Imagine a fund manager who loses money on eighteen out of every twenty investments she makes. Not loses a little: loses everything, the company folds, the equity is worth precisely nothing, the term sheet ends up in a drawer as a kind of relic. Imagine, further, that she does this quarter after quarter, fund after fund, for a working life of thirty years, and that at the end of it she is not merely forgiven for the record but celebrated for it, sought out by pension funds and university endowments, invited to speak at conferences, remembered as one of the great investors of her generation. In any other branch of finance this biography describes a disaster, possibly a criminal one. In venture capital it describes a very good career.
This plain fact makes venture capital strange, and the history only makes sense once the arithmetic is in view. A mutual fund manager who lost money on ninety percent of her positions would not survive a single bad quarter, let alone a career. A bank loan officer with that default rate would be escorted from the building, and possibly out of the profession, before lunch. Insurance is built, at enormous cost and seriousness, to make catastrophic loss happen as close to never as actuarial science can arrange. Venture capital inverts all of this. It is the one corner of organised finance in which failure, on a heroic scale, is not a symptom of bad judgment but the expected and budgeted cost of doing business at all. A venture fund that returns money to its investors by finding two or three extraordinary winners among twenty or so companies it backed, while the other seventeen or eighteen wrote off their capital entirely, has not merely done acceptably. It has very likely delivered one of the best returns available anywhere in the financial system that decade. Venture capital is optimism institutionalised, a machine for converting confidence in the future into companies, built on a distinctly American relationship with failure, and its real returns are rarer, stranger, and more concentrated than its mythology admits. Both facts belong in the same sentence, which is where the industry least likes to see them.
Nobody stumbled into this shape by accident; it was worked out, over a single American century, by a small number of identifiable people, starting with a Harvard professor who had spent the Second World War organising the funding of unproven military technology, and ending, for now, on a stretch of road in Northern California that has become shorthand for the whole enterprise.
The story has a prehistory, though, and a thoroughly American one. Bhu Srinivasan, in Americana, his history of American capitalism told industry by industry, opens with the observation that the country itself began as a venture: the Mayflower crossed the Atlantic as a speculative project, financed by London investors who put up the capital and waited out a horizon of years for a return that might never come, while other people bore the physical risk of the crossing, an allocation of duties the industry has honoured ever since. Venture capital's history belongs squarely inside that framing. So the named inventors of venture capital can wait. A trade that never used the word venture and never heard of Silicon Valley ran, without knowing it, on exactly the same mathematics.
The Ships of New Bedford
In the middle of the nineteenth century, the richest town in America per head, by most contemporary reckonings, was not New York or Boston. It was New Bedford, Massachusetts, a whaling port whose fortune rested entirely on financing voyages that stood a very good chance of coming back empty and a smaller chance of coming back astonishingly rich. The town found the arrangement agreeable. A whaling voyage lasted anywhere from one to four years. It sent thirty or so men into the Pacific or the Atlantic or the Arctic in a wooden ship, hunting an animal that could sink the boat that was hunting it, in pursuit of oil that lit the lamps of the industrial world before petroleum took the job over. Some ships came home with holds so full of whale oil and spermaceti that the owners' agents, the men who had financed the voyage, made fortunes that built the mansions still standing on County Street. Other ships came home clean, having found nothing, the whole voyage a dead loss of years and capital. And some ships did not come home at all.
