Part IThe Story
At some point during a large Workday event in 2024, Aneel Bhusri stopped his keynote to say hello to his mother. The moment survives in only a few lines of trade-press reporting, affectionate and short on stage directions. We do not know what he said or how long the pause lasted. We know why she was there. Mother and son had both wanted to mark the occasion, which was supposed to be Bhusri's last keynote as chief executive of the company he had co-founded nineteen years earlier and led, in one configuration or another, for fifteen. He was handing the title to Carl Eschenbach, a former Sequoia partner brought in to run the machine, and returning to what he had always said he loved: product, strategy, the long view. "I don't have any training for running a multibillion-dollar corporation," he had said more than once. People who know him report that he meant it.
Two years later, on February 6, 2026, Workday's board appointed him chief executive again.
The announcement came the following Monday, February 9. The stock fell more than 8 percent that day, to its lowest level since November 2022. Software as a category was coming apart under a single word, AI, and Workday had lost roughly $40 billion in market value from an all-time high near $80 billion. Eschenbach left the board, received a $3.6 million severance payment, and agreed to stay on as an adviser. To bring the founder back, the company assembled a compensation package worth $138.8 million. Some $75 million of it paid out only if Bhusri hit undisclosed stock-price targets over five years. Roughly $60 million more, Fortune noted, required nothing except that he stay for four.
HR software, the category Bhusri spent three decades building, has a word for an employee who leaves and comes back: a rehire. The software logs the original start date, the separation, the return, and the terms. It is a deterministic process, with a beginning and an end. Every step can be audited.
There is something fitting about the founder of a company that keeps the world's employment records becoming, at sixty, a line in one. He has been leaving and returning to the same rooms his whole career. One room was PeopleSoft. Another was a partnership with a man twenty-five years his senior who first took him out for beers. Another was the chief executive's chair, which he never quite wanted and never quite gave up.
By the Numbers
The Workday Ledger
33Bhusri's age when PeopleSoft named him vice chairman in 1999
500Early Workday employees he personally interviewed
74%First-day gain in Workday's October 12, 2012 IPO
68%Voting control held by Bhusri at his 2026 return, per Fortune
$9.55BWorkday revenue for fiscal 2026
11,500+Customers cited by Bhusri on returning as CEO
$138.8MPay package to bring the founder back
A Small Little HR Software Company
The public record on Aneel Bhusri's childhood is thin, and the gaps matter. He was born on February 14, 1966, in Pittsford, New York, a suburb in the Rochester area, into an Indian-American family of Punjabi descent. Beyond that, the official biographies go quiet. His parents are not named in the standard accounts, and I have found no interview in which he describes the household he grew up in. What can be described is the history that preceded him.
Punjabis were the first South Asians to arrive in the United States in meaningful numbers. Most were Sikh farmers. Around 2,000 Indians worked for the Southern Pacific and Western Pacific railroads between 1907 and 1908. On September 5, 1907, a mob in Bellingham, Washington, drove out the "Hindoos," as locals called them, regardless of their actual religion. In 1923 the Supreme Court ruled in United States v. Bhagat Singh Thind that Indians were not "free white persons" and could not become citizens. The Luce–Celler Act of 1946 reopened the door to a quota of one hundred Indians a year. The Immigration and Nationality Act of 1965 opened it wide. Bhusri was born the following year. The 2020 census counted 7,564 Asian Indians in the entire Rochester combined statistical area. By any measure, he was one of very few.
He went to Brown, where he took degrees in electrical engineering and economics. The pairing signals someone who wants to understand how the machine works and also what it is worth. After college he became a corporate finance analyst at Morgan Stanley in New York and stayed several years. Then came Stanford's Graduate School of Business, where a summer job turned out to matter more than anything taught in a classroom.
Between his first and second years, Bhusri worked at Norwest Venture Partners for a man named George Still. Still happened to sit on the board of PeopleSoft. He became a mentor, and as Bhusri was finishing business school in 1993, Still told him to go see Dave Duffield. Bhusri has described what he joined in unglamorous terms: "a small little HR software called PeopleSoft."
Dave Duffield is the second protagonist of this story, and in some ways its origin.
He was born in Cleveland in 1940 and raised in Ho-Ho-Kus, New Jersey. His mother taught at the elementary school he attended, and his father was an engineer without a college degree. At Ridgewood High School he threw the first no-hitter in school history and won the top math award for the Class of 1958. The prize was a new slide rule. He pitched at Cornell, earned an electrical engineering degree and an MBA there, joined IBM in 1964 as a marketing representative, and then started companies the way other men change jobs: Information Associates, Integral Systems in 1972, Business Software Corporation in 1979, and PeopleSoft in 1987, when he was forty-six.
Duffield and Bhusri trained in the same discipline and arrived at business by the same detour. What the young man noticed first, though, was not the résumé. Duffield took the new hire out for beers. Bhusri was struck that a chief executive would spend an evening getting to know someone at the bottom of the company, and he chose PeopleSoft largely because of it.
He started as director of planning, close to entry level. He did not stay there. He took over products, then marketing, and became, in the phrasing of one profile, "the No. 2 person at PeopleSoft." By then the company had grown from an HR vendor into a full suite of enterprise resource planning (ERP), customer relationship management, and human resources software. In 1999, when Bhusri was thirty-three, PeopleSoft named him vice chairman and put him on its board. CNET ran the news as a "Short Take," three sentences long. He also sat on the boards of Marimba, WiData, and STC.
That same April, Bhusri became a partner at Greylock, the venture firm that would later back Facebook, Dropbox, and Airbnb. For years he held both roles at once, insider and allocator. He was an operator who wrote checks and an investor who knew what it cost to ship a payroll module. That doubleness became one of the defining features of his career, and it began before he was thirty-five.
The Pink Slip's in the Mail
In 2003, Larry Ellison came for PeopleSoft.
