Part IThe Story
At six o'clock one morning in February 2001, a month before the American recession would officially begin, Tom Siebel sat at home in his sweats and logged on to his company's internal website. It was ritual. Every morning the founder of Siebel Systems reviewed his sales force's forecasts for the quarter: every deal in play, its dollar value, its distance from signature. For seven straight years those forecasts had told essentially one story, which was that the company would double. In 2000 it had, to $1.8 billion in sales and $258 million in profit. Siebel held 70 percent of his core market, and his stock had risen fiftyfold since its 1996 offering. The plan for 2001, set in ink only weeks before, called for $4 billion.
What the screen showed him that morning, in blue and red and black, was a field of stalled deals. There were hundreds of them, ranging from a few thousand dollars to several million, frozen in place. The backlog of pending sales was thinner than it had been days earlier. In the free-text notes his salespeople typed into the system (the software was his, the salespeople were his, and the discipline of writing everything down was very much his), a new vocabulary had appeared: budget eliminated. All IT spending frozen. Decision deferred to following quarter.
"Here we were, pedal to the metal, and suddenly the pipeline looked really bad," he said afterward. Hours later he called in his six senior executives and told them the growth plan was dead. He sent them out to visit key customers and close whatever could still be closed before the rest of the world figured out where the economy was going, and he visited nine accounts himself over the following weeks. On February 27 he warned Wall Street analysts on a conference call of a possible technology "depression." Then he cut. He laid off 800 people, shut down three money-losing business units, slashed travel, marketing and hiring, and took 20 percent off the pay of the top brass. "By week one of April, we turned a $4 billion company back into a $2.6 billion company," he said. "We turned this Titanic around." Economists needed nine months to confirm that a recession had started in March. Forbes ran the story in January 2002 under the headline "The Man Who Sees Around Corners."
Keep that image in view: a man alone before dawn, reading a screen, seeing first.
Early in 2025, Siebel was seventy-two, worth somewhere near $4 billion, and running his second public software company. That year he was diagnosed with giant cell arteritis, an autoimmune inflammation of the blood vessels that, in his case, caused significant visual impairment. In July he announced he would seek a successor. In September he gave the chief executive title of C3 AI to a former federal procurement official and moved to executive chairman, explaining that the illness had kept him from participating "as effectively as I used to in the sales processes." It was not a retirement. In May 2026, eight months later, he took the job back.
The man who saw around corners could no longer see well. He returned anyway. Everything that follows happened between those two mornings, or before them.
By the Numbers
Two Companies, Two Booms
$50,000Capital used to found Siebel Systems in 1993, with no venture money
45%Siebel Systems' peak share of the global CRM market, 2002
$5.85BPrice Oracle agreed to pay for Siebel Systems, September 12, 2005
$311MTom Siebel's gains from exercising stock options, 1999–2001
19Reconstructive surgeries after a 2009 elephant attack in Tanzania
$177.47C3 AI's all-time high closing price, December 22, 2020
$10.40C3 AI's share price in February 2026
The Note on the Back of a Business Card
He did not intend to become a salesman, and he did not intend to become an entrepreneur. He has said both things often enough that they are part of the legend, and they hold up. "I am a computer scientist," he told a room of University of Illinois students in September 2024, by way of introduction. He was seventy-one, a billionaire, and the namesake of the school they attended. He meant it as a statement of identity rather than a credential. When he finished graduate work in relational database theory, he was recruited by a small California company he had never heard of. "I had actually never heard of Oracle," he admitted. He took the job because of the "very bright people with high levels of energy" who worked there. "Turned out to be a pretty good idea."
The size of that company depends on who is counting. Siebel has put Oracle at about twenty employees in the United States when he arrived in 1984, and in one interview he agreed that he was employee number twenty. A reference biography puts the headcount at forty. Everyone agrees on the larger point: the worldwide market for relational database software, by Siebel's estimate, might have been $5 million. He started as a presales representative in the Chicago office, setting up database demonstrations so that actual salesmen could close deals with prospective clients. Within a year he was Oracle's top salesman, outselling colleagues with something few of them had. He understood the machine. He could take the jargon of a technology still mostly confined to graduate seminars and turn it into the language of purchase orders.
Larry Ellison, the young founder who recruited him, was by then already the kind of figure who gets compared to other figures. He had passed through the University of Illinois himself without staying, built a company on a database theory IBM had published and neglected, and sold software before it fully existed. It is tempting to see the two men as mentor and protégé. The record suggests something cooler: two salesmen who recognized each other. When Marc Benioff, an early investor in Siebel Systems who went on to compete with it, wanted to describe Siebel's gift, he reached for the one comparison that would mean the most in Silicon Valley.
— Marc Benioff, to Forbes, December 2000Tom can sell anything to anybody. He's as good a salesman as Larry Ellison.
By 1988 Siebel was running Oracle's Direct Marketing Division, and that August a headhunter sent him a candidate for division controller. Mei Lin Fung had come from Intel, where she had been the alpha user of the chipmaker's distribution marketing system. That was a world of stagger charts and forecast-variance analysis and what she called "rigorous accountability for making projections and tying them to what actually happened." Siebel, she recalled years later, had been unable to find anyone who fit his peculiar specification: finance, marketing, sales, software, people skills, "and a propensity for risk-taking." What he described to her in that hour was a vision of technology that could serve "the prima donnas of sales" while accommodating each salesperson's quirks. In other words, he wanted to give Oracle's own sales force what Oracle had been promising its customers.
Oracle's internal information-systems group told him it couldn't be done. Fung's paraphrase of their verdict was "fergedaboutit." So in September 1988 Siebel put Fung and Kevin Kraemer, the division's top presales technical man, to work on the project on the side. The result was OASIS, the Oracle Automated Sales Information System. Fung would call it the genesis of the CRM industry. People who had worked in the division and used OASIS were later hired away by other software companies so they could replicate what became known as the "Oracle Sales Model." By Fung's count the division produced two billionaires, Siebel and Vinny Smith. A third, she predicted in 2006, would be the young man who took over the division when Siebel left: Marc Benioff.
In 1989 Siebel proposed turning OASIS into a commercial product. Ellison declined; he said he saw no commercial potential. Siebel remembers the exchange as a suggestion that Oracle "build this out as a division" and Ellison as having "no interest." What he saw and Ellison didn't was simple, and he has described it many times in nearly the same words. In the early 1990s, he said, "the state-of-the-art in sales was still a note on the back of a business card." Accounting had been computerized, and so had manufacturing and human resources. The department that brought in the revenue was still running on memory and handshakes and cardboard. "I thought that was highly unlikely to remain the case."
In 1990 he took a leave of absence from Oracle, and he never went back. The next year Oracle nearly collapsed. Forbes described it as "the victim of an overzealous sales force and creative bookkeeping," failures of exactly the discipline Fung had been hired to impose on his division. Siebel later said he saw the train wreck coming and got out of the way. One of the more revealing things he has ever said about those years came in a 2019 interview, as an aside: "I learned a great deal about how to run a company and I think I might have actually learned even more about how not to run a company."
A History Major North of Green Street
The forecasts, the screen, the discipline of tying projections to outcomes: all of it came later and from somewhere else. Thomas M. Siebel was born in Chicago on November 20, 1952, the sixth of seven children of Arthur Francis Siebel, a Harvard-educated corporate attorney, and Ruth Schmid Siebel, a homemaker. He grew up in Wilmette on the North Shore, in a household one biographer describes as upper-middle-class. When he was fifteen his father sent him to Shattuck Military Academy in Faribault, Minnesota. He disliked the regulations, but he stayed to please his father, and over time he came to value the self-reliance the place forced on him.