No single investor could sensibly underwrite a whaling voyage alone; the risk of a single catastrophic loss was too large for any one purse, however deep. What New Bedford's merchants worked out instead, well before anyone had a word for venture capital, was a system of shared ownership that spread the risk of any one voyage across many small stakes, so that a bad captain or an unlucky season sank only a sliver of any investor's fortune. Ownership of a ship and its voyage was divided into fractions, an eighth, a sixteenth, a sixty-fourth, denominations otherwise encountered mainly in sheet music, and a prudent Nantucket or New Bedford investor held slivers of a dozen different voyages at once rather than the whole of any one. The crew were paid on the same speculative logic: not wages but a lay, a fixed fraction of the value of the eventual catch, scaled by rank, so that everyone aboard, from the captain down to the greenest hand, held what a later century would call equity in the outcome, and nobody was paid merely for time served. The mathematics of the arrangement did not require every voyage to succeed. It required only that the fleet as a whole, taken across enough voyages and enough years, produce more spectacular winners than the failures could offset, and it generally did, for as long as there were whales left to justify the trip.
There was, alongside the shares themselves, a whole supporting apparatus that later venture capital would reinvent in its own idiom without ever quite acknowledging the debt. Marine insurers in Boston and New York would cover a portion of a voyage's value, pricing the premium on the captain's record and the ship's condition, an early and none too scientific form of founder diligence. Agents ashore, who took a cut of the proceeds for organising a voyage, outfitting the ship, and hiring the crew, functioned as something close to the general partners of the arrangement: they did not go to sea themselves, but they staked their reputation and their own capital on picking captains capable of bringing a ship home full rather than empty. A captain with a run of successful voyages behind him could raise backing for his next one with an ease that a first-time captain, however able, could not, a pattern anyone who has watched a venture capitalist chase a founder's second company will recognise at once. The trade eventually collapsed, not because the mathematics stopped working but because the underlying commodity did: petroleum, cheaper and more abundant than whale oil, made the entire industry obsolete within a generation, a reminder, to be filed away for later, that even a well-run wager on the future can be undone by a better technology arriving from an unexpected direction.
What had not yet happened, in 1850, was the institutionalisation of that logic: its conversion from the private wisdom of merchants who knew each other at the docks into a repeatable, professionalised practice with its own vocabulary, its own legal structures, and its own class of specialists whose working life was organised around finding the next voyage worth backing. That conversion took the better part of a century, and it needed, as most institutional inventions do, a particular person in a particular institution at a particular moment. His name was Georges Doriot.
Georges Doriot and the Idea of Patient Capital
Georges Frederic Doriot was born in Paris in 1899, the son of Auguste Doriot, a pioneering automobile engineer who had raced and built cars for Peugeot before founding a motorcar firm of his own, a background unlikely to have made caution look glamorous, and he came to the United States in the early 1920s to study at Harvard, where he stayed for the rest of his working life, teaching at the Business School from 1925 until his retirement decades later. He became a naturalised American citizen, and in his manner and his convictions he was, by most accounts, more thoroughly American in his devotion to enterprise than many of the native-born businessmen who sat in his classroom. During the Second World War he was commissioned into the United States Army and rose to become a brigadier general, running the Military Planning Division of the Quartermaster Corps, which put him in direct, practical charge of funding the development of untried military technology and equipment at wartime speed and scale. The experience formed everything that came next. Doriot spent the war years deciding, again and again, whether to commit real resources to unproven ideas proposed by scientists and engineers, under conditions in which the ordinary caution of peacetime commerce would have been fatal. He learned, in other words, what it feels like to bet institutional money on people rather than on balance sheets, because the alternative, waiting for certainty, was not available.
When the war ended, Doriot did not want that habit of mind to end with it. The United States in 1946 was full of engineers who had spent the previous four years working on radar, on rockets, on code-breaking machines, on the countless small technical miracles that had helped win the war, and who were now being demobilised into an economy that had no particular institution designed to let them keep inventing things. The traditional avenues of American finance, the commercial banks, the trust departments, the wealthy families who financed businesses out of their own pockets, were built to lend against collateral and established cash flow, not against a plausible idea and a talented team. The Boston trust department of the period was an institution designed to ensure that nothing interesting ever happened to the money in its care, and it performed this duty superbly. A young engineer with a good idea for, say, a new kind of computing machine had almost nowhere sensible to go. Doriot's answer to this problem, worked out with a small group of collaborators that included Karl Compton, the president of MIT, and Ralph Flanders, the president of the Federal Reserve Bank of Boston and shortly afterwards a United States senator, was American Research and Development Corporation, founded in Boston in June 1946 with initial capital commonly cited at around three and a half million dollars, and generally regarded as the first institutional venture capital firm in American history.