The details of the siege now read like settled history. Oracle launched a hostile takeover bid, which Wikipedia's account puts at $13 billion, and PeopleSoft's board rejected that offer and the ones that followed. The U.S. Department of Justice sued to block the deal on antitrust grounds. None of it held. In December 2004 Oracle announced a definitive agreement to acquire PeopleSoft for approximately $10.3 billion, and the deal closed in January 2005. The company Duffield had founded, then the world's second-largest application software company, became a product line inside the firm that had spent two years trying to swallow it.
Bhusri had left the vice chairmanship in September 2001 to concentrate on Greylock. In September 2004, as the fight entered its final months, he came back to the role and held it until December. This was his first return: he showed up at the end, when there was nothing left to win. On January 14, 2005, CNET ran a headline about Oracle's plans for PeopleSoft's workforce: "Oracle to PeopleSoft: The pink slip's in the mail."
The defeat is the hinge of the story, and both men reacted in ways that reveal them. Duffield was sixty-four. He had five companies behind him and a fortune by any standard, along with a foundation he and his wife Cheryl had set up in 1994 with $200 million. It was named for their miniature schnauzer, Maddie, whom they called a "lighthouse during the stormy period" of their careers. When Maddie died of cancer in 1997, Duffield kept a promise he had made to her years before: "If we ever make some money, I promise we will give it back to you and your kind so others can be as happy as we are today." A man who writes that sentence and then funds it with nine figures is not someone who retires quietly after being bought out by Larry Ellison.
Bhusri was thirty-nine and had a career at Greylock that did not depend on any single company. He could have watched from the investor's side of the table and backed whatever came next. Instead, in March 2005, he and Duffield founded Workday. The money came from Duffield and from Greylock, the firm where Bhusri was a partner.
George Anders, writing in Forbes in 2013, summarized the arc in one line: "Larry Ellison booted Dave Duffield and Aneel Bhusri from their company. They returned the favour by creating the biggest threat to Oracle in years." It makes a good revenge story, and it is mostly true. Bhusri himself told it with more caution. "One of most important things for us," he said in 2008, "was that we had to make sure we were doing this for the right reasons — not just for nostalgic reasons. We had a great run at PeopleSoft, but there really had to be an opportunity to build the next great applications company."
The word nostalgic stands out. He was naming the temptation he wanted to avoid, which is a reasonable guide to what he expected to be tempted by.
Legacy ERP Guys, Switching Sides
The phrase "legacy ERP guys" was his own. He used it in 2008 to describe himself and Duffield to CIO magazine, which framed their project with some skepticism. The men who had sold on-premise systems to chief information officers for twenty years now wanted those same buyers to rent software delivered over the internet. Their target was not some abstract incumbent. "Square in their sights," the magazine wrote, were "companies like Oracle and its PeopleSoft applications, which Duffield and Bhusri had evangelized for so many years."
They were trying to make obsolete the product they had built, now owned by the man who had taken it from them. The revenge narrative leaves out the strangeness of that. Every PeopleSoft installation Bhusri had helped sell was now a possible Workday customer. Each one was also proof that the old way worked well enough for people to keep paying for it.
"After the hostile takeover," Bhusri told CIO, "we felt that the next wave of ERP was going to be on-demand." He listed the components as an engineer would: on-demand delivery, web services, object-oriented technologies, "what had happened with the consumer Internet." They were not adding a feature to the old platform. "We started the company to be unequivocally the replacement platform for HR and ERP systems that had been delivered in an on-premise way."
Ray Wang, a Forrester analyst who had also worked at PeopleSoft, laid out the economics. Implementation took days and weeks instead of months and years. Pricing was a monthly fee per user instead of $100,000 to millions up front. Upgrades arrived every three to six months instead of every two to three years. Years later, Andreessen Horowitz would describe the founding insight more sharply. Duffield and Bhusri had bet that the move from client-server to multi-tenant cloud would reset the category. They saw that Oracle and SAP would struggle to close the architectural gap. They also saw that large enterprises were shifting from big upfront capital outlays to predictable annual subscriptions. Workday was built for both changes from the first day.
The early timeline is short and has the feel of a company that knew exactly what it was building. After fifteen months of development, Workday shipped its first product, core HR, in November 2006. Financials followed in July 2007, and payroll was scheduled for May 2008. The company ended 2007 with thirty-five customers, each averaging 5,000 to 10,000 employees. By early 2008 it had about 170 people, and roughly 80 percent of them had come from PeopleSoft.
That 80 percent figure matters. Oracle had bought PeopleSoft's code, contracts, customers, and buildings. It had not bought the people, and the people were moving, a few at a time, to a startup in Walnut Creek.
In February 2008 Workday bought Cape Clear Software, a Dublin integration company, which signaled that a cloud product would still need to connect to everything a large enterprise already ran. In May 2008 it signed a large human capital contract with Flextronics, which InformationWeek called the software industry's biggest software-as-a-service deal. That December the company moved its headquarters from Walnut Creek to Pleasanton, California. The local paper ran the story under the headline "Duffield Is Back." Pleasanton was where PeopleSoft had been based. The legacy ERP guys had gone home.
On April 29, 2009, with credit markets frozen and much of venture capital waiting to see what would survive, Workday announced $75 million in funding led by New Enterprise Associates, with Greylock and Duffield also participating. Forbes called it "Duffield Scores Biggest Round." That year the two founders formalized what had already been true in practice: they became co-CEOs. The hyphenated title would recur throughout Bhusri's career.
Both Sides of the Table
For most of Workday's first decade, Bhusri was the rare enterprise executive who appeared on the Midas List, Forbes's ranking of venture investors. He was No. 16 in January 2008, at forty-one, and No. 8 a year later. The notes on his returns reflected his PeopleSoft background. PolyServe, a storage software company, sold to Hewlett-Packard for $200 million. Data Domain, a backup-device maker whose board he chaired, was then worth $1.1 billion. He spent five years on the board of OutlookSoft, a financial planning company that SAP later acquired, and he would later cite that experience when explaining why Workday needed to own its planning product. He sat on the boards of Cloudera, Pure Storage, and Sumo Logic. In April 2013, AllThingsD described him in its headline as "Workday CEO and Greylock Partner."