Hold on to that detail. Decades later, in the Silicon Valley of dogs in the office and jeans at the board meeting, Siebel would run a company where salesmen wore suits and cubicles were kept neat. The press would trace the habit back, almost reflexively, to a boy in uniform in Minnesota. "Siebel, who spent his high school years at a suburban Chicago military academy, brought along his Midwestern values," the Associated Press wrote in 2001, misplacing the school but not the point.
He went to the University of Illinois at Urbana-Champaign and studied history. He joined the tae kwon do club. His most vivid memories of college are of sitting on a bench on the Quad with a book and wandering the library stacks. "That's where I studied," he said of the Stacks in the Main Library. "Immediately surrounded by five million volumes that approximated the sum of all human knowledge, you could get a visceral hint of what it must have felt like to be a polymath like Thomas Jefferson or Benjamin Franklin who had an immediate grasp of everything that was known." Then, drily: "But Franklin and Jefferson had immediate recall in their minds. In the Stacks, you might have to walk a few flights of stairs to get at it."
He took his bachelor's degree in 1975 without knowing what he wanted to do, so he drove west in a pickup truck and worked as a ranch hand in Idaho. (Montana, where he would one day own working cattle ranches covering tens of thousands of acres, comes into the story here too. In 2006 he was described as having a history in the state "spanning 35 years.") Eventually he went back to Chicago and worked for a trade book publisher. Then he enrolled in the MBA program at Illinois.
There a single text redirected him, which is a very Illinois-history-major way for a career to turn. As a business student in the early 1980s he read the Harvard sociologist Daniel Bell. Bell's 1973 The Coming of Post-Industrial Society predicted an economy organized around information rather than goods. Bell, Siebel said, "predicted that computers and information technology were going to change everything about the way that we communicate, the way that we educate, the way that we entertain ourselves and the way that we work. This was before the mini-computer, before the personal computer, before the Internet, certainly way before the iPhone." So he crossed campus, "north of Green Street," as he put it, and enrolled in computer science, "so that I could have the tools to play in what was going to be a very big and very real game."
He programmed Control Data CDC Cyber mainframes on punch cards and a PDP-8 on paper tape. The university's records date his computer science master's to 1985, and a reference biography dates it to 1983. Both degrees hang on the wall of his California office. When he spoke to Illinois students by video in 2024, all four of his Illinois degrees were displayed behind him, the honorary doctorate of engineering from 2006 among them.
His curriculum shaped everything that came later: history first, then business, then the machine. In 2024 he told Illinois students that the pressure to study only STEM subjects was "a bunch of bunk." He reads history, he says, to prepare for the future. "I have a book in my hand all the time."
The Worst Address on the Planet
Siebel Systems was not his first company. After leaving Oracle in 1990 he took over Cayenne Systems, a small, privately held multimedia software company, and renamed it Gain Technology. By one account his main job was to prepare it either for a sale or for a public offering. It got the sale: Sybase bought it in December 1992 for roughly $100 million. (Forbes says $105 million.) It was a competent, unglamorous exit, the kind that buys a man the freedom to do what he actually wants.
What he wanted was OASIS without Larry Ellison. The partner he chose was Patricia House, whom he had met at Oracle in 1986 and who left the company in 1992. House's public persona was as intense as Siebel's, if quieter. She later co-wrote a book with him, Cyber Rules, and in 2000 Fortune put her on its list of the fifty most powerful women in business. Speaking at Wharton that December, she described what the company did in a way that sounds like a joke until you read the follow-up: "It's brain surgery," she said, and then, seriously, "It really is." She and Siebel, she said, had been "maniacally focused." The phrase was hers. It applied to both of them.
They founded Siebel Systems in July 1993 with $50,000 of their own money and twelve employees. House called their East Palo Alto office "the worst address on the planet." The office furniture came from bankruptcy auctions. Employees were offered their pay in cash or in stock, and according to Siebel every one of them chose stock. He called it, laughing at himself, "my Marxist dream." When the company went public, roughly seventy of those early employees became millionaires. Siebel said employees owned 85 percent of the company at the offering and he owned the rest.
There was no venture capital, and Siebel was blunt about why. "We didn't need money, connections or product development help," he said. He and House did not want to give up control either. That preference would follow him for the next three decades, and a dual-class share structure would eventually make it legal and permanent.
When they started, roughly 400 vendors were competing in what was then called sales force automation, and most of them were selling glorified electronic Rolodexes. Siebel wanted something else. He set three requirements for the software. It had to work as well for a 50-person sales force as for a 5,000-person one. It had to work across languages and currencies. And it had to be customizable for different industries. To build it, he hired William Edwards, who had run engineering at Frame Technology. The first product shipped in 1994, priced between $3,500 and $6,500 per seat and running on Windows NT. Microsoft co-marketed it.
The early customers were the real genius of the operation, and Siebel picked them the way a novelist picks a cast. In the fall of 1994 he sat down with Charles Schwab, the man who had turned a discount brokerage into a revolt against Wall Street's commission system. Schwab left the meeting with a seat on the Siebel board and a 2.5 percent stake, worth $14 million by 1997. More than 4,000 Schwab brokers ended up using Siebel software to check customers' trading profiles mid-pitch, and the time saved let them make three times as many sales calls. Then came Andersen Consulting. In 1995 Andersen rewrote its own bylaws so that George Shaheen, its managing partner, could join Siebel's board. Andersen took a 10 percent stake and began recommending Siebel's software to its clients, and before long it had more than 300 technicians whose only job was installing Siebel. In March 1997 Compaq agreed to equip its sales reps with Siebel software and collaborate on hardware design, and Siebel agreed to promote Compaq machines in return. Cisco gave Siebel software to more than a thousand salespeople.
So every important early customer was also, in some way, an owner, a director, a distributor, or a partner. In 1997 Forbes compared the company to the Borg, the hive civilization in Star Trek that assimilates its victims into a single mind. The comparison was only partly unkind. The software, at up to $6,000 per salesman, let "squadrons" of salespeople operate as one: who on the team had talked to whom, who actually made the decision at Chrysler "and who the lackeys are," how close the deal was to signature. Revenue went from $8 million in 1995 to $39 million in 1996. Siebel summed up the whole theory of the product in one sentence, as tersely as anything he has said: "If I can answer your questions, we're gonna do business."
A Salute from General Patton
In December 2000 Forbes opened its profile of the forty-seven-year-old chief executive with a set of questions for anyone hoping to work for him. Want to bring your dog to work? Wear jeans to visit a customer? Play ball in the hallway? "Then don't bother applying to Siebel Systems," the magazine wrote, "where the 47-year-old chief executive demands conformity that would have earned a salute from General Patton."
Siebel did not dispute any of it. He made it a doctrine. "The notion of dressing in jeans and a T-shirt to greet the CEO of a major financial institution who just got off the plane from Munich is not acceptable," he told the Harvard Business Review in 2001. The magazine called him "buttoned-down" and said he put "discipline ahead of inspiration." To the Associated Press he listed his priorities in an order that sounded heretical in the year of peak Nasdaq: customers first, employees second, shareholders third. "I think a lot of CEOs spend a lot of their time worrying about their stock price and worrying about their investors," he said. "That strikes me as not a good use of a chief executive's time." He thought a chief executive should be meeting customers, designing products, recruiting employees.
