back: "Workday can't do businesses based on taking orders." The argument was framed as one about software. Underneath it was the same question his abstention had posed: whether a company is its ledger or its people.
He gave up the co-CEO title in May 2014 and became chairman, and in April 2021 he stepped down from that role too, becoming chairman emeritus and then CEO emeritus. Workday went on without him in the way he had designed it to. By 2023 it had more than 10,000 customers. In 2024 it entered both the Fortune 500 and the S&P 500. Fiscal 2025 revenue reached $8.4 billion, the company served more than 60 percent of the Fortune 500, and it counted over 70 million users across 11,000 organizations. When its then–chief executive, Carl Eschenbach, was asked what had survived the company's growth, he named the six values. "A lot has changed," he said. "But the values of the company have never changed." In 2026, Bhusri returned as CEO. The younger founder came back to the company he had started with the older one, as the older one had once come back to PeopleSoft.
The House on the Hill
The niceness had limits, and the limits were often about houses.
A house had financed PeopleSoft. Another house, on Lake Tahoe, had twice hosted the entire company for a weekend. In 2005, the year of the Oracle sale and the Workday founding, Duffield and Cheryl bought 21 acres in Bryan Ranch, a quiet development in Alamo, California, at the edge of Mount Diablo State Park. The land was endangered-species habitat surrounded by public trails. Their proposed house measured 72,000 square feet, larger than Hearst Castle and, as the local press noted, double the size of the Taj Mahal.
The reaction was predictable. The neighbors organized as Save Bryan Ranch. The environmental group Save Mount Diablo objected. Duffield cut the design by 75 percent, to about 17,000 square feet, though Seth Adams, the group's director of land programs, suggested the real number was closer to 21,000. The homeowners association's architectural review committee, realizing it had never capped house size, passed an amendment limiting homes to 10,000 square feet and rejected the plan. Duffield sued the association's board members, arguing the amendment targeted him, was unfair, and had not been properly adopted. It was overturned.
"Many of the people who are opposed have been bullied into silence," Adams told a local paper in 2008. "We're not willing to be intimidated by a bully." Steve Hill, the project manager, said the association and most neighbors were now "fully behind the project." Jim Dugdale, Duffield's spokesman, said there was no controversy left, only a tree permit for cutting down twenty-some trees. The man who had asked employees to treat bagel delivery people like Jerry Garcia was being called a bully by his neighbors over a house, and the facts on record do not settle the question in either direction.
That is useful to keep in mind. Every profile of Duffield risks sentimentality: the schnauzer, the macarena, the $10,000 checks. The Alamo fight belongs alongside them, together with the $1.2 million he and Cheryl gave to Donald Trump's 2020 campaign, a donation that sits uneasily with the inclusive memos of the 1990s for some of the people who once drank the Kool-Aid. The warm-hearted founder and the man who litigates were the same person, operating on the same principle he had stated plainly years before. Being nice and competitive were two different things. He did both, and on any given day the only question was which one was facing you.
The Duffields lived in Alamo from 2005 to 2016. The Wall Street Journal eventually covered the property's sale under a headline that said a great deal about the household: "Dog Heaven." Then they moved for good to the Nevada side of Lake Tahoe, to Incline Village. There Duffield had already chaired the local private school, the Lake Tahoe School, and helped secure it a new building. Cheryl Duffield, local press reported, helped decorate it. The couple eventually gave it $25 million. They gave $27 million more to the public school district to expand Incline High School. In April 2020 their foundation made a $350,000 grant to the Incline Village Community Hospital for its pandemic response. When fires burned through the Tahoe region in 2021, it gave $4 million to more than thirty charities, $2 million of it to the Red Cross.
Wherever he lived, he built schools for the place, which is perhaps what you would expect from someone whose mother taught at his.
Bored Silly
In 2017, at seventy-six, Duffield founded his sixth company.
Ridgeline built cloud software for investment management firms, a fragmented, conservative industry that, one Ridgeline executive later suggested, was stuck in an "ice age" of patched-together legacy systems. Duffield recognized the conditions. An entrenched incumbent stack, a customer base that tolerated bad software because switching felt impossible, and an architectural shift the incumbents were slow to make: these were the conditions under which he had built PeopleSoft out of Integral and Workday out of PeopleSoft's remains. He based the company in Incline Village, near his home, with offices in Reno, New York, and the Bay Area. His son Mike joined to lead go-to-market.