ARDC was unusual in its structure in a way that mattered a great deal to Doriot personally. Rather than a private pool open only to a handful of very rich men, ARDC was organised as a publicly traded, closed-end investment company, so that ordinary investors, insurance companies, pension trustees, even individuals of comparatively modest means, could buy shares and take part in financing the postwar wave of invention. Doriot believed, with real conviction, that the financing of new enterprise should not remain the private amusement of a small number of wealthy families, and that a properly run institution could democratise access to it while exercising the discipline needed to make it work. Whether this was mostly high-mindedness or mostly a practical solution to a fundraising problem is a question his biographers have debated; it was probably some of each, which is true of most good ideas.
Doriot's temperament shaped ARDC as much as his theory of finance did, and it became, whether the people who inherited his industry knew it or not, the founding creed of everything that followed. At Harvard Business School his course was formally listed as a course on manufacturing, and it did cover manufacturing, but generations of students remembered it as a course on the entire discipline of building something from nothing: how to organise people, how to hold a team together under pressure, how to tell a founder who could be trusted with money from one who could not. Doriot was famous for a conviction he repeated to his students across three decades of teaching: that a mediocre idea in the hands of a strong, disciplined team would very often succeed, while a brilliant idea in the hands of a weak or undisciplined one would very often fail. Character, in his account, was not a soft consideration alongside the harder facts of a business plan. It was very close to being the whole of the analysis. This is not a comfortable thing to say to a room full of people trained to believe that ideas are what matter, and Doriot said it anyway, for thirty years, to students who went on to run a great deal of American industry. The classroom manner in which he said it became legend in its own right. He lectured in a French accent that five decades in Boston never wore away, and which had therefore, one must assume, been retained on purpose. He told his students that business, properly practised, was the most demanding of callings, and assigned homework no other course required: study a company, certainly, but study above all the people who ran it, their habits, their correspondence, the way they treated subordinates. Sayings of his circulated among alumni for decades afterwards, polished by repetition: that a creative man merely has ideas while a resourceful man makes them practical; that somewhere, at every moment, somebody is building the product that will make yours obsolete. Whether every attributed saying is word for word his own matters less than the conviction generations of students carried away: the person, not the plan, was the thing to study.
He was, by the testimony of nearly everyone who worked with him, a demanding and occasionally imperious presence, given to long silences in board meetings and to visiting the companies ARDC had backed with a frequency that startled their founders, who were used to bankers who checked in once a quarter if at all. He thought of himself not as a passive supplier of capital but as an active partner in building the company, sitting on boards, recruiting executives, intervening in operational decisions when he judged intervention necessary. This made him, in the vocabulary later generations of venture capitalists would adopt, one of the first practitioners of what came to be called value-added investing, though Doriot himself would very probably have found the phrase both ugly and slightly beside the point: of course you helped the companies you had staked your reputation on, what other kind of investor would you want to be.
He distrusted cleverness for its own sake and had a well-earned reputation for being unmoved by a dazzling pitch delivered by a founder who could not answer a plain question plainly, a reputation that shortened a good many meetings. Students who took his course decades later still recalled being asked, not what the product did, but whether they would want to spend the next ten years of their life in a small room with the person proposing to build it, a question that has nothing to do with market size and everything to do with the discipline Doriot considered the true subject of his teaching.
ARDC did not move quickly, and it did not make very many bets by the standards that Silicon Valley would later set. Over roughly its first fifteen years it invested in something on the order of two dozen companies, a pace a modern venture partner would find sedate. But among that small number of bets was one that, on its own, would justify the entire institution, and settle the question of whether Doriot's theory of patient, engineer-backing capital could produce spectacular returns rather than merely worthy ones.