The dress code drew the jokes, but the economics mattered more. In Siebel's telling, the company did not play a volume game. It played for market share and references. "We wanted to own at least 50% of every market we entered," he wrote in late 2000, "and to secure that, we turned down new customers until we were absolutely certain that our initial customers were going to be successful." He called this a "paradoxical strategy." At a time when competitors would sell to anyone with a budget, Siebel was turning money away so that each customer could become the reference for the next one. By his account the company owned two-thirds of every market it was in, and half its revenue came from repeat business. It beat Wall Street's expectations in every quarter from its 1996 IPO through at least the summer of 2001.
The essay in which he laid this out, published in Forbes ASAP in November 2000, is the best key we have to how he understood himself at the peak. It opens with a sentence few founders at the top of a bubble would write:
— Tom Siebel, Forbes ASAP, November 2000There is no substitute for being in the right place at the right time. Many of Silicon Valley's so-called entrepreneurial geniuses tend to forget this. I am continually amazed by the self-perpetuating myths that Highway 101 is afloat with geniuses, makers of insanely great technology. The story of Siebel Systems is quite different. We got lucky.
He meant it, and the modesty was also a sales technique. He listed the technologies that had happened to arrive together in the mid-1990s: replication, high-performance relational databases, Windows, 32-bit processing, object-oriented programming, high-bandwidth communications. Then he stated what he called "Siebel's fundamental law of business management": When you find yourself in the right place at the right time, don't foul up the opportunity. Twenty years later, speaking to business students, he had loosened the wording but kept the idea. "When you're in the right place at the right time — and some of you will be there, and you will know it — things will feel a little bit magical, and maybe surreal," he said. "And when that happens to you, you gotta not goof it up."
The dot-com crowd got his scorn, sometimes mixed with something close to affection. "There has been just a wonderful blossoming of creativity and entrepreneurism," he told the AP in August 2001, and called the previous two decades in the Valley "an economic miracle unprecedented in the history of mankind." Then he turned: "That being said, I think there has been a lot of just rank amateurism out there during the last five years." His advice to a room of Wharton and Harvard alumni around the same time was almost insultingly simple: "Just figure out how much revenue comes in, and then spend less than that." People in the audience, a reporter noted, scribbled frantically.
He also hired, he said, people smarter than himself. That claim deserves the same skepticism and the same credit as the luck. He was the most technically fluent salesman in the industry, he demanded conformity, and he logged in at six every morning to read the forecasts personally. It is hard to picture anyone in the building feeling smarter than him for long. But he did delegate, eventually and painfully. He told a visiting group of Wharton fellows that he had rebuilt his management team three separate times. And he learned something that genuinely surprised him from two board members, Schwab and Shaheen. "These people prided themselves on doing absolutely nothing," he said, with something like wonder. By 2002 he had reached his own version of the idea: "If I'm doing anything at all, there's a problem."
What the Pipeline Knew
Go back to that February morning, now that we know what it cost him to be right.
The forecasting system that showed him the recession was more than a product he sold. Through it he watched the whole company and, through the company, the economy. Fung had brought Intel's stagger charts to his Oracle division in 1988. Twelve years later the same discipline, rebuilt as commercial software and sold to thousands of companies, was running in his own house, and on a few mornings it gave him what looked like a view of the future. The first quarter of 2001 brought in $335 million in license revenue, $30 million less than the fourth quarter of 2000. It was the company's first sequential decline ever, and it was still 72 percent above the first quarter of 2000. Wall Street took it in stride. Siebel had braced, and the bracing worked.
The same instinct got him in trouble that November. On November 5, 2001, at an invitation-only Goldman Sachs conference, he told the audience he was optimistic and that business was returning to normal. Three weeks earlier he had publicly called the IT market tough and said the company expected that climate to last the rest of the year. Siebel's stock rose about 20 percent from the previous day's close. The trouble was that most investors had no access to what he had said. Under Regulation FD, a rule that had taken effect in October 2000 to stop companies from selectively disclosing material information, that was a violation. The SEC's first enforcement actions under the rule named three companies, and Siebel Systems was one of them. The company agreed to a $250,000 penalty.
The episode was small and it was also revealing. Siebel had read the market correctly once again; in January 2002 he told Forbes the recession had bottomed at the end of November and predicted IT spending would recover in the first half of the year. His mistake was that he could not keep a reading to himself. The same quality made him a magnificent salesman, an alarming interviewee, and a periodic legal liability. At a 2002 trade show keynote he told an audience that CRM, the industry he had more or less invented and still dominated, was "dead." He later explained that he meant a new generation of applications would make the current ones obsolete. Hardly anyone in the room heard it that way.
He was also never more himself than in the months after September 11, when Siebel Systems brought out a homeland security product that let government agencies track suspects the way a company tracks its customers. The Transportation Security Administration was using it, he said, to screen and train newly hired airport security staff. Asked by a reporter in 2002 whether computer profiling could go too far, he answered with the September 11 timeline, recited the way a salesman walks a customer through a pipeline review. "We knew the day Mohammed Atta met with Iraqi agents in Prague," he said. "We knew the day he entered the country. We knew the day his roommate (Marwan) Al-Shehhi got a $100,000 wire transfer from an al-Qaida operative. We knew all these facts. If they had been able to associate these facts, people never would have blown up on Sept. 11." Then came the close: "It's not a question of if we're going to share information across agencies about these things. The only question is how many people are going to die before we do it."
When the reporter pressed him on privacy, he snapped: "Hello. We're doing everything but body cavity searches." (The Prague meeting he cited was among the most contested intelligence claims of the period. He stated it as plain fact, the way he stated most things.) Critics at the time accused him of exploiting the war on terrorism to drum up business. He did not appear to see much of a line between national security and a well-kept database, and his record suggests he never has.
That same year, asked whether he worried about competitors, he answered in a way that was half boast and half confession. "I'm worried about everyone gaining on us every minute of every day. Are you kidding? This is what I do for a living." In another interview he said simply: "I stay awake at night worrying about it." One headline about him that season read: "Paid to be paranoid."
Handing Out Options Like Candy
His paranoia covered competitors and the economy. It did not extend to his own pay.
Between 1999 and 2001, according to SEC filings, Tom Siebel realized $311 million in gains from exercising stock options. That was the stretch when Siebel Systems stock went from Wall Street darling to a peak of $119.88 in 2000. By the summer of 2003 it had fallen 92 percent, erasing nearly $47 billion in shareholder wealth. In September 2002 the Teachers' Retirement System of Louisiana sued the company's board. The fund alleged that the directors had breached their fiduciary duties by authorizing excessive compensation for the chief executive, and that options had been granted below fair market value in violation of the company's approved plan. Stuart Grant, the Delaware attorney representing the pension fund, offered the most quotable description: "They were handing out stock options like candy."
The board in question was a who's who of the American establishment, which is how it came to be called clubby. Besides Siebel it included Charles Schwab; Eric Schmidt, then CEO of Google; Marc Racicot, the departing chairman of the Republican National Committee, who was poised to run President George W. Bush's re-election campaign; and A. Michael Spence, the Nobel-winning economist and former Stanford business school dean, who years earlier had co-written a Stanford case study about Siebel Systems. The suit said Schmidt, a member of the compensation committee, had made $13.2 million in option gains between July 2000 and February 2001, and that Spence, the committee's other member, had made $22.6 million. The scornful version of the shareholders' argument fit in one line: the company's largest shareholder hardly needed more options to be motivated.