He ran it himself as chief executive until May 2024, when Dave Blair took over and Duffield became chairman. That October, Bloomberg ran a story whose headline did most of the work: the Workday billionaire was "Bored Silly" in retirement and chasing a third IPO. The phrase fits nearly everything he ever said about stopping. "I'm sort of a foot-on-the-gas kind of guy," he said once. At another point, explaining why he would not settle into leisure, he said: "Yeah, I could go rock on the back porch and do crossword puzzles — but I've got six kids, ages 9 to 16, and someone in the family should work. That's me."
The money he put behind Ridgeline was not a hobbyist's money. By September 2026 he had invested more than $400 million of his own capital and led a $250 million funding round at a $1.4 billion valuation. That made Ridgeline the third company he had built to a billion-dollar valuation, as he approached his late eighties. Clients had committed more than $750 billion in assets under management to the platform, up 50 percent in a year, and the company expected to pass $1 trillion by early 2027. Blair said the new capital was going largely to AI research and development. Asked about an IPO, Blair said he would "love" for Ridgeline to become the third of Duffield's companies to list.
Blair also mentioned, in passing, that he and Duffield meet for breakfast every week.
That is all the record says, and it is enough. A man past eighty-five, worth somewhere between ten and thirteen billion dollars depending on which index you trust, sits down weekly with the executive running his sixth company to talk about an industry most people find tedious. He had said years earlier that "great companies are usually built at a special point in time," and that you don't understand this when you're young. He seems to have spent his old age trying to stay present for each of those moments as it arrived.
To You and Your Kind
In 1994, three years before Maddie died, David and Cheryl Duffield gave $200 million in PeopleSoft stock to create a foundation in her name. At the time it was the largest single donation ever made to an organization devoted to companion-animal welfare. Its goal was stated with the bluntness of a product roadmap: a no-kill nation, meaning that no healthy, adoptable dog or cat in an American shelter would be put down for lack of a home. At the time, about five million were euthanized each year.
To run Maddie's Fund he chose Richard Avenzino, who had turned the San Francisco Society for the Prevention of Cruelty to Animals into a model for the field. Avenzino was a reformer from inside a sentimental industry who talked about outcomes the way a sales vice president talks about quota. "You get millions of dollars but the kicker is," he explained, "you have to increase adoptions, you have to decrease deaths community wide, and you have to spay and neuter a lot of animals." The first large project was in Utah, where the fund committed $8 million to a statewide adoption program led by Best Friends Animal Sanctuary in Kanab. It enlisted all 126 of the state's animal-rescue organizations and veterinary hospitals and aimed to save about 15,000 dogs and cats a year. Skeptical shelter officials said true no-kill was impossible, with too many animals, too many problems, and too few homes. Avenzino thought the fund could eventually help save three million lives a year.
Duffield had a ready answer for anyone who thought the money would be better spent on people. "99.3 percent of that money is going to human causes," he said of American charitable giving, "and I think less than 1 percent is kind of a fair contribution" for pets. It is a very Duffield argument: a figure, a proportion, a claim about fairness, and a refusal to be embarrassed. In 2005, Forbes included him in a list of "quirky" billionaire philanthropists, beside Bill Gates fighting AIDS and Gordon Moore protecting salmon. He seems not to have minded.
Maddie died of cancer in 1997. Other dogs followed her into the house: Sadie, a German shepherd; Lucy, a Brussels griffon; and inevitably another miniature schnauzer, Abigail, who appeared with him in a 1998 newspaper photograph. By 2023, Maddie's Fund had awarded more than $265 million in grants. In 2024, chaired by his daughter Amy Zeifang and holding about $260 million, it revised its approach to fund multi-site projects with grants of $200,000 to $1.5 million. The foundation he and Cheryl started in 2016, the Dave & Cheryl Duffield Foundation, carried the same commitment into new areas. After the 2023 Maui wildfires it worked with the Maui Humane Society to care for displaced pets. It also began planning Liberty Dogs, a 27-acre campus in Reno to train service dogs and pair them with military veterans living with PTSD.