Digital Equipment and the Proof
In 1957, two young engineers named Kenneth Olsen and Harlan Anderson, who had worked together at MIT's Lincoln Laboratory on early digital computing projects, came to Doriot with an idea for a company that would build computers, though they were careful, in their pitch, not to use the word computer too loudly. IBM so dominated the industry that any company proposing to build a computer risked dismissal as either delusional or doomed, and Olsen and Anderson, sensibly, framed their pitch around smaller, cheaper digital modules rather than announcing an intention to compete head to head with the largest computer company in the world. ARDC put in seventy thousand dollars, a modest sum even by the standards of the day, in exchange for a stake commonly recorded as roughly seventy percent of the new company, which they named Digital Equipment Corporation, DEC for short. The terms would strike a modern founder as confiscatory, and struck nobody at the time as anything but generous, since no other institution in America was offering unproven engineers terms of any kind at all.
Doriot's habitual involvement extended to this deal in ways that became part of the company's own founding lore. He is remembered as having pushed Olsen and Anderson to spend their early capital cautiously, to prove out a real, profitable product before expanding ambitiously, and to resist the temptation, common among engineers with a good idea and someone else's money, to chase every interesting technical direction at once. DEC built small computers priced and packaged so that individual laboratories, universities, and engineering departments could buy one without the elaborate procurement an IBM mainframe required. The company was profitable within its first year, an almost eccentric distinction among venture-backed firms, and its PDP-8, introduced in 1965 at a price commonly cited as around eighteen thousand dollars, became the machine that turned the minicomputer from a curiosity into a category, cheap enough for a single laboratory to buy in the way it might buy an oscilloscope. This turned out to be exactly the right bet on how computing would actually spread through American institutions, and DEC grew, through the 1960s, into the second-largest computer company in the country.
When DEC went public in August 1966, ARDC's original seventy-thousand-dollar stake was abruptly worth tens of millions of dollars on paper, and it kept growing; by the early 1970s the holding was commonly valued at somewhere between three hundred and fifty and four hundred million dollars, a return on the original investment running to several thousand times what had been put in. A single deal, done by a single institution working to principles Doriot had spent years articulating to his students, had generated returns no conventional theory of prudent investing would have predicted, and no conventional financial institution, bound by fiduciary caution and a horror of concentrated risk, would knowingly have pursued. The episode earns its place near the centre of Srinivasan's Americana, whose tour of American speculation runs from the Mayflower's London backers to the modern technology industry; set in that line, the DEC return is the moment the whaling merchant's arithmetic re-entered the record in modern corporate form.
The popular memory of Doriot as the founder of Silicon Valley flattens a more interesting and more complicated truth. The first complication is the least comfortable: strip the DEC position out of ARDC's quarter century of investing and the record that remains, by most later reconstructions, shrinks to something an ordinary trust department could have matched without leaving the safety of listed securities. One investment carried the institution. That is not a blemish on the story; in a sense it is the story, the first full-scale demonstration of the lopsided arithmetic that would come to govern the entire industry, in which the case for the whole practice rests, again and again, on one holding in twenty. But it means the DEC return proved the power law before anyone had named it, rather than proving that Doriot's firm could produce winners on demand. ARDC, for all its historical importance, also remained a comparatively conservative, East Coast institution, closer in temperament to a patient merchant bank than to what venture capital would become a generation later on the other side of the continent. It invested slowly. It held its positions for a long time, often a decade or more, waiting out the slow maturation of a technology company rather than pushing for a quick exit. Doriot himself distrusted speculative excess and had little patience for founders who seemed more interested in a quick payday than in building something durable; he wanted, those who knew him agree, to build institutions that would last, not merely to generate a return and move on to the next deal. This patience served ARDC well with DEC, but it meant the firm never achieved the volume, the speed, or the sheer number of swings that would come to define venture capital as it developed in California. ARDC made, across its entire working life, a total number of investments that a single successful Sand Hill Road partnership would later make in five or six years of ordinary business. And its public structure, the feature Doriot was proudest of, proved its undoing: a publicly traded investment company, watched by regulators and shareholders, could not let its own young staff take personal stakes in the companies they nursed, which meant the men who found and tended the winners drew salaries while the winners made fortunes for others. The ablest of them noticed. Bill Elfers, for years Doriot's senior lieutenant, left in 1965 to found Greylock as a private partnership, and the exodus he began settled, in practice, the question of what legal form the industry would take. The limited partnership did not win because it was elegant. It won because it let the people doing the picking keep a share of what they picked.