In January 2003 Siebel surrendered about 26 million options, valued at roughly $56 million. That was every option the board had granted him over the previous four years, and it cut his stake from 13.5 percent to 10.7 percent. He had already limited his salary and bonus to $1 each for the two prior years. None of it stopped the lawsuit. In July 2003 a San Mateo County Superior Court judge, Quentin Kopp, ruled that the directors could be held liable for punitive damages if the shareholders proved their case. "We will be looking for a very large number," Grant said. "These guys are all worth a ton of money." The company settled in August, before a trial scheduled for November 3, by adding an outside director, expanding its compensation and governance committees, limiting director options, submitting its option plans to annual outside review, and paying the plaintiffs' lawyers up to $900,000. Grant called it "a very large step in the right direction for a board that had a reputation for being one of the worst around." Tom Siebel, the AP reported, defended his gains as his just deserts for building a software giant.
The business had been sliding the whole time. After peaking at $2.1 billion in 2000, by the AP's count, annual sales fell to $1.35 billion in 2003. The company cut thousands of jobs, ending up with about 5,000 employees, and still lost a combined $128 million over two years. The products that had once let squadrons of salespeople act as one mind were now criticized as too expensive and too complex to install, in a corporate America that had grown careful about technology spending. In 2002 Nucleus Research published a report suggesting most Siebel customers were not getting a return on their investment. Siebel called the study "methodologically unsound, statistically insignificant and factually incorrect," and then could not resist a professional compliment: "This was one of the best pieces of guerrilla marketing I've seen in my 25 years in the industry. I was a little bit impressed." When Computerworld asked whether the company was in trouble, he answered with the balance sheet. "Let's call a spade a spade. We have $2 billion in cash and have been generating a half-billion a year in free cash. Does that look like an ongoing concern to you?"
At the low end, meanwhile, Salesforce.com was coming up fast. It was run by Marc Benioff, who had once taken over Siebel's division at Oracle and had been an early investor in Siebel Systems. He had built a company on renting CRM over the internet rather than installing it on customers' servers. In the 2002 interview Siebel brushed off the threat from both the big suites and the small upstarts. "There is no market for integrated software," he said. Three years later the integrated-software company bought him.
On May 3, 2004, Siebel stepped down as chief executive and handed the job to Michael Lawrie, IBM's head of sales and distribution, while staying on as chairman. "After giving it an awful lot of thought, I was absolutely certain that this is the right decision for the company," he told analysts. Patrick Mason of Pacific Growth Equities put it more plainly: "He just got worn down." A month later, in a videotaped deposition during Oracle's hostile pursuit of PeopleSoft, Larry Ellison said that if the PeopleSoft deal failed, Siebel was next on his list, and that the interest ran both ways. "Tom Siebel came to my house and tried to sell me Siebel," Ellison said.
On September 12, 2005, Oracle announced it would buy Siebel Systems for $5.85 billion. The deal closed in January 2006. In what one account called a rare move, Ellison kept the name. More than twenty years later, Oracle still sells an enterprise CRM product called Siebel, versioned by year. The company Siebel left in 1990 because it would not build OASIS ended up owning OASIS's descendant, along with his name.
Two Hundred Yards
In September 2006, months after the Oracle deal closed, a reporter from the Helena Independent Record found him on horseback in the middle of a cattle drive on the Dearborn Ranch in Wolf Creek, Montana. A herd of black Angus was moving slowly from one pasture to another. A few confused calves kept breaking from the herd and had to be headed off by Siebel or one of the hands, whom the reporter named as Todd Krenning, Shane Hoopes and Pete Petersen. A young chocolate Lab named Pedro was trying, without much success, to be a cattle dog. Siebel got off his horse and started picking up trash: baling wire, rusted metal, glass discolored by decades in the weather. "Hey, this could be a meth lab from 1929," he said.
The joke was about the Montana Meth Project, which he had founded and into which he had put more than $10 million to fight what he called "the devil's drug." In 2006 the White House Office of National Drug Control Policy recognized it as the most influential drug program in the country. The campaign used graphic public service advertising aimed at teenagers, which made it a sales campaign in the strictest sense: a product, a target market, measurable results. Siebel owned the Dearborn and also the N Bar Ranch in Grass Range, both working cattle ranches. Forbes had put the family's Montana holdings at 77,000 acres in 2000, where he spent time "roping and riding with his wife Stacey." She was a former Oracle sales representative. Even his marriage started in the pipeline.
The giving was large and, as with everything else, organized around a few places he trusted. Illinois came first. He gave $32 million toward the Thomas M. Siebel Center for Computer Science, which opened in 2004. In 2006 came $4 million to endow two chairs, one in computer science and one in the history of science, the second a quiet acknowledgment of his first degree. In 2007 came a $100 million gift, a record for the university, for fellowships, faculty chairs, and work in alternative energy and stem-cell research. In 2016 he gave $25 million for the Siebel Center for Design. The Siebel Scholars program, which began with a corporate gift from Siebel Systems, now recognizes about eighty graduate students a year in business, computer science and bioengineering across five countries. Barron's ranked him among the world's twenty-five top philanthropists in 2009 and 2010, and in 2010 he received the Woodrow Wilson Award for Corporate Citizenship at a dinner in Palo Alto. Condoleezza Rice received the public service award that same evening, and she later wrote the foreword to his book.
In August 2009 he was in Tanzania with a guide, watching a group of elephants from about 200 yards away. One of them charged. It went at the guide first and then turned on Siebel. It broke several of his ribs, gored his left leg and crushed his right. Over the next two and a half years he had nineteen reconstructive surgeries, including a stretch of sixteen operations and an Ilizarov apparatus, an external frame of rings and rods bolted through the limb, to mend, lengthen and reshape the tibia of his right leg. He recovered fully. In 2013 National Geographic retold the story in a television series called Dead or Alive: Trampled on Safari.
Two hundred yards is the kind of number a careful man puts between himself and a risk he has assessed and decided he can manage, and the elephant covered it anyway. He has rarely talked about the attack in terms of fear or chance. What he took from it, he told the radio host Alan Fleischmann years later, was a rule for leadership: never consider the negative. Asked by an Illinois student in 2024 how to deal with setbacks, he answered like a man who had learned to walk again. "The world is about setbacks," he said. "You need to be on the balls of your feet; you need to have knees bent." He had once been a member of the tae kwon do club, and the stance is the same.
Measure, Mitigate, Monetize
He started his next company before the elephant attack, and he did it the way he did everything after Oracle: quickly, with his own network, and without asking venture capitalists for permission.
"Back in 2007 or 2008, everybody was talking climate, climate, climate," he told SiliconANGLE in 2023. He gathered a group of people, by another account about fifty, and spent roughly a year with them working out how to make a meaningful contribution to the conversation. "It became apparent that the best way to do it was not through philanthropy, but by appealing to the profit motive." So, in his telling, "I sent out an email on a Friday in December 2008 and we raised $20 million by Monday." The company was called C3. The C stood for carbon and the 3 for "measure, mitigate and monetize." He was chairman from January 2009 and became chief executive in July 2011. Pat House came along.