The other great beneficiary was Cornell. In January 1997, the year Maddie died, he gave his alma mater $20 million for interdisciplinary science and engineering, and the resulting building, a nanoscale research facility, was named Duffield Hall. Further gifts went to the veterinary college, including the Maddie's Shelter Medicine Program and a $12.1 million Duffield Institute for Animal Behavior, and $5 million in 2020 funded scholarships for engineering students whose families were hit by the pandemic. On September 4, 2018, the College of Engineering gave him its inaugural Distinguished Alumni Award, its first ever, in 101 Phillips Hall. Phillips had housed the School of Electrical and Computer Engineering since 1955, three years before he arrived in Ithaca as a freshman who wanted to be an engineer.
In 2025 he pledged $100 million, the largest gift in Cornell Engineering's history, to expand Duffield Hall and renovate Phillips, joining the building where he received his award to the building that bore his name. Space would be added for quantum computing, artificial intelligence, robotics, and semiconductor research. In January 2026 he pledged another $371.5 million, the largest gift in the university's history, and his total commitments to Cornell rose past $520 million. The College of Engineering was renamed. Students would now graduate from Cornell Duffield Engineering, carrying, among other things, the name of a man whose father practiced engineering without a degree.
All of that, the buildings and endowments and the no-kill nation, the gym in Ho-Ho-Kus and the service-dog campus in Reno, and in a different key the apology to the employees he could not save, traces back to one moment, which he described in nearly the same words for more than twenty years.
It was during the stormy period, around 1987, when the house was mortgaged, the company existed mostly as an argument, and the money had not yet come. On one particularly good day, he picked up the ten-pound schnauzer and held her at eye level.
"I remember picking her up one day," he said, "just looking directly into her eyes; she was looking at me, and I said, 'You know, Maddie, if I ever make any money I'm going to give it back to you and your kind.'"
Part IIThe Playbook
What follows is an attempt to extract method from a career that its subject often described in terms of feeling: fun, harmony, happy employees, a promise made to a dog. The warmth was real. It also rested on a set of repeatable decisions about capital, control, architecture, succession, and loyalty that are worth studying apart from the personality. Duffield founded six companies across six decades. These principles come from the evidence of all six, with particular weight on the two that went public.
The principles
- 1.Treat your early companies as rehearsals.
- 2.Bet on the architecture your incumbents call too risky.
- 3.Fund the bet yourself to choose your board.
- 4.Write the culture down and rank it.
- 5.Put employees ahead of customers, out loud.
- 6.Point the warmth inward and the war outward.
- 7.Make equity the reward people choose.
- 8.Design the second company against the first one's killer.
- 9.Pair with a younger operator and share the title.
- 10.Own the ending you couldn't prevent.
- 11.Give like an operator, with a metric attached.
- 12.Treat retirement as a pause between companies.
D
Six Companies, Sixty Years
Each company reused the customer knowledge of the last and bet on the next architecture.
1964
Joins IBM as marketing representative and systems engineer, serving mainframe customers including universities.
1968
Founds Information Associates, selling university software including exam scheduling.
1972
Founds Integral Systems to commercialize a Rutgers-built HR and payroll package; moves it to Walnut Creek after a 1976 UC Berkeley contract.
1979
Starts Business Software Corporation for real-time HR and payroll in commercial enterprises.
1987
Leaves Integral after its board rejects client-server; mortgages his house and co-founds PeopleSoft.
1992
PeopleSoft goes public; financials module launches.
2004
Returns as PeopleSoft CEO in October; abstains as the board approves Oracle's $10.3B offer in December.
2005
Co-founds Workday (incorporated as North Tahoe Power Tools) with Aneel Bhusri.
2012
Workday lists on the NYSE with co-founders holding majority voting control.
2017
Founds Ridgeline for investment management at age 76.
Principle 1
Treat your early companies as rehearsals
Duffield's own summary of his first two decades as an entrepreneur is unsparing. He wanted a harmonious workplace, and he "failed the first two or three times." Those failures were not commercial. Integral Systems reached $55 million in sales and led its niche. What failed was the part of company-building he cared about most and found hardest to get right. Only at the fourth attempt, he said, did he take core values seriously enough to make them work.
The lesson is not that early ventures are disposable. Each of them taught him something PeopleSoft needed. Information Associates gave him a market, university administration. Integral gave him a product, mainframe HR software, and a geography, the East Bay. Business Software Corporation showed him that commercial enterprises wanted the same real-time payroll tools that universities did. By 1987 he had the customer knowledge, the domain expertise, and a clear account of his own cultural failures. PeopleSoft's speed in its first five years, from zero to IPO, came largely from those rehearsals.