That interview, conducted over a meal in a San Francisco restaurant, shows his mind clearly in a single paragraph:
"When you think about what's happening in the enterprise, it's the most disruptive time in 25 years. Apps are moving to the cloud. Arguably, the relational database is going to look like a mainframe in 10 years, as transactions move into in-memory databases and Hadoop and other noSQL databases for Analytics. Storage is going from disk to flash. The legacy enterprise companies aren't innovating, but they have cash and they have distribution, so they can buy their way into this new generation of innovation."
He then described the dilemma facing founders: whether "this generation of entrepreneurs sells out to the big guys, or do they go it alone." Between 2000 and 2006, he said, with the exception of Salesforce, startups had competed "at the edges." Now they were "going right for the jugular."
Workday itself had already answered that question. When SAP bought Workday's rival SuccessFactors in late 2011 and Oracle answered by buying Taleo, the incumbents were trying to buy back the future, and Workday was the company not for sale. The governance reflected that. When the company went public, Duffield held voting rights on shares worth $3.4 billion and Bhusri on shares worth $1.3 billion. Together they controlled 67 percent of the vote. The Wall Street Journal noted that this "makes the event of a hostile takeover much less likely." It is hard to read that line without thinking of Ellison. The founders had lost one company to a hostile bid and arranged the second so it could not happen again.
The last private round set up the public listing. On October 24, 2011, Workday announced $85 million in new funding at a $2 billion valuation, which brought total capital raised to $250 million. The investors included Morgan Stanley Investment Management, the asset arm of the firm where Bhusri had started as an analyst, along with Janus, Bezos Expeditions, and T. Rowe Price.
T. Rowe's check was written by Henry Ellenbogen.
Ellenbogen was the grandson of a Viennese immigrant who became a Pittsburgh congressman and quietly helped thousands of European Jews escape in the 1930s. He had run a congressional campaign at nineteen and spent a summer at Goldman Sachs before deciding he did not want to be an "insurance adjuster" pricing other people's risk. At T. Rowe he had built what colleagues called a "rogue effort" to put mutual fund money into private companies, a practice the Investment Company Act of 1940 barely allowed.
Ellenbogen's first big private bet was Twitter, a company with thirteen employees and no clear chief executive. Workday came in 2011. It had about $100 million in revenue, was growing 80 percent a year, and was valued at $2 billion, which, Colossus reported, some respected investors thought was crazy. Ellenbogen and his analyst Corey Shull saw something else: two founders who had already built a business in the same category and understood how the industry worked, with a clear view of how cloud infrastructure would improve it. When Ellenbogen later looked for proof that older founders could start over, he pointed to Duffield, who was sixty-four when he founded Workday. Ellenbogen called such people "Act 2 entrepreneurs."
— Henry Ellenbogen, on T. Rowe Price's early private bets, as reported by ColossusWe got the technology right with Twitter. With Workday, we got the business model and the people.
If both bets had failed, Colossus reported, T. Rowe would have shut down the program. They did not fail. Workday priced its IPO on the New York Stock Exchange at $28 a share on October 11, 2012, and closed its first day of trading, Friday, October 12, at $48.69. The 74 percent gain gave the company a market value of nearly $9.5 billion including unexercised options. It sold 22.75 million Class A shares and raised $637 million, more than any American technology listing since Facebook that May.
The company that went public that day had lost $30.94 million on $81.52 million in revenue in its most recent quarter. The market did not mind. Within two years Bhusri was a billionaire, and Forbes called his and Duffield's fortunes the fastest-growing on its entire list.
Five Hundred Interviews
Reid Hoffman, a Greylock partner, eventually interviewed Bhusri for Masters of Scale, and the detail listeners remembered was a number. Bhusri had interviewed each of Workday's first five hundred employees himself.
This deserves to be taken literally. By the time Workday was hiring its five-hundredth person, it was a funded company with Fortune 500 customers, a payroll product in development, and a co-CEO who was also a venture partner sitting on outside boards. Five hundred interviews means five hundred hours at minimum, probably far more. A founder who spends that time is betting that the company's culture is set in its first years and cannot be fixed afterward.
The values were written down early and repeated for twenty years: employees, customer service, innovation, integrity, fun, and profitability. The order is deliberate. "We've always been employee-first," Bhusri said in 2020. "I know a lot of companies say customers number one, but I've never seen a company that has happy customers and unhappy employees. You've got to start with your employees."
— Aneel Bhusri, in conversation with Marc Benioff, June 2020I've never seen a company that has happy customers and unhappy employees. You've got to start with your employees.
The practical expressions were modest. When Workday hit its targets, employees got a week off and a stipend. Dogs often slept next to their owners' desks, a detail that is easier to understand once you remember who was paying for Maddie's Fund. Duffield's title at the new company was "chief customer advocate." The pair created the Workday Foundation in 2012 with a gift of 500,000 shares.
Bhusri could also be cutting. Interviewed in 2018 by Fortune's Adam Lashinsky, in a ballroom deep inside the Marriott Marquis in San Francisco at a Great Place to Work conference, he said that if employees didn't care about fun, they could go work for Oracle. Workday was No. 7 on the Fortune list that year. In 2020 it was No. 5.
The more revealing part of that conversation was a problem he admitted. A few years earlier, he told Lashinsky, he had watched Workday's ratings slip on Glassdoor. As the company grew, not all of its new employees, especially managers, were living its values. Five hundred interviews could set a culture, but they could not hold it at ten thousand people. The response, as Lashinsky summarized it, was "to dramatically add processes and training the growing company needed." The man who had personally vetted his early hires now had to build a system that would vet people for him.
He added that he was willing to let go of a top performer who did not meet the company's standard for collegiality. When Lashinsky asked whether he could prove this helped shareholders, Bhusri gave an answer Lashinsky called sagacious: financial results are probably a lagging indicator of a bad culture. By the time the damage shows in the numbers, it is probably too late to fix.