He has resisted every invitation to call the founding a revelation. "I read a lot. I have historical perspective. I know what's going on," he told an interviewer. "There was no epiphany." The company moved from measuring corporate carbon footprints to analytics for electric grid operators. Siebel explained the grid with the enthusiasm of a man who loves physical systems: generation, long-distance transmission, medium-distance transmission, the meter, the last few feet to the outlet. Utilities were wiring smart meters, thermostats, substations and transformers with sensors that reported their condition in real time, data arriving "60 or 90 hertz cycles — that would be 90 times a second — and from hundreds of millions of devices." Clean tech went out of fashion, and C3 became C3 Energy. Then, in February 2016, it became C3 IoT, for the Internet of Things. "I've been through mainframe computing, mini computing, personal computing, relational database computing, enterprise computing, and the Internet," he told Fortune that month. "The next generation of computing is all about smart, connected devices. The future is all about IOT." The new company's list of target industries, Fortune noted with some amusement, was "exhaustive": oil and gas, manufacturing, aerospace, automotive, chemicals, pharmaceuticals, telecom, retail, insurance, financial services, government, defense and intelligence.
Then it became C3.ai. He presents each renaming as the same company noticing it had been right all along. "We realized that what we were doing for grid operators we could do for everybody," he said. The underlying idea, which he described to Illinois students as "designing a platform instead of applications," came straight out of the Siebel Systems years, with one difference. At Siebel he had built applications on top of other people's databases. At C3 he wanted to own the layer underneath. In a 2002 interview he had predicted that the next breakthrough in enterprise software would come "when we begin to view an application not so much as software, but as collections of best practices." He predicted in the same interview that within ten years the computer itself would be "like a paper clip in terms of form factor and cost." He was early on both, but not wrong.
In 2019 he published Digital Transformation: Survive and Thrive in an Era of Mass Extinction, with a foreword by Rice. It was his fourth book, after Virtual Selling (1996, with Michael Malone), Cyber Rules (1999, with House) and Taking Care of eBusiness (2001). The title tells you most of the argument. The phrase "digital transformation" had irritated him when he first heard it in boardrooms around 2010 in Shanghai, Rome, London, New York, San Francisco and Chicago. "I wondered if they had any idea what they were saying," he said, "digital transformation as opposed to what, analog transformation?" Eight years of talking to executives later, he had settled on a definition: elastic cloud computing, big data, the Internet of Things and AI, converging. He framed the stakes in Darwinian terms. More than half the Fortune 500 had vanished since 2000, he noted. "Most CEOs have no idea how massively disruptive the coming digital transformation will be or how quickly it will happen," he said around then. "My message to them: whatever you think you're doing, it's not enough."
He made his own employees take the advice. In 2019 C3 began paying the full cost of the University of Illinois's online Master of Computer Science for any employee who was admitted and didn't already have an advanced degree. Graduates got a 15 percent raise, a $25,000 cash bonus and more equity. By early 2020 four employees had been admitted. The company had also spent $1 million on bonuses for curated Coursera certifications. "It's some of the best money we've ever spent," he said. By his count, 83 percent of C3's employees already held advanced degrees. Perhaps it reflects a man who once crossed Green Street to retool. It is also cheaper than buying a forecasting system that tells you your engineers are falling behind.
Ticker Symbol AI
In May 2020, two months into the pandemic, Siebel talked with Fortune's AI newsletter. He was as outspoken as advertised. "No one here has any idea what they are talking about," he said of the American response to COVID-19, and he included many of the world's leading epidemiologists in "no one." C3 had spent about four weeks assembling what it called the world's largest COVID-19 data lake: twenty-two datasets from the World Health Organization, the World Bank, the Census Bureau, Johns Hopkins and others, cleaned and combined into one searchable knowledge graph and made freely available to researchers. "So the next time someone says something," he said, "they will know what the hell they are talking about instead of listening to this, candidly, idle speculation."
In the same conversation he predicted that eight out of ten AI startups would go bust. "You clear the silliness out of the market," he said. "All the Masa Sons are gone." He was referring to Masayoshi Son of SoftBank, whose $100 billion Vision Fund he considered the "poster boy" of irrational exuberance. His oil customers, he said, now faced a simple choice: work out how to make money at $26 a barrel or go out of business.
Seven months later, on December 9, 2020, he took C3.ai public on the New York Stock Exchange. He had secured a ticker that no amount of brand consultancy could have topped: AI. The offering priced 15.5 million shares at $42 to raise about $651 million. Spring Creek Capital, affiliated with Koch Industries, bought $100 million at the IPO price in a concurrent private placement, and Microsoft bought $50 million. Two weeks earlier the company had done a 6-for-1 reverse stock split. It also adopted a dual-class structure in which each Class B share carried fifty votes, and after the offering Siebel and related entities controlled about 72 percent of the voting power while holding about a third of the stock. The founder who had refused venture capital in 1993 to keep control had found the public-market version of the same refusal.
The shares more than doubled on the first day, closing at $92.49 for a market capitalization of $8.9 billion. On December 22 they closed at $177.47, and the next day they touched $183.90 intraday. The company was briefly valued above $17 billion. Every condition the old Siebel had mocked was in place: near-zero interest rates, a pandemic boom in cloud software, a market that had lost its sense of proportion. He had predicted that silliness would be cleared out of the market, and for a few weeks he was its main beneficiary.
The stock fell 77 percent in 2021, a good year for software generally, and another 64 percent in 2022. Siebel was criticized for selling hundreds of millions of dollars' worth of shares in the months after the offering. Investors sued, alleging that before the IPO the company had misrepresented its access to a 12,000-person sales force through a partnership with the oilfield services company Baker Hughes. Short sellers accused it of inflating margins, misclassifying revenue, aggressive accounting and opacity about how it counted customers. The company denied the allegations and accused the shorts of trying to "cover the short and pocket the profits." In June 2023, with generative AI suddenly everywhere and C3's shares up 210 percent for the year, Siebel sat down with CNBC, denied the lawsuit's allegations, and called his company "increasingly recognized as the gold standard in enterprise AI." In a season when the word "AI" alone could move markets, he owned the ticker.
CNBC also reached more than two dozen former employees. Most spoke anonymously, and they described a culture of fear that came from the top. One former leader compared Siebel to Logan Roy, the media patriarch of HBO's Succession: charming and charismatic, and a "tyrant" who "humiliates people." The comparison is unfair in the way television comparisons always are, and it is also hard to dismiss. A 2018 list of his honors includes a 100 percent CEO approval rating on Glassdoor from 2017. The same man inspired both. Like Siebel Systems in 2000, C3 was fully in the office in 2022, when most of the industry was not, and Siebel went on a podcast to say why. He also told the host he couldn't understand why newspapers were covering a celebrity trial while the world faced potential famine.
The politics got louder too. In February 2022 hacked records showed he had given $90,000 to the Ottawa convoy protest against Canadian COVID-19 mandates. In 2024 he gave $500,000 to Donald Trump's re-election campaign and organized a fundraiser for him in September. In 2025 he gave $1 million to the effort opposing California's Proposition 50, a redistricting amendment, which made him the third-largest single donor against it. His second cousin, Jennifer Siebel Newsom, is married to California's governor. He is a director of the Hoover Institution at Stanford. In 2026 he appeared with Rice in a Hoover documentary series called Only in America to discuss what he regards as the enduring foundations of American exceptionalism.
One more detail belongs in the record, and he would probably object to it more than to anything else here. The company he founded to help corporations measure and reduce their carbon footprints still carries the C for carbon in its name. In 2022, according to the tracking project ClimateJets, cited by Mother Jones, its founder ranked first among Americans for carbon emissions from private jet use.