Founders who treat a modest early success as their final verdict, whether for good or ill, waste the most valuable thing it gave them, which is a detailed list of what they still don't know how to do.
Tactic: After each company or major project, write down the one capability you failed to build, and make it the explicit design goal of the next one.
Principle 2
Bet on the architecture your incumbents call too risky
Three times Duffield bet on a platform shift that the people around him considered premature. In 1986 the Integral board rejected client-server as too risky, so he left and built PeopleSoft on it. Within PeopleSoft, version 8 tore down the client-server suite and rebuilt it for the web. In 2005 he started Workday on the cloud at a moment when his former company's product, now owned by Oracle, was the on-premises standard he would be competing with.
The pattern has a precise shape. He kept the customer and the problem, HR and finance for large institutions, and changed the delivery architecture beneath them. This is a narrower and less romantic kind of disruption than inventing a new category. It requires knowing a domain so thoroughly that you can see what the new architecture will let customers do that the old one never could, and then accepting years of losses, as with Workday's $329.5 million accumulated deficit, while the market catches up.
The old rivals never caught up with PeopleSoft. The new rivals spent a decade catching up with Workday. In both cases the advantage came from rebuilding something everyone already needed, not from imagining something new.
Tactic: Identify the domain you know better than anyone, then ask which platform shift your incumbent's board would veto, and build there.
Principle 3
Fund the bet yourself to choose your board
The founding of PeopleSoft followed directly from a governance failure. Duffield left Integral because his board would not approve the bet he believed in. His answer was to mortgage his house, put about $2 million into the new company, and reportedly fund it himself for two years. PeopleSoft's only venture investor was IBM, a strategic partner rather than a firm with a fund clock.
He repeated the pattern decades later at a far larger scale. He put more than $400 million of his own money into Ridgeline and led its 2026 round himself. Self-funding is not mainly frugality or bravado. It is a way of choosing who sits at the table when the hard decisions come. Founders who raise money from people who doubt the thesis eventually have that doubt in the boardroom.
Most founders cannot write a $400 million check. Almost every founder can make a smaller version of the same choice: to accept less capital, or capital on worse terms, from people who already agree about the risk being taken.
Tactic: Before raising, list the single decision your company will most likely face in three years, and take money only from investors who would vote your way on it today.
Principle 4
Write the culture down and rank it
Workday's six core values were employees, customers, innovation, integrity, fun, and profitability, in that order, and the order was the point. Ranking values turns a culture from an atmosphere into a decision procedure. When the choice is between pleasing a customer and protecting an employee, or between a profitable quarter and an innovative product, the list settles it before the argument begins.
Duffield paired the written values with costly signals. At Workday, he and Bhusri interviewed the first 500 hires personally. At PeopleSoft, he worked from a cubicle, answered his own phone, and opened his own mail. Values that cost the founder nothing tend to mean nothing to anyone else. His version of the idea: "You can say that you care and then show that actually you don't, or you can say that you care and do something about it. The whole company is going to operate according to what you care about."
The durability shows in the record. Nearly two decades after Workday's founding, with the founders largely gone from daily operations and revenue above $8 billion, its CEO named the same six values as the thing that had not changed.
Tactic: Write your company's values as a ranked list, and attach to each one a personal behavior you will keep doing at your own expense.
Principle 5
Put employees ahead of customers, out loud
"The most important thing for us is our employees," Duffield said. "Customers are a close second. Employees are number one." Most executives believe something like this privately and say the opposite publicly, because putting customers first sounds better to customers. Duffield said the honest ranking in public and built his companies around it.
The reasoning is causal, not sentimental. Enterprise software is sold once and serviced for years. The people who answer the support calls, write the patches, and walk a payroll department through an implementation are the product's real interface. A 3 percent turnover rate, as PeopleSoft had in 1997, means the person who learned your account last year is still there this year. Customer loyalty follows from employee continuity.
The principle also shaped how he defined success at the end. His 2004 letter apologized to employees, not to customers or shareholders. His 2005 fund paid former employees. In 2010, the reason he gave Fortune for wanting an IPO was that employees could "make some money here, and do good work with it, and help their families out."
Tactic: State your stakeholder ranking explicitly to your team, and then measure the employee metric that most directly predicts customer retention in your business.