A man who runs on leading indicators sells software that, at its core, records lagging ones: who was hired, what they were paid, what was spent. Bhusri spent the next decade trying to turn that record into prediction.
The Pied Piper
He calls himself "a products guy," and he says it modestly, as if it were a limitation.
The clearest picture of how he operates comes from a story he told the Cloud Wars columnist Bob Evans, who has known him for about fifteen years. When machine learning began to matter in enterprise software, Bhusri decided Workday needed to understand it at every level. He did not issue a mandate or set up a task force. He asked one question, everywhere. "Every time I see one of our engineers or developers, I ask, 'what are you doing on machine learning?' Or 'what do you think about machine learning?' And 'what should we be doing with machine learning?'"
"Pretty soon they're all saying, 'Okay, before I meet with Aneel, I know he's going to ask about machine learning so I should have my act together.'"
Internally, he said with a laugh, he described his role as "the pied piper of machine learning." He asked every employee to buy Prediction Machines, the economists' book on how cheaper prediction reshapes business decisions, and to expense it to Workday. He saw it as relief from a burden as much as a strategy: "For me there's actually something very gratifying when I can say, okay, not going to try to get the engineers to work on five different things." He said machine learning would prove more disruptive than the cloud, which is a striking claim from someone who had built a company on the cloud.
The acquisitions tell the same story in sequence. In February 2014 Workday bought Identified and its artificial intelligence engine, Syman. In 2015 came the first AI capabilities, Workday Talent Insights. SkipFlag, an AI knowledge-base startup, followed in January 2018, and Skills Cloud launched that same year. Skills Cloud broke jobs down into skills and normalized them in a canonical format across customers so that hiring could become a matching problem. Bhusri later said Workday had access to data from more than sixteen million records on the HR side. The bet was that the system of record, the dull ledger of who works where, would become the training data for the next generation of software.
He also bought things that were not AI. In June 2018 Workday agreed to acquire Adaptive Insights, a cloud financial planning company, for $1.55 billion. The diginomica writer Dennis Howlett, who had argued for twenty years that the spreadsheet had to die, suggested it was a defensive move. Bhusri rejected that. "It's absolutely an offensive play and not defensive in any way," he said. He cited his five years on the OutlookSoft board: the implementations were cumbersome, the usability poor, and moving data in and out was one of the hardest parts. The planning products of the client-server era had promised distributed, business-user planning and delivered spreadsheets pasted back into a central system at the end. "So, basically in many ways, I'd argue cloud delivered on the promises of client-server."
The line works in two directions. It is a claim about technology, and it is also what a man says when he is completing a project he started thirty years earlier.
By then the scale had changed. In February 2016 Workday reported its first fiscal year with more than $1 billion in revenue, and Business Insider reported that it was displacing Oracle software as it did so. The following year Bhusri opened the platform to developers and partners and moved Workday into platform-as-a-service. He was building infrastructure and, like the incumbents he had once criticized, buying his way toward the next era. The acquisition list grew: Platfora, Rallyteam, Stories.bi in Prague, Scout RFP for $540 million, Peakon in Copenhagen for $700 million, VNDLY for $510 million. Bhusri had predicted in 2013 that incumbents with cash and distribution would do exactly this.
The Hyphen in Co-CEO
Few company founders have changed their own title as often as Bhusri. His Workday biography reads like a ledger of configurations. He was co-CEO with Duffield from 2009 to 2014, chairman from January 2012 to May 2014, sole CEO from 2014 to 2020, co-CEO again from 2020 to 2024, executive chair from 2024 to 2026, and CEO and chair since then.
Duffield gave up the co-CEO role in May 2014 and became chairman. That same month, Bhusri shared a stage with Marc Benioff at the Code Conference. Benioff spent most of his time urging tech executives to support their cities. Bhusri described a coming "post-server world" in which nearly every business application would run in the cloud. Recode noted that he had just become sole CEO earlier that month. He was forty-eight and running alone for the first time.
It lasted six years. In August 2020 he named Luciano "Chano" Fernández co-CEO. When Fernández announced his departure in December 2022, Bhusri chose Carl Eschenbach, who came from Sequoia, the venture world Bhusri also belonged to. The plan was announced in advance: Eschenbach would become sole CEO, and Bhusri would move to executive chair. Bhusri was openly pleased. Eschenbach's arrival, he told Evans, would let him get back to building products and setting strategy.
Bhusri seems to have treated the chief executive's job as an organizational design problem rather than something to hold onto. In a 2022 conversation he described "how a co-CEO model can solve more problems." The structure kept the work he loved, product, close and assigned the work he felt untrained for to someone else. Some founders cannot share a title. Bhusri seemed unable to hold one alone for long.
Duffield, meanwhile, kept doing what he had always done. In 2017, at seventy-six, he founded his sixth company, Ridgeline, an investment management software platform based in Incline Village, Nevada, near Lake Tahoe. In 2024 Bloomberg described him as "bored silly" by retirement and pursuing a third IPO. He stepped back from Workday's chairmanship and became chairman emeritus, then CEO emeritus. In 2026, he and his wife pledged $371.5 million to Cornell, the largest gift in the university's history, and the College of Engineering was renamed for him. The slide-rule boy became the name on the engineering school. The man who took a new hire out for beers never stopped starting companies.
The Workday press materials call the two co-founders best friends, and Fortune called Duffield Bhusri's "best friend and mentor." They describe a twenty-five-year age gap, a shared engineering education, two founded companies, a lost war, and a won one. They differed sharply on one point. Duffield loved being CEO and kept coming back to it in new companies. Bhusri found the job heavy and kept coming back to it at the same company.