The Quarter Was Dreadful
The illness came first, and then the numbers.
Giant cell arteritis inflames the arteries, and when it reaches the vessels that feed the eyes it can take sight. Siebel was diagnosed in early 2025 and disclosed the condition that year. On July 24, Bloomberg reported that he would look for a successor. In early August the company put out preliminary results for its fiscal first quarter, and in September the final figures: revenue of $70.3 million, down 19 percent from a year earlier, and a net loss of $116.8 million, against $62.6 million a year before. It was C3's nineteenth quarter as a public company and the first in which it missed its own revenue guidance.
He handled it as he had once handled Nucleus Research and the Goldman conference and a 2002 question about solvency. He was blunt to the point of rudeness, and the rudeness was aimed at himself. "The quarter was dreadful, okay?" he said. "Now we need to pick ourselves up, dust ourselves off and get on with business." Then he went through the possible causes the way he would once have gone through a forecast at dawn:
— Tom Siebel, on C3 AI's fiscal first quarter 2026 results, September 2025The financial results of the first quarter were completely unacceptable and completely unacceptable in virtually every respect. I've given this a lot of thought as to what the root cause of this is. Is there a market? The market is huge. Is there some new competitor that changed the competitive dynamics of the space? There is not. Is there some secular change in the market that we haven't seen before? There is not. The fact of the matter is this boiled down to poor sales execution and poor resource coordination.
New sales and services leadership had arrived mid-quarter and confused the sales process, he said. Then he named the other cause, in a way few chief executives would. "As a result of these health issues, I was unable to participate as effectively as I used to in the sales processes and the coordination of resources necessary to make these sales processes successful and come to closure." Twenty-three years earlier he had said that if he was doing anything at all, there was a problem. This was the other side of that remark, and it was almost unbearable to read. At the company he had built most recently, the problem turned out to be that he couldn't do enough.
The new chief executive, effective September 1, 2025, was Stephen Ehikian. He had been acting administrator of the General Services Administration, the federal government's purchasing office, where he had been responsible for implementing the President's AI Action Plan. The government had always been one of C3's best customers, and that December the Army selected C3 for a contested-logistics project and the Department of Health and Human Services chose it as an enterprise AI platform. Ehikian said publicly that "the ability to learn from Tom Siebel, who invented this entire Enterprise AI market," had made the job easy to accept. Siebel became executive chairman and promised to focus on strategic partners, strategic customers, and "direction and product strategy." Diginomica's headline put it more plainly than the press release did: this was not "a retirement story by any manner of means."
By February 2026 the stock was trading at $10.40, roughly a quarter of the IPO price and a seventeenth of the December 2020 peak. On May 12, 2026, the company announced preliminary results for the fourth quarter and the fiscal year in the same press release that announced Siebel's return as chief executive, with Ehikian becoming president. A product memo published under his name as chairman and CEO put the company's cumulative software investment at about fifteen years and more than $3 billion. At a fully loaded cost of about $375,000 per engineer-year, the memo said, that came to roughly 8,000 person-years of work. As of April 2025 he controlled 52.4 percent of the votes. The company his money had built over 8,000 person-years answered, in the end, to him, and he came back to run it.
It would be easy to end with a moral: the oracle humbled, the seer with failing eyes, the forecaster outrun by events. He would hate that, and it wouldn't be accurate. Asked in 2023 why a multibillionaire hadn't retired into venture capital, he said: "The honest answer is this is my idea of a good time." He added a line that sounds like a joke and isn't: "I'm not good at golf." To Illinois students in 2024, when the illness was still months away, he described C3's ambition to lead enterprise AI globally and added, deadpan: "I think there's some risk we might actually do that."
What a man who has kept a book in his hand his whole life does when the print goes soft, the press releases don't say, and I won't guess. The record does contain this. In October 2024, at the dedication of the Siebel School of Computing and Data Science at Illinois, he told the crowd that whatever he had learned about persistence and overcoming adversity, "to try to attain the unattainable," he had learned there. In the same remarks he offered a line that may be the one thing he has said that wasn't a sales pitch: "The best supercomputer is still the size of an apple and runs on sugar between our ears."
Some years earlier, asked about his favorite place on that campus, he had not named either of the buildings carrying his name. He named the Stacks in the Main Library, five million volumes that approximated the sum of human knowledge, where everything a person might need to know was within reach, provided he was willing to climb a few flights of stairs on a leg rebuilt nineteen times to get at it.
Part IIThe Playbook
What follows is an attempt to separate the method from the man, though with Siebel the two are unusually tangled. He built two category-defining companies a decade and a half apart, each on a thesis he had formed long before the market caught up, and each ended up as a cautionary tale about the costs of his own temperament. The principles below come from the evidence in Part I. Some are worth copying outright. Others are worth copying with the warning label attached.
The principles
- 1.Build what your employer refused to build.
- 2.Read the pipeline yourself, before dawn.
- 3.Turn customers away until the first ones win.
- 4.Make your distribution own a piece of you.
- 5.Refuse capital that only buys you supervision.
- 6.Use discipline as a product feature.
- 7.Cut early, cut once, cut deep.
- 8.Rename the company before the market does it for you.
- 9.Learn the plumbing; the CEO cannot outsource the technology.
- 10.Pay people to become more valuable to you.
- 11.Engineer your own redundancy, then test it.
- 12.Remember that your mouth is a material event.
S
Three Acts, One Thesis
The same idea, that the processes closest to revenue were the last to be computerized, recurs across four decades.
1984
Joins Oracle in its Chicago office as a presales representative; top salesman within a year.
1988
Launches the OASIS skunkworks with Mei Lin Fung and Kevin Kraemer after Oracle's MIS group calls it impossible.
1989
Proposes commercializing OASIS; Larry Ellison declines.
1992
Sells Gain Technology to Sybase for roughly $100 million.
1993
Founds Siebel Systems with Pat House and $50,000, no venture capital.
2001
Reads the recession in his own pipeline data; cuts 800 jobs by April.
2004
Steps down as CEO after a sales slump and a shareholder suit over options.
2005
Oracle agrees to buy Siebel Systems for $5.85 billion.
2009
Founds C3 to help companies measure, mitigate and monetize carbon.
2020
Takes C3.ai public under the ticker AI; shares more than double on day one.
2025
Hands the CEO title to Stephen Ehikian after a giant cell arteritis diagnosis.
2026
Resumes the role of chief executive.
Principle 1
Build what your employer refused to build
Siebel Systems is the commercial version of a tool Siebel had already built, used, and proven inside Oracle. OASIS existed. Oracle's own salespeople ran on it. Rival software companies hired away people who had used it so they could copy the "Oracle Sales Model." When Ellison declined to commercialize it in 1989, Siebel walked away with what matters most to a founder: a validated design, a list of real users' complaints, and a precise sense of what the incumbent wouldn't do.
This is different from the familiar advice to "scratch your own itch." The itch had already been scratched, and his own division was the customer. His 1993 company wasn't a guess about what sales organizations might want. It was a rebuild, with better enabling technology, of something that had already worked in production at one of the fastest-growing companies in the world. Gain Technology in between was a useful rehearsal in running a company and arranging an exit, but the thesis came from Oracle.
The risk is also part of the lesson. Founders who build what their former employer refused are betting that the employer's refusal was a judgment error and not a strategic choice, and that the employer won't change its mind once the market is proven. Oracle did change its mind, sixteen years and $5.85 billion later.