Principle 6
Point the warmth inward and the war outward
The 1997 Forbes profile that celebrated the macarena and the Kool-Aid also described Duffield gleefully reading insulting press clippings about Larry Ellison to a cheering staff. Its title was the doctrine: Have fun, kill the enemy. "Being nice and competitive are two different things," he said, and he practiced both at full intensity, aimed in opposite directions.
A named external rival does work inside a company. It gives a casual, family-like culture an edge and a sense of urgency. It makes ordinary market share feel like defending something. It lets a founder be kind internally without the organization going soft. PeopleSoft's warmth and its competitiveness were one system.
The doctrine has a failure mode, and the PeopleSoft record shows it. Duffield hired Craig Conway partly for his "similar disdain for Larry Ellison," and the board eventually fired Conway in part because the war had become personal. An enemy that motivates a company at 2,300 employees can blind it at 11,000.
Tactic: Name a specific competitor for your team to beat, and set a review date on which the board explicitly asks whether the rivalry still serves shareholders or only serves feelings.
Principle 7
Make equity the reward people choose
PeopleSoft's outstanding-service award offered a choice between $500 in cash and 100 stock options. Employees "invariably" chose the options. The design matters more than the amount. A forced equity grant is compensation. A chosen equity grant is a statement of belief, made publicly by the employee and renewed each time someone else makes the same choice.
The mechanism worked because the stock performed, returning 115 percent a year from the 1992 IPO through early 1997. It also created a feedback loop between culture and capital structure. People who chose ownership behaved like owners, and owners were the employees most hurt by Oracle's takeover. That helps explain why Duffield felt the loss as a personal debt to them.
Tactic: Wherever you offer a cash bonus, offer an equity alternative of comparable expected value, and track the share of people who choose equity as a measure of internal conviction.
P
PeopleSoft vs. Workday: What Changed the Second Time
Duffield's second public company was built as a direct correction to the first.
| Dimension | PeopleSoft (1987) | Workday (2005) | Read |
|---|---|---|---|
| Architecture | Client-server, later rebuilt for the web | Cloud from day one | Scale |
| Control | Acquirable; defended by poison pill | Co-founders held majority voting stock at IPO | Risk |
| Leadership | Solo founder-CEO, then hired successor | Co-CEOs with a younger co-founder | Try |
| Hiring | Rapid growth, culture diluted at scale | Founders interviewed first 500 hires | Scan |
| Values | Fun and integrity as slogans | Six ranked core values, profitability last | Scale |
Principle 8
Design the second company against the first one's killer
PeopleSoft did not die from bad products or bad customers. It died, in Duffield's words, because it "ran out of runway." A determined acquirer, a public shareholder base, and an eighteen-month siege outlasted its defenses. Its most inventive protection, the customer refund guarantee, was a tactical response bolted on after the threat appeared.
Workday was built so the threat could not arise. Its S-1 disclosed that the co-founders controlled a majority of voting stock. Outside observers drew the obvious conclusion that the company was structured to be immune to hostile takeovers. Whatever one thinks of dual-class governance in general, the case for it here was specific and earned. The founders had watched a company with good revenue and loyal customers taken away anyway.
Second-time founders usually overcorrect for the mistakes they made. Duffield corrected for something done to him, which is a more precise target.
Tactic: Identify the external force that ended or nearly ended your last venture, and write the structural defense against it into your next company's founding documents before you raise your first round.
At sixty-four, after the most painful loss of his career, Duffield did not start Workday alone. He started it with Aneel Bhusri, who had joined PeopleSoft in 1993 and risen to vice chairman, and the two shared the CEO title. In 2014 Duffield stepped back and Bhusri led. At Ridgeline he followed a similar pattern, holding the CEO role until 2024 and then handing it to Dave Blair while continuing to meet him for breakfast every week.
The pairing solves several problems at once. The older founder brings domain credibility, customer relationships, and capital. The younger operator brings energy, a longer horizon, and a natural succession plan. A shared title makes the eventual handoff a gradual shift in weight between two people rather than a rupture. Workday's 2026 decision to bring Bhusri back as CEO shows the arrangement producing an heir who also feels like a founder.
Tactic: If you are the experienced founder, recruit a co-founder at least a decade younger and plan the title transition in writing on the day you start.