ARDC itself did not last as an independent institution long enough to enjoy its own vindication for very long. In 1972, the same year Kleiner Perkins and Sequoia were founding themselves on the other side of the country, ARDC was acquired by the conglomerate Textron, its publicly traded shares absorbed into a much larger and much less specialised industrial holding company, a quiet, almost bureaucratic ending for the institution that had started the whole trade. There is something fitting, if slightly melancholy, in the timing: the year the East Coast pioneer was folded into a conglomerate was the very year its Californian successors were being born a few miles from Stanford, as though the baton had to be dropped on one coast before it could properly be picked up on the other.
Doriot's model did not stay on the East Coast. It travelled west, into the hands of men with a very different temperament from Doriot's own, who would strip out the patience and the institutional caution and keep the underlying logic, running it faster, looser, and on a far larger scale than the Harvard professor with the Napoleonic bearing would have found comfortable to watch.
West to Stanford
The man most responsible for building the physical and institutional soil into which that transplanted model would grow was not, on the face of it, a financier at all. Frederick Terman was a professor of electrical engineering at Stanford University, later its dean of engineering and then its provost, and he is now remembered, with only mild exaggeration, as the father of Silicon Valley, a title he earned not by inventing anything himself but by deciding, with unusual persistence, that his students should build their futures in California rather than shipping themselves east the moment they graduated, the well-worn path of the previous generation.
Terman came by his convictions by inheritance and by training. He was the son of Lewis Terman, the Stanford psychologist who developed the Stanford-Binet intelligence test, and he took his own doctorate at the Massachusetts Institute of Technology under Vannevar Bush, the engineer and administrator who would later run the American government's wartime scientific research effort. Terman returned to Stanford to teach radio engineering, and from the late 1930s onward he made it something close to a personal mission to keep his best students from leaving. In 1939 he lent encouragement, practical introductions, and a small amount of money to two of his former students, William Hewlett and David Packard, who set about building audio oscillators in a rented garage behind a house on Addison Avenue in Palo Alto, with starting capital commonly put at five hundred and thirty-eight dollars and a company name whose word order, Hewlett-Packard rather than Packard-Hewlett, was settled by a coin toss. That garage, preserved today as a minor shrine and generally credited as the literal birthplace of Silicon Valley, produced a company, Hewlett-Packard, that became the region's founding legend and its first proof that a serious technology company could be built and headquartered in Northern California rather than in Boston or New York.
Decades of retelling have polished the garage into something close to a nativity scene, complete with a commemorative plaque, the kind of consecration that ought to make anyone who has read a little history suspicious of how neatly it fits the story everyone already wanted to tell. Hewlett and Packard's actual first product, an audio oscillator used to test sound equipment, sold in modest numbers to modest customers; the most famous early order, commonly given as eight oscillators, came from Walt Disney Studios, which used them in developing the sound system for the film Fantasia. It was not a dramatic breakthrough. It was a well-made, unglamorous instrument sold by two disciplined engineers who paid attention to their customers, a fairly good description of most successful early companies and a poor description of the myths that later grew around them. Terman's real contribution was not the modest loan or the introductions but permission: the standing example, from a respected faculty member, that staying in Palo Alto to build something of one's own was a sensible career, not an eccentric detour from the sensible path east.