Over these years, Bhusri served on the boards of Intel from 2014 to 2019, then General Motors, Cruise, Okta, and Pure Storage. Okta's co-founder Frederic Kerrest later included him in Zero to IPO, a book of hard-won founder advice alongside Marc Andreessen, Eric Yuan, and Melanie Perkins. Bhusri became a Stanford trustee, a director of Memorial Sloan Kettering, and a Crown Fellow at the Aspen Institute. Asked in a 2022 interview what advice he gives most, he offered two words: collect mentors. He first did that on a summer job at Norwest.
A Company Needs a Soul
The pandemic gave Bhusri a vocabulary, and for a while he used it constantly.
In April 2020 he gave $1 million to Give2SF, San Francisco's coronavirus relief fund. In June he appeared with Benioff to discuss what Benioff called a pandemic, an economic crisis, a racial crisis, and a leadership crisis all at once. Bhusri said that a crisis "forces you to your core values … I think you never know if your value system really works until you hit a crisis." He described three town halls in a few weeks with Workday's Black employees, who told him they felt alone and did not get the mentorship their colleagues received. Taking a leadership role, he said, felt risky to them. "We have to fix that."
Late that year he told Evans: "Working from home is not for everyone. It's definitely not for me." Workday had vaccinated employees against the flu in an office closed for eight months. "Ironically, people came into the office to get vaccinated," he said. "But that's the right thing to do."
The word he kept using was soul. "I believe it's more important than ever, for companies to have a soul," he said at Workday's 2021 virtual event, which opened with Ayesha and Stephen Curry discussing their Eat. Learn. Play. foundation, whose board Bhusri would join, and featured Amal Clooney and Mary Barra. On Salesforce's The Inflection Point he cited Churchill: "There's no way to start a company without optimism," followed by the quote about choosing to be an optimist because the alternative is not very good.
His idealism had limits, and he was clear about them. At Workday Rising in Florida in September 2022, the Fortune editor Alan Murray asked him about political attacks on ESG. Bhusri said he believed in environmental, social, and governance concerns as "part of stakeholder theory," but added: "I'm not a fan of the ESG construct, because it seems like a catchall… The SEC is now looking to measure and regulate it. I think that's not the best idea." At Davos in 2019 he had said every company using data to make decisions needed an ethics officer. He believed in conscience as a function inside a company, not as a category for regulators to define.
That distinction was tested by the product itself. In 2023 a job applicant named Derek Mobley filed a federal class action alleging that Workday's AI-driven screening tools discriminated against applicants based on age, race, and disability. Workday denied it, saying its systems were not designed to consider protected characteristics, while critics argued that bias came in through proxy variables. In May 2025 a federal judge allowed the age claims to proceed as a collective action, and later rulings required Workday to disclose which customers used its AI hiring features. The software built on employee-first values now stood between millions of applicants and the people who might have hired them. When Workday announced Bhusri's return as CEO on LinkedIn, one commenter, a company that helps candidates get past screening systems, wrote that it existed "because Workday's (and every ATS's) system works too well at rejecting people before humans ever see their resume."
In February 2025, with Eschenbach as CEO and Bhusri as executive chair, Workday cut 1,750 jobs, 8.5 percent of its workforce, to fund an AI push amid a "softer macroeconomic environment." A company that had once given employees a week off for hitting targets now announced layoffs for the same reason everyone else did. In 2024 it entered the Fortune 500 and the S&P 500. By its own account, it had become what large companies become.
Then, on April 28, 2026, Andreessen Horowitz published an essay by Joe Schmidt titled "Workday's Last Workday?" Its first sentence was harsh: "Workday is arguably the most important and least loved product in enterprise software."
The essay is worth reading as a portrait, not just a short thesis, because it describes the moat Duffield and Bhusri built in the terms of the people who live inside it. It describes HR administrators running reports across three screens and managing compensation cycles in Excel because the system is too complicated. It describes integrations built in Workday Studio, a proprietary tool, taking six to eighteen months with certified consultants, of whom there were more than 10,500. Workday's tools have no market outside Workday, Schmidt wrote, so "both the developers and the customers who employ them are locked in by their own résumés." He argued that near-100 percent renewal rates should not be read as satisfaction: "leaving is close to impossible."
He was describing PeopleSoft, or the version of it the legacy ERP guys had promised in 2008 to replace with something users would choose. Another startup with venture money was now aiming at the incumbent.
Deterministic
The return came at the worst moment and was presented as the best one.
"We're now entering one of the most pivotal moments in our history," Bhusri said in the February 9 press release. "AI is a bigger transformation than SaaS — and it will define the next generation of market leaders." The lead independent director, Mark Hawkins, went further: "As we enter a defining moment shaped by AI, there is no one better than Aneel to lead this next chapter." Workday confirmed that the appointment was permanent, not an interim arrangement during a search. Fortune noted that Bhusri would combine majority voting control with operational authority as CEO, power few founders hold. Fortune also did the math on his stake. He held more than eight million shares, worth $2.6 billion at the 2024 peak and about $1.3 billion now. The skeptics noticed too. "This is the language of a man who saw the tide turning," one blogger wrote, and decided to steer through it himself.
The week before, Workday had announced about 400 more job cuts, roughly 2 percent of the workforce. Forrester's analysts were careful but blunt. Under Eschenbach, they wrote, Workday had pursued an aggressive, acquisition-led AI strategy, buying HiredScore, Evisort, Paradox, Flowise, Sana, and Pipedream, and was "struggling to fully integrate these assets." The firm warned of "integration drag." The Sana deal alone had cost about $1.1 billion. In 2013 Bhusri had said incumbents with cash and distribution would buy their way into each new generation. Now he had to show that buying could also become building.
On February 24 he held his first earnings call back in the role. The numbers were solid: fourth-quarter revenue of $2.53 billion, up about 15 percent, and full-year revenue of $9.552 billion with $693 million in net income. Analysts paid more attention to a paragraph he seems to have been preparing for decades.
"You've all heard the narrative out there that HR and ERP will be replaced or relegated to the background by AI. I personally just don't see that happening. Our application domains are really, really hard to build. I've been working in the HR and ERP space for over 30 years. These are true systems of record that must process transactions with absolute accuracy and speed, enforce complex security models, and comply with statutory and regulatory requirements all over the world."