Principle 10
Own the ending you couldn't prevent
On December 12, 2004, Duffield abstained from the vote to sell PeopleSoft. On December 13 he wrote to 11,200 employees calling it a sad day and apologized to those who would struggle to find new work. Then he backed the apology with action. He called other technology companies to place laid-off workers and created a fund that paid $10,000 grants to former employees who earned under $150,000 and had been out of work for three months.
None of this was required. Shareholders were paid and the deal was done. What he did established, beyond doubt, that the culture he had promised was not a recruiting device to be dropped once the company changed hands. It also mattered commercially. When he started Workday months later, the people he needed to recruit had watched how he behaved when he had nothing to gain.
The ending of a company is the last thing its employees see of a founder, and it is the part they remember when deciding whether to follow you again.
Tactic: Before any sale or shutdown, set aside a defined, personally funded transition program for the employees least able to absorb the loss, and announce it in your own words.
Principle 11
Give like an operator, with a metric attached
Maddie's Fund began with a deeply sentimental gesture, a promise made to a dog, and was run with operating discipline. Its stated goal was a no-kill nation. Its director, Richard Avenzino, tied grants to measurable outcomes: higher adoptions, lower community-wide euthanasia, more spaying and neutering. Its first major program in Utah enlisted all 126 of the state's rescue organizations and veterinary hospitals around a numerical target. In 2024 the fund revised its strategy toward multi-site projects, a reorganization of the kind a company makes when the old plan has stopped producing results.
Duffield even defended his choice of cause with a market-share argument: since 99.3 percent of charitable dollars went to human causes, a fraction for animals was "kind of a fair contribution." His Cornell gifts follow the same logic, committing large sums to specific buildings for specific research fields, such as quantum computing, AI, robotics, and semiconductors, rather than general endowments.
Tactic: For each major gift, define one outcome you will measure annually and a date by which you will revise the strategy if the number doesn't move.
Principle 12
Treat retirement as a pause between companies
Duffield stepped back from PeopleSoft's day-to-day operations in 1999, visibly exhausted. Five years later he was back as CEO and, by his own account, "having a good time, frankly." Months after that he started Workday. At nearly seventy he was in a cubicle. At seventy-six he founded Ridgeline, and in his mid-eighties he was leading its funding rounds. Bloomberg's headline, "Bored Silly," was accurate.
For founders whose work is their main source of meaning, a permanent retirement may not be a real option, and pretending otherwise can produce the worst of both conditions: restless absence from the work and half-hearted involvement in it. Duffield's pattern was to step back from operations when the company needed a different leader, keep the board seat and the relationships, and stay ready for the next point in time when the architecture shifted and an industry needed rebuilding.
Tactic: When you step away from a CEO role, keep one structured weekly commitment to the business, such as a standing meeting with your successor, so you can tell the difference between rest and withdrawal.
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
— David DuffieldThe most important thing for us is our employees, hire the best, treat them well, expect a lot from them. They're the champions of the business.
— David DuffieldI aspired to have a harmonious work environment and failed the first two or three times, and would have failed the fourth time had I not woken up and took this core values system and beliefs seriously.
— David DuffieldI'll sweep the floors. I've picked up napkins in the men's room. I'm not above anything.
— David Duffield, Fortune, 2010It's a rite of passage. It's a special thing to be a public company.
— David DuffieldYou don't know this when you're young, but over time, you see that great companies are usually built at a special point in time.
Maxims
- Count your rehearsals. A profitable company that fails at the thing you care about most is a draft, not a verdict.
- Rank your values. A list in order settles arguments before they begin, while an unranked list just restates them.
- Employees first, customers close behind. In businesses built on long relationships, the people who serve the customer are the product.
- Nice and competitive are different settings. Aim the warmth at your own people and the aggression at the market, and don't confuse the two.
- Choose your board by funding your bet. Capital from people who doubt your thesis will eventually show up in the boardroom.
- Rebuild the familiar on new ground. Keep the customer and the problem and change the architecture underneath them.
- Design against what killed you. Your second company should be structurally immune to the force that ended your first.
- Share the title before you need a successor. A co-founder ten years younger is a succession plan that doesn't feel like one.
- Pay for the ending. How you treat employees when you have nothing left to gain decides whether they follow you to the next company.
- Keep the promise you made before the money. Commitments made in hard years are the ones worth funding once you're rich.