Terman's own wartime experience mirrored Doriot's in an interesting way, though it pointed toward a different institutional conclusion. He spent the Second World War directing the Radio Research Laboratory at Harvard, working on radar countermeasures for the Allied war effort, and the experience left him convinced that the three-way collaboration he had seen work so effectively there, government funding, university research, and industrial application feeding into one another continuously, could and should be recreated permanently in peacetime. Where Doriot took from the war a conviction about funding unproven people, Terman took from it a conviction about building unproven ecosystems. As dean of engineering from 1945 and then as Stanford's provost from 1955, Terman set about turning that conviction into physical reality. He built Stanford's engineering departments into a genuine research power, chasing federal defence research funding aggressively and using it to recruit and retain top faculty. And he created the Stanford Industrial Park, opened in 1951, leasing university-owned land on long terms to technology companies willing to locate near the campus, so that firms like Hewlett-Packard and later Lockheed's missile division could operate within walking distance of the laboratories that were training the engineers those firms wanted to hire, an arrangement that suited everybody except the East Coast.
The Industrial Park filled up steadily through the 1950s with tenants that read, in retrospect, like a roll call of the coming industry: Varian Associates, founded by Stanford-trained physicists working on microwave tubes, was among the very first arrivals, and Lockheed's missile and space division eventually became the single largest private employer in the valley, a fact modern accounts tend to underplay in their eagerness to get to the part about garages and venture capitalists. Silicon Valley's earliest and, for a long stretch, largest paymaster was the United States government, spending defence dollars on radar, missile guidance, and reconnaissance technology, and the venture-backed companies that later defined the region's popular image were built, in no small part, on engineering talent and contracting relationships the Cold War had already assembled and paid to train. This is not a criticism of the venture model so much as a useful corrective to the story it likes to tell about itself, in which brilliant individuals and daring investors conjure an industry out of nothing but nerve and a garage. The nerve and the garage were real. So was the Pentagon.
None of it looked remotely like a revolution: the men conducting it wore sports jackets, drove station wagons, and were indistinguishable at forty paces from insurance agents. What the region still lacked, in 1955, was the specific financial technology that could take that ambient talent and turn it, systematically, into new companies rather than merely into new products developed inside existing ones. That piece of the puzzle arrived, in a characteristically indirect way, in the person of a brilliant, difficult, and increasingly erratic physicist who came home to California to found a company that would fail on its own terms and succeed, spectacularly and almost by accident, on every other term that mattered.
The Traitorous Eight
William Shockley had co-invented the transistor at Bell Laboratories in 1947, alongside John Bardeen and Walter Brattain, a piece of work for which the three men shared the Nobel Prize in Physics in 1956. Shockley was, by any measure, one of the most important physicists of his generation, and he was also, by nearly every account left by the people who worked closely with him, an extremely difficult man to work for: brilliant, competitive, deeply suspicious of the loyalty and competence of the people around him, and prone to management methods that would raise eyebrows in almost any era. The episode everyone who worked there seems to have remembered involved a secretary who cut her hand at a laboratory door; Shockley convinced himself the injury was sabotage and ordered members of his staff onto a lie detector, over an incident most employers would have settled with a plaster and an apology. In 1955, with his Nobel recognition still fresh, Shockley returned to the Northern California of his boyhood and founded Shockley Semiconductor Laboratory in Mountain View, financed by the chemist and instrument maker Arnold Beckman as an arm of Beckman Instruments, on ground Terman had spent the previous decade preparing.