Tactic: Make a list of the internal tools your current organization has built for itself and refuses to sell. Those are pre-validated product hypotheses with a known first customer profile.
Principle 2
Read the pipeline yourself, before dawn
The February 2001 episode is a story about delegation as much as foresight. Siebel's executives could read the same forecasts he read. What set him apart was that he read them himself, every morning, and he read more than the numbers. He read the salespeople's free-text notes, where phrases like "budget eliminated" turned up before any aggregate figure moved. A CEO reading only dashboards would have seen a modest dip in the backlog. A CEO reading the raw notes saw the language of a coming recession.
The deeper point is that the instrument has to be designed so the leader can read it. Fung's Intel-trained practice of "tying projections to what actually happened" was in place at Siebel's Oracle division in 1988, and his company spent a decade turning it into software. By 2001 he had a forecasting system he trusted enough to bet the company on: within hours he threw out a growth plan that had been set weeks earlier. Economists took nine months to confirm what he acted on in a day.
The trap is believing the instrument sees everything. The same habit of personal reading made him confident enough to call the bottom early, and expressive enough to say so at a private Goldman conference.
Tactic: Pick one raw, unaggregated feed of frontline language, such as sales notes, support tickets, or lost-deal reasons, and read it yourself weekly. The words will change before the numbers do.
Principle 3
Turn customers away until the first ones win
Siebel described the early Siebel Systems strategy as "paradoxical." The company declined new customers until it was "absolutely certain" its first customers would succeed. In a land-grab market with 400 competitors, that looks reckless. In practice it built a compounding asset: every customer became a reference for the next one. He claimed it led to two-thirds share in every market the company entered and half of revenue coming from repeat business.
The early customers were treated as "product development partners" and involved in research and development of the first release. That changes what a customer is. It is no longer a source of a single transaction but a co-author whose success becomes the marketing. The approach works best in enterprise markets, where buyers call each other before signing and one failed implementation can poison a category.
R
Volume Game vs. Reference Game
How Siebel Systems' early go-to-market inverted the land-grab instinct.
| Dimension | Volume game | Siebel's reference game |
|---|---|---|
| Objective | Maximize logos signed | Own at least 50% of each market entered Scale |
| New customer policy | Sell to anyone with budget | Decline until first customers succeed Try |
| Role of first customers | Revenue | Product development partners Scan |
| Primary marketing asset | Advertising and sales coverage | Each customer as reference for the next |
| Failure mode | Churn and bad word of mouth | Slower early growth; complexity later hurts satisfaction Risk |
The strategy has a lifespan. By 2002 the complexity and cost of Siebel deployments had become exactly the kind of customer dissatisfaction the reference strategy was designed to prevent, and Nucleus Research was publishing it. Reference discipline at ten customers doesn't automatically carry to 4,500.
Tactic: Before you scale sales capacity, set a hard rule that no new segment opens until a defined number of existing customers in the current segment will take a reference call on your behalf.
Principle 4
Make your distribution own a piece of you
Charles Schwab received a board seat and 2.5 percent of the company, and then deployed the software to more than 4,000 brokers. Andersen Consulting rewrote its bylaws so George Shaheen could join the board, took 10 percent, and built a 300-person installation practice around the product. Compaq equipped its sales force and co-designed hardware. Microsoft co-marketed the first release. Each important early relationship combined equity, governance and distribution.
Siebel's reasoning was explicit: "anyone who owned a piece of his business would have an incentive to see him do well." This is a sharper idea than a partnership program. A systems integrator that owns 10 percent of a software company doesn't just recommend it. It builds a business unit whose economics depend on it. By late 2000, Siebel Systems reportedly had more than 700 alliance partners, but the shape of the ecosystem was set by the three or four partners who were also owners.
The same arrangement carried a later cost. A board made up of the people who were your biggest partners, customers and investors can drift into the "clubby clique" the Louisiana pension fund described. Aligned incentives in 1995 turned into a governance problem in 2002.
Tactic: For your two or three most important channel partners, explore structures that give them economic upside in your success, and pair every such arrangement with at least one board member who has no commercial relationship with you.
Principle 5
Refuse capital that only buys you supervision
"We didn't need money, connections or product development help." That was Siebel's whole case against venture capital in 1993, and it works as a diagnostic: if what an investor offers is money, connections and product help, and you already have all three, then what you are actually selling is control. He and House started with $50,000, furnished the office from bankruptcy auctions, and offered employees stock instead of salary. All of them chose stock.
Two effects followed. The employees who owned the company before the IPO (85 percent by his count) had every reason to run it as frugally as he did. And the founders kept the room to impose a culture that venture investors of the era might have dismissed as eccentric. At C3 he raised $20 million over one weekend in December 2008, by his own telling, from people who knew him, and then built a dual-class structure that gave him 50 votes per Class B share.
The counterpoint is in C3's later history. Founder control protects a thesis against impatient capital. It also shields a founder from the people who might tell him a quarter is going badly while he is too sick to see it himself.
Tactic: Before raising, list exactly what an investor would add beyond cash; if you can't name something specific, raise from people who already know your work, or don't raise.
Principle 6
Use discipline as a product feature
The dress code was never really about clothes. When Siebel said it was "not acceptable" to greet the CEO of a major financial institution in jeans, he was describing his customer. Siebel Systems sold to banks, brokerages, telecoms, insurers and governments, buyers whose own cultures prized predictability and who were deciding whether to trust a young software company with their most important relationships. Business attire, neat cubicles and an unbroken record of meeting Wall Street's expectations were all evidence that the company could be trusted.
His stated priority order (customers, employees, shareholders) and his advice to "figure out how much revenue comes in, and then spend less than that" belong to the same strategy. In 2001 he ran a profitable, cash-generating business while dot-coms burned money, and that was itself something he could sell. By 2002 he had $2 billion in cash, and when asked whether the company was in trouble, he could simply cite it.
The principle doesn't say "be formal." It says to match your internal culture to what your buyer needs to believe about you. At C3 the expression changed. A full return to the office in 2022 and an aggressive education benefit replaced the suits. The logic stayed the same.
Tactic: Write down the three things your most important customer must believe about your reliability, and audit whether your internal norms make those beliefs visible to them.
Principle 7
Cut early, cut once, cut deep
Between his February 2001 morning screen review and the first week of April, Siebel had cut 800 jobs, closed three money-losing units, frozen discretionary spending, and reduced senior pay by 20 percent. The company went from planning for $4 billion to planning for $2.6 billion. The first quarter showed the company's first-ever sequential decline in license revenue, and the market absorbed it.
The method had three parts. He acted on leading indicators, not lagging ones. He told the market himself, on February 27, before the numbers forced him to. And he made one large adjustment rather than a series of small ones, which is what turns a downturn into a slow-motion crisis of morale. The pay cut at the top mattered as a signal. Two years later he went further and took $1 in salary and bonus.
What the episode didn't fix is just as instructive. The company's revenue kept falling from its 2000 peak to $1.35 billion in 2003, and the cuts couldn't solve a product that customers increasingly found too complex and too costly. Decisive cost action buys time to fix the product. It doesn't fix the product.
Tactic: Pre-write your downturn plan, including which units close and what pay changes at the top, while times are good, so the decision on the bad morning is execution, not deliberation.