Then came the line that went everywhere:
— Aneel Bhusri, Workday fiscal Q4 2026 earnings callNo amount of vibe coding is going to produce an HR or ERP system.
He continued, and the next part is the more interesting argument. "Our underlying business processes are deterministic by nature. There is a start and end to a business process. Its goal is to deliver consistent, auditable outcomes. AI, for all of its incredible capabilities, is probabilistic by nature… You can't have probabilistic outcomes in running a payroll, it needs to be 100% accurate and completed, 100% of the time."
This is the philosophy of a man who started as director of planning. Some things in a company can be guesses, and some must be ledgers. He predicted that the future would be a "marriage" of the two: deterministic applications with probabilistic intelligence on top, prompting instead of clicking, and agents doing work that people now do across three screens.
He called it Chapter Four. The first chapter was the founding in 2005, "based on a new idea of HR and Finance in the cloud." The second was hypergrowth. The third, starting in 2023, was "operational excellence and efficiency," which was Eschenbach's era, described politely. The fourth was a return to innovation, which he also called a "re-founding."
Three months in, he said what he had been studying. "I watched a lot of Steve Jobs videos where he talked about his management approach when he returned as CEO." What stayed with him was not the products. It was a 2010 AllThingsD interview in which Jobs called Apple "the biggest start-up in the world." Bhusri drew the conclusion: "fewer layers, faster decisions, the best ideas winning, more ownership." He set up an "AI agent factory." He named Joel Hellermark, the young founder of Sana, as chief AI officer. "With AI, we are essentially a start-up again," he said. "That's how we are leading Workday now and it's how Dave and I led the company early on."
He had also been doing what he did in the beginning, going out and listening one meeting at a time. "I've met with dozens of customers since I've been back," he said, "and I keep hearing the same thing — not one of them is looking to replace Workday with something they're building internally or from a start-up." The man who had once insisted on interviewing every hire himself was now gathering his own evidence in person.
The question underneath all this cannot be answered from the outside. Bhusri may be right that payrolls will always need a deterministic ledger and that the ledger's owners will capture the value of the agents built on top. Or Schmidt may be right that renewal rates reflect captivity more than loyalty, and that a platform shift larger than the one that created Workday will treat it the way Workday treated PeopleSoft. He has been on both sides of this before. In 2005 he was the insurgent arguing that the incumbent's architecture could not cross the gap. In 2026 he is the incumbent arguing that the gap is smaller than it looks.
On March 5, 2026, Workday filed a Form 8-K with the Securities and Exchange Commission. It reads like any personnel record. In connection with his appointment as chief executive, Mr. Bhusri was granted a restricted stock unit award covering 437,602 shares of Class A common stock under the company's 2022 Equity Incentive Plan. He would not be eligible for additional equity awards until fiscal 2028. The award would vest over four years, one-quarter at a time, the same schedule an engineer receives on the first day, as if the co-founder were starting over.
Part IIThe Playbook
What follows is an attempt to draw usable principles from a career that rarely announced its methods. Bhusri does not write manifestos. His approach shows in structural choices: the governance he put in place after losing PeopleSoft, the titles he kept rearranging, the question he repeated until engineers prepared for it. Each principle below is grounded in that record and tested against where it failed or strained.
The Bhusri Playbook
- 1.Choose the mentor, then the company.
- 2.Make your own legacy product the target.
- 3.Build the second company so the first loss cannot repeat.
- 4.Sit on both sides of the table.
- 5.Hire the culture by hand, then replace your hands with systems.
- 6.Treat financial results as a lagging indicator of culture.
- 7.Steer with one repeated question.
- 8.Treat the org chart as a variable.
- 9.Separate what must be deterministic from what can be probabilistic.
- 10.Buy on offense, but count the integration bill.
- 11.Don't confuse retention with love.
- 12.Return as a founder, not a caretaker.
Principle 1
Choose the mentor, then the company
Bhusri's career depends on two choices of people over institutions. The first came at Norwest, where a summer job put him in front of George Still, who sat on PeopleSoft's board and steered him toward Duffield. The second came when he chose PeopleSoft, then a small HR software company, largely because its founder took a new hire out for beers. Neither choice was strategic in a narrow sense. PeopleSoft was not the obvious destination for a Stanford MBA with Morgan Stanley experience. But the bet on Duffield paid off three times: a rise to vice chairman by thirty-three, a co-founder for Workday, and a partner whose reputation, according to Forrester's Ray Wang, "brought significant interest to the company" when it had no track record.
Most career advice focuses on the company: its market, its trajectory, its brand. Bhusri's record suggests that the person you will learn from compounds more reliably than any company's prospects, and that a mentor's network becomes your own over time. His later advice, "collect mentors," is a description of his own path, not a slogan.
Tactic: When comparing two offers, rank them by the founder or manager you will learn from most directly, and treat the company's prospects as the tiebreaker.
Principle 2
Make your own legacy product the target
Workday did not attack a competitor's product. It attacked PeopleSoft, which Duffield and Bhusri had "evangelized for so many years," now owned by Oracle. Bhusri called the new company "unequivocally the replacement platform" for on-premise HR and ERP. The advantage was unusual: they knew every weakness of the incumbent because they had designed many of them. They knew which customers suffered most through upgrade cycles, which implementations ran long, and which buyers had always wanted to stop paying large license fees up front.
The risk was nostalgia, rebuilding the old thing on a new platform, and Bhusri named it at the founding. He said the test was whether there was "really... an opportunity to build the next great applications company," not whether they wanted their old jobs back. Eighty percent of early hires came from PeopleSoft, but the architecture did not. Multi-tenant cloud and subscription pricing were built in from the start, so the institutional knowledge was applied to a product the old company could not have built.
Tactic: Write down the three biggest structural weaknesses of the product you know best, and ask whether a new architecture, not a new feature set, would remove them.