Shockley's one great and lasting achievement in founding the company was recruitment. On the strength of his reputation alone, he assembled one of the most talented groups of young scientists and engineers ever gathered under one roof in American industry, among them Robert Noyce, a physicist from Iowa with a doctorate from MIT; Gordon Moore, a chemist with a similarly sharp analytical mind; and Jean Hoerni, a Swiss-trained physicist who would shortly make one of the decisive technical breakthroughs of the whole semiconductor era. What Shockley could not do, it turned out, was manage the people he had recruited. He grew suspicious of his own staff's competence and loyalty in ways that had little grounding in reality, insisted on pursuing technical directions that his own engineers considered scientifically unpromising, and generally created an atmosphere that made the laboratory, for all its concentration of talent, an almost impossible place to get anything shipped. As a study in how to lose eight geniuses in a single building, the laboratory has never been bettered.
On 18 September 1957, eight of Shockley's most senior scientists and engineers resigned together, having concluded, after failed appeals to Beckman to remove or sideline the man whose name was on the door, that nothing lasting could be built under his direction. Shockley, incensed, referred to the group as the Traitorous Eight, a label meant as an insult that history has kept, mostly because nobody involved seems to have minded it very much in retrospect. The eight were Robert Noyce, Gordon Moore, Jean Hoerni, Eugene Kleiner, Jay Last, Julius Blank, Victor Grinich, and Sheldon Roberts, and their departure poses, in miniature, the central practical problem any theory of venture financing has to solve: eight talented people had decided to leave together and start something new, with no employer willing to take them as a group and no capital of their own to start a semiconductor company.
The letter describing their situation and their credentials landed, through a family connection of Eugene Kleiner's, on the desk of a young investment banker in New York named Arthur Rock, who worked at the small firm Hayden, Stone & Co. Rock tried, in the conventional way, to place the group with an established electronics firm willing to fund and absorb them. No company wanted to hire eight people at once on the strength of a letter and a reputation, however impressive the reputation. What Rock did next is, on any fair reading, the single most consequential improvisation in the whole story of American venture capital. Rather than continuing to look for a company willing to hire the team, he began looking for a company willing to fund the team to build their own. After some thirty-odd established firms had declined, by the count usually given, he found Sherman Fairchild, an inventor and financier who happened to be the largest individual shareholder in IBM, his father having helped found it, and whose Fairchild Camera and Instrument Company agreed to stake the group to a semiconductor operation of their own in Mountain View: a commitment commonly cited at just under one and a half million dollars, against an option to buy the founders out later for three million. Each of the eight put up five hundred dollars of his own money for his stake, and in a story told often enough to have earned its keep, the eight scientists and their two bankers from Hayden, Stone sealed the arrangement by signing their names on a set of dollar bills, the closest thing to a founding contract the deal ever had. No court was ever asked for its view of the instrument, which is probably as well.
The distinction between these two arrangements, hiring a team versus funding a team to build a company of its own, sounds almost too small to matter. Doriot's ARDC had funded unproven companies, but companies that already existed, with a founder at the helm and a plan drawn up: the traditional posture of an investor evaluating a proposition brought to him. What Rock had done with Fairchild Semiconductor was subtly different and, as it turned out, far more generative: he had gone looking for money on behalf of a team of people rather than a company or a plan, on the theory that the team itself, its talent and its cohesion, was the asset worth financing, and that the company and the plan would follow from the team rather than the other way around. This was Doriot's own emphasis on people over ideas, but pushed a full step further, toward financing the raw human talent directly and letting the corporate form assemble itself around it afterward. It is the seed of the modern venture model, and it happened, characteristically for this story, almost by accident, as the solution to a practical recruiting problem rather than as the deliberate execution of a theory.