Principle 8
Rename the company before the market does it for you
C3, C3 Energy, C3 IoT, C3.ai. Each change could be mocked as trend-chasing, and in some quarters it was. Siebel's own account is more interesting. The company built a scalable platform to handle carbon and grid data, then noticed the platform was general-purpose. "What we were doing for grid operators we could do for everybody." Each new name described a market the existing technology could already serve, not a new technology.
The renamings tracked the vocabulary buyers were using: climate in 2008, IoT in 2016, AI later. A company that keeps an outdated name makes buyers translate, and buyers don't translate; they move on. The ticker AI, secured for the 2020 IPO, was the final step: the brand now matched the word investors were searching for. It was a remarkable branding achievement, and during the 2023 AI rally it was also a source of volatility.
C
Two Booms Compared
Siebel's two companies rode two different manias; the patterns rhyme.
| Dimension | Siebel Systems | C3 AI |
|---|---|---|
| Founding capital | $50,000, founders' own money | $20 million raised over one weekend, by Siebel's account |
| Founding thesis | Computerize sales, marketing and service | Measure and manage carbon, then all sensor data |
| Peak | Shares at $119.88 in 2000 Scale | Close of $177.47 on December 22, 2020 Scale |
| Drawdown | Down 92% by 2003 Risk | About $10.40 by February 2026 Risk |
| Controversy | Option grants; Reg FD fine | Post-IPO share sales; Baker Hughes suit; short reports |
| Control mechanism | Founder stake near 15%, later about 11% | Class B shares with 50 votes each |
| Outcome | Sold to Oracle for $5.85 billion | Founder returns as CEO in May 2026 Try |
Tactic: Once a year, describe your company in the exact words your best prospect uses for their problem; if your name and positioning require translation, change them.
Principle 9
Learn the plumbing; the CEO cannot outsource the technology
Siebel's advantage at Oracle was that he was "a salesman who understood the technology he was selling." He had a computer science degree in a world of salesmen who didn't, and he had written relational database theory as a graduate student. At Siebel Systems in 2002 he required every executive, the CFO included, to master one Siebel product and one competitor's product. "I'm a strong believer that knowledge goes a long way," he said. At C3 he turned the belief into an argument aimed at every CEO: "whatever you think you're doing, it's not enough."
What makes this less of a platitude is how he went about learning. He didn't read white papers. He read Daniel Bell and decided to go get a second degree. He keeps a book in hand constantly, reads history to prepare for the future, and has published reading lists for executives that start with information theory and go through machine learning. The technology knowledge sits on a base of historical perspective, which lets him see the cloud and AI as continuations of a seventy-year pattern rather than as novelties.
Tactic: Require every member of your executive team, including finance and legal, to be able to demo your product and explain your top competitor's architecture to a customer without help.
Principle 10
Pay people to become more valuable to you
The C3 education benefit is unusually generous and unusually specific. It pays full tuition for an Illinois online master's in computer science or data science, and on completion it adds a 15 percent raise, a $25,000 bonus and more equity. Before that, the company had spent $1 million on bonuses for Coursera certifications in AI, deep learning and elastic computing. Siebel called it "some of the best money we've ever spent."
The design is the point. It isn't a reimbursement capped at a few thousand dollars, and it isn't wellness spending. It is tied to a credential the company directly needs, and the payout comes only when the credential is earned. It also reflects his own history: he got rich because he went back for a second degree. In a field where technical skills go stale in a few years, paying employees to retool is cheaper than replacing them and builds loyalty a signing bonus can't buy.
Tactic: Identify the single credential that would most increase an employee's value to your business and fund it fully, with a compensation step-up triggered only on completion.
Principle 11
Engineer your own redundancy, then test it
Siebel's management philosophy at the height of Siebel Systems was "If I'm doing anything at all, there's a problem." He had learned it, with surprise, from Schwab and Shaheen, who "prided themselves on doing absolutely nothing." He rebuilt his team three times to get there. And he admitted the change cost him: "I have had to move on from the day when I was absolutely satisfied I could do any job better than anyone in the company."
The story's harshest lesson comes twenty-three years later. When illness kept him out of C3's sales processes in 2025, the company missed revenue guidance for the first time in nineteen quarters, and he publicly blamed his own absence along with mid-quarter leadership changes. Whatever the philosophy said, C3's ability to close deals depended on its founder more than anyone had tested. Within eight months he was back in the job.
Becoming dispensable isn't something you declare; you have to prove it under load. A founder who is still the best salesman in the company has a single point of failure in plain sight, and the only way to know how serious it is is to step away on purpose, while healthy, and watch what happens.
Tactic: Take yourself out of your company's most important recurring process for a full quarter, deliberately and while things are going well, and measure the result before illness or circumstance runs the test for you.
Principle 12
Remember that your mouth is a material event
The $250,000 SEC penalty in 2002, one of the first enforcement actions under Regulation FD, came from a single optimistic remark at a private conference that moved Siebel's stock 20 percent. The keynote line that CRM was "dead" took months of clarification. The September 11 profiling comments drew accusations that he was exploiting a national tragedy. His COVID remarks, his dismissal of the six-month AI research pause as "kind of wacky," and his political giving all came from the same source: a CEO who says exactly what he thinks, at volume, wherever he is.
Candor is an asset. It makes him trustworthy to customers and gives journalists a reason to call. His public line on the dreadful 2025 quarter, "completely unacceptable in virtually every respect," is a model of executive accountability. But a public company's CEO has no private opinions about its prospects. Every forecast he offers counts as a disclosure. Every provocation he aims at his own industry becomes something competitors can use.
Tactic: Treat any forward-looking statement you make in any room as if it were being simultaneously webcast, and if you wouldn't webcast it, don't say it.
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
— Tom Siebel, to Forbes, May 1997If I can answer your questions, we're gonna do business.
— Tom Siebel, LEADERS Magazine, 2019In the course of being there for about a decade, I learned a great deal about how to run a company and I think I might have actually learned even more about how not to run a company.
— Tom Siebel, to SiliconANGLE, April 2023The honest answer is this is my idea of a good time. I enjoy assembling really talented people and building industry-leading companies. I've done it a couple of times before and I think there's some probability we might do it again. I'm not good at golf. If I'm any good at anything, this is it.
— Tom Siebel, Reflections | Projections fireside chat, University of Illinois, September 2024The world is about setbacks. You need to be on the balls of your feet; you need to have knees bent.
— Tom Siebel, Siebel School of Computing and Data Science dedication, October 2024The best supercomputer is still the size of an apple and runs on sugar between our ears.
Maxims
- Don't foul up the opportunity. Luck puts you in the right place at the right time; whether it pays off depends on the discipline you bring once you're there.
- The refusal is the signal. When an incumbent declines to build something you have already proven internally, you may be holding its future.
- Read the words before the numbers. Frontline language changes before aggregate metrics do, so read the raw notes yourself.
- Every customer is the next customer's reference. Turn down growth that would put your first customers' success at risk.
- Make your channel an owner. Partners who hold equity build businesses around you; partners who don't just refer.
- Capital without a contribution is a claim on control. If an investor offers only money, connections and help you already have, decline.
- Culture is evidence. Run the company so its habits prove to buyers what they need to believe about you.
- Cut once, and early. One decisive adjustment on leading indicators beats a year of small retreats.
- Know the machine. A leader who can't explain the technology is relying on someone else's judgment about the future.
- Test your absence before fate does. Delegation isn't real until the company has run without you and been measured.