Principle 3
Build the second company so the first loss cannot repeat
Oracle took PeopleSoft through a hostile bid that its board rejected repeatedly and the Justice Department tried to block. Workday's IPO structure looked like a direct response: dual-class shares that gave the two founders 67 percent of the vote, an arrangement the Wall Street Journal said "makes the event of a hostile takeover much less likely." Two decades later, that control let Bhusri return as CEO with authority few founders have. Fortune put his voting control at 68 percent.
The lesson is not that every company needs dual-class stock. It is that founders should design governance around the specific way they were hurt before, while the scar is still fresh. Many founders write charters around abstract best practices. Bhusri and Duffield wrote theirs around Larry Ellison.
Tactic: Before your next financing, list the ways you have seen a company lost, whether to acquirers, boards, or dilution, and make sure the documents address the one you fear most.
⚖
Two Companies, Two Outcomes
How the PeopleSoft loss shaped Workday's design.
| Dimension | PeopleSoft (1987–2005) | Workday (2005–) |
|---|---|---|
| Architecture | Client-server, on-premise | Multi-tenant cloud from day one |
| Pricing | Perpetual license, $100K to millions upfront | Per-user, per-month subscription |
| Upgrade cadence | Every two to three years | Major release every six months |
| Takeover defense | Board rejection; lost to Oracle | Founders held 67% of votes at IPO |
| Ending | Acquired ~$10.3B | Independent ~$9.55B annual revenue |
Principle 4
Sit on both sides of the table
From April 1999, Bhusri was a Greylock partner while also serving as an operating executive. He made the Midas List while helping start Workday. He chaired Data Domain's board, spent five years on OutlookSoft's, and sat on Cloudera, Pure Storage, and Sumo Logic. This was not a distraction. It gave him a view across the whole stack. His 2013 description of the enterprise, apps moving to the cloud, relational databases becoming the new mainframe, storage moving from disk to flash, came from someone who saw pitches in every layer.
The investor's view shaped the operator's choices. His OutlookSoft years told him why legacy planning software failed and made the Adaptive Insights deal look like completing a thesis rather than chasing a market. His observation that incumbents would "buy their way into" each new wave predicted, accurately, the defensive acquisitions SAP and Oracle made around Workday.
Tactic: Take one outside board or advisory seat in a layer of the stack adjacent to yours, and keep a written log of what its problems tell you about your own market.
Principle 5
Hire the culture by hand, then replace your hands with systems
Bhusri personally interviewed Workday's first five hundred employees. That is an expensive way to set a culture, and it worked for a while. Then the company grew past what one person could screen, Glassdoor ratings slipped, and Bhusri admitted that new managers in particular were not living the values. The fix was not more founder interviews. It was what Fortune described as dramatically adding the processes and training the company needed.
The two phases belong together. The founder's personal involvement defines the standard, and the system scales it. Founders who never do the first build processes around a culture nobody has defined. Founders who never move to the second become a bottleneck and watch the culture weaken beyond their reach. Bhusri did both and was candid about the gap between them.
Tactic: Personally interview every hire until you can write down why you rejected the last ten, then turn that list into the training your managers receive.
Principle 6
Treat financial results as a lagging indicator of culture
When pressed to justify culture spending to shareholders, Bhusri reframed the question. Financial results, he said, are probably a lagging indicator of a bad culture, and by the time the effects show up it is probably too late. His "employee-first" ordering, employees ahead of customers, follows from this: "I've never seen a company that has happy customers and unhappy employees." He backed it by being willing to fire top performers who lacked collegiality.
The principle has a limit, and the record shows it. Workday's 2025 cut of 1,750 jobs was justified by macroeconomic conditions and AI investment, the kind of financial reasoning the culture argument was meant to resist. Leading indicators only help if leadership acts on them before the lagging ones force a decision.
Tactic: Pick one culture metric you trust, such as manager-level engagement scores, and give it a standing slot in every quarterly business review, ahead of revenue.
Principle 7
Steer with one repeated question
Bhusri's machine learning push did not start with a reorganization. It started with a question he asked every engineer he met: "What are you doing on machine learning?" Within weeks, engineers were preparing answers before meeting him. He added a small, concrete gesture, buying every employee Prediction Machines, and called himself the "pied piper." Years later, when generative AI arrived, he had a decade-long record as an AI advocate that analysts cited when assessing his return.
The method works because it costs the leader almost nothing and focuses the organization without a directive. It also spares the leader from pushing five priorities at once, which Bhusri said he found "very gratifying." Repeating a question signals what you will notice, and people tend to work on what the boss will notice.
Tactic: Choose the single capability your company most needs in two years and ask about it, in the same words, in every meeting for ninety days.
Principle 8
Treat the org chart as a variable
Bhusri held at least six configurations of the top job: co-CEO, chairman, sole CEO, co-CEO again, executive chair, and CEO and chair. He openly said he lacked "training for running a multibillion-dollar corporation" and used the co-CEO model to pair his product focus with operators like Duffield, Fernández, and Eschenbach. Ego seemed to play little role. He described Eschenbach's arrival as a release.
The pattern carries a cost. Each transition took time, and the 2024 handoff reversed within two years at a much lower stock price. A founder who treats the top job as adjustable must also accept that the market reads each change as a signal. The flexibility worked best when it was planned in advance, as Eschenbach's path to sole CEO was announced a year early, and worst when it happened suddenly.
◐
The Configurations of the Top Job
Bhusri's titles at Workday, 2005–2026.
2005
Co-founds Workday with Dave Duffield; Duffield is CEO, Bhusri is president.
2009
Bhusri and Duffield become co-CEOs.
2012
Bhusri becomes chairman; Workday goes public.
2014
Duffield steps back in May; Bhusri becomes sole CEO.
2020
Chano Fernández named co-CEO in August.
2022
Carl Eschenbach of Sequoia named co-CEO in December.
2024
Eschenbach becomes sole CEO; Bhusri becomes executive chair.
2026
Board appoints Bhusri CEO again on February 6; Eschenbach departs.
Tactic: Write down which parts of the CEO job you do worse than a hired operator would, and design the leadership structure around that list, announcing transitions well before they happen.
Principle 9
Separate what must be deterministic from what can be probabilistic
Bhusri's strongest strategic argument on his return rests on a distinction anyone can use. Business processes such as payroll, compliance, and security "are deterministic by nature. There is a start and end." AI is "probabilistic by nature." "You can't have probabilistic outcomes in running a payroll." So AI should be layered on top of a deterministic core: prompt-based interfaces, automated workflows, and agents that act within guardrails the ledger enforces.
The distinction matters outside enterprise software. Every company has processes where a mistake is fatal and processes where a good-enough guess is valuable. Leaders who apply the same tooling, risk tolerance, and speed to both will either be reckless with the first or overly cautious with the second.
◇
Where Each Mode Belongs
Applying Bhusri's distinction to operating decisions.
| Process | Mode | Why |
|---|---|---|
| Payroll execution | Deterministic | Must be correct and complete every time. |
| Statutory compliance | Deterministic | Outcomes must be auditable. |
| Candidate sourcing | Hybrid | AI can rank; humans decide, and the bias risk is real. |
| Skills matching | Probabilistic | Pattern-matching across normalized data adds value. |
| Interface and navigation | Probabilistic | Prompts can replace multi-screen workflows. |
Tactic: Classify each of your company's core workflows as deterministic, probabilistic, or hybrid, and require human sign-off wherever a probabilistic system touches a deterministic outcome.
Principle 10
Buy on offense, but count the integration bill
Bhusri defended the $1.55 billion Adaptive Insights deal as "absolutely an offensive play," a chance to deliver what client-server planning had promised and failed to provide. That framing distinguishes acquisitions that extend a thesis from those that patch a hole. The later wave of AI deals, HiredScore, Evisort, Paradox, Flowise, Sana, and Pipedream, showed the cost. Forrester warned of "integration drag" and a company "struggling to fully integrate these assets."
Offensive acquisitions are still only as good as the integration that follows. Bhusri's own 2013 observation, that incumbents "buy their way into" each new wave, applies to Workday too. An acquisition program can look ambitious and still produce a set of loosely connected products.
Tactic: For every acquisition, write the date by which the acquired capability must run natively inside your core workflows, and treat missing it as a failed deal.
Principle 11
Don't confuse retention with love
Andreessen Horowitz's critique hits hardest on one point. Near-100 percent renewal can reflect captivity rather than satisfaction. Workday's moat, as the essay described it, included proprietary configuration tools, integrations that take six to eighteen months, and a group of more than 10,500 certified consultants whose skills only work inside Workday. That moat protected revenue for years. It also recreated the conditions Workday was founded to fix.
The lesson for any successful incumbent is uncomfortable. The metrics that look healthiest, retention and expansion, may be measuring switching costs rather than value. The company that once attacked PeopleSoft's lock-in now faces the same argument from the next startup.
Tactic: Survey your most locked-in customers about whether they would choose you again if switching were free, and treat any gap between that answer and your renewal rate as risk.
Principle 12
Return as a founder, not a caretaker
When Bhusri came back, he did not promise continuity. He promised a "re-founding." He studied how Steve Jobs ran Apple after returning, "the biggest start-up in the world," and took specific operating changes from it: fewer layers, faster decisions, clearer ownership. He hired a founder, Sana's Joel Hellermark, as chief AI officer and set up an "agent factory." He also returned to his earliest habit of going out personally, meeting dozens of customers to test the claim that none planned to replace Workday.
A returning founder's main asset is permission, the authority to undo what professional managers built. Bhusri's voting control gave him that authority formally. His framing of Chapter Four gave him the story to use it. Whether it works depends on whether the start-up practices get adopted or only described.
Tactic: If you return to lead something you built, name the three layers, processes, or committees you will remove in your first ninety days, and announce them publicly.
IV
Workday's Four Chapters, in Bhusri's Telling
How the co-founder frames the company's history.
| Chapter | Period | Focus | Phase |
|---|---|---|---|
| One | 2005 onward | HR and finance reimagined in the cloud | Found |
| Two | Post-IPO | Hypergrowth and product leadership | Scale |
| Three | 2023–2026 | Operational excellence and efficiency | Optimize |
| Four | 2026– | Re-founding around AI | Reinvent |
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
— Aneel Bhusri, to CIO magazine, 2008One of most important things for us was that [we had to make sure] we were doing this for the right reasons — not just for nostalgic reasons. We had a great run at PeopleSoft, but there really had to be an opportunity to build the next great applications company.
— Aneel Bhusri, to AllThingsD, April 2013The legacy enterprise companies aren't innovating, but they have cash and they have distribution, so they can buy their way into this new generation of innovation.
— Aneel Bhusri, to Cloud WarsInternally I described my role to the company as 'the pied piper of machine learning.'
— Aneel Bhusri, June 2020I think you never know if your value system really works until you hit a crisis.
— Aneel Bhusri, on returning as CEO, 2026With AI, we are essentially a start-up again.
Maxims
- Follow the person. A mentor who buys the new hire a beer can be worth more than a company with a better market.
- Aim at your own past work. No one knows a legacy product's weaknesses better than the people who built it.
- Design governance around the specific way you were hurt before. Abstract best practices protect less than lessons learned firsthand.
- Hire by hand, then build the system. The founder sets the standard and the process scales it, and both steps are necessary.
- Culture shows up in the numbers late. By the time financial results reflect a bad culture, the damage is done.
- Ask one question until people prepare for it. Repeating a question can steer an organization more effectively than a mandate.
- Adjust the title to fit the work. Arrange the top job around what you do well, and announce changes early.
- Keep the ledger reliable. Put probabilistic intelligence on top of deterministic processes, never in place of them.
- High retention is not the same as satisfaction. High renewal can mean customers cannot leave rather than that they want to stay.
- Come back to rebuild, not to maintain. A returning founder's advantage is permission to undo what others built.