Fairchild Semiconductor, once funded, justified the bet with a run of technical achievement that has few rivals in the industrial history of the century. In 1959, Jean Hoerni worked out what came to be called the planar process, a method of manufacturing transistors that protected the delicate junction beneath a layer of oxide, making transistors dramatically more reliable and, crucially, suitable for mass production rather than the fragile, hand-finished devices the industry had relied on before. Robert Noyce, building directly on Hoerni's planar process, saw that the same flat, protected structure could be used to connect several transistors and other components together on a single piece of silicon rather than wiring discrete components together by hand, and in doing so arrived, within months of Jack Kilby at Texas Instruments, at the integrated circuit. Kilby had reached a version of the same idea first, in circumstances that are themselves a small parable about how invention happens: too new at Texas Instruments to have earned any summer holiday, he was left almost alone in the laboratory while senior colleagues took their vacations, and he used the quiet to work through the tyranny of numbers problem then vexing the whole industry, the fact that as circuits grew more complex, the sheer quantity of components and the wiring between them threatened to make further advance practically unmanageable. His insight was that a single block of semiconductor material could furnish every part of a circuit at once, dissolving the wiring problem rather than solving it piece by piece, and in September 1958 he demonstrated a sliver of germanium carrying a complete oscillator. But his device relied on fragile hand-soldered gold wires and could not readily be manufactured; Noyce's planar version, arrived at independently months later and built on Hoerni's process, was the one that could be built at scale, and it is Noyce's approach that became the ancestor of every integrated circuit made since. Fairchild and Texas Instruments spent the following decade disputing the patent rights before eventually agreeing to share credit and cross-license the underlying technology, a resolution that satisfied the lawyers considerably more than the engineers, and one that history has, on the whole, ratified: Kilby received the Nobel Prize in Physics for the invention in the year 2000, by which time Noyce had been dead for a decade, the Nobel committee's rules against posthumous awards having quietly settled a priority dispute that patent lawyers had spent years failing to resolve. When the Apollo program needed a guidance computer reliable and light enough to fly to the moon, it was Fairchild's integrated circuits that went into the Apollo Guidance Computer, an arrangement that made the federal government, for several years, the young chip industry's most important customer, and one for whom cost per unit was not the pressing question.
A further piece of Fairchild's legacy shows how directly the company's own commercial experience fed back into the industry's understanding of itself. In 1965, Gordon Moore, studying the pace at which Fairchild and its competitors had been packing ever more components onto a single chip, wrote a short article for the trade journal Electronics observing that the number was roughly doubling every year, and predicting, with a confidence that turned out to be remarkably durable, that the trend would continue for at least another decade. The observation was later canonised as Moore's Law, a considerable rank for a magazine article to reach, and though it describes a regularity in manufacturing rather than a law of physics, it became one of the load-bearing assumptions of the entire venture industry: an investor backing a semiconductor or computing company in 1970 could reasonably assume that the underlying hardware would keep getting cheaper and more capable on a predictable schedule, which made a certain kind of long-horizon technology bet far less speculative than it would otherwise have looked. Moore's observation, that is, quietly underwrote a great deal of the optimism that later financed the industry he had helped found.
As a technical enterprise, then, Fairchild Semiconductor was a triumph almost without qualification. As a corporate home for the people who had built the triumph, it proved to be something closer to a very expensive lesson in what happens when a parent company keeps the profits and pays its most valuable engineers a salary rather than a meaningful ownership stake. In 1959, with the division suddenly and visibly valuable, Fairchild Camera and Instrument exercised its option and bought the founders out for the agreed three million dollars, which came to something like three hundred thousand dollars a man: real money for a salaried physicist, and a fixed, final sum that severed the eight from everything the division created afterwards. From that point on the parent company, run from New York, took the semiconductor earnings and redirected a great deal of them east, into its other businesses, while the scientists and engineers who had actually built the technology watched the value they went on creating accrue to shareholders they had never met. It did not take a genius to notice the pattern, though it happened, fittingly, that the people who noticed it were geniuses: if you wanted to actually own what you built, you needed to leave and start your own company, the way the original eight had once left Shockley. Fairchild's engineers absorbed the lesson thoroughly, and over the following decade they put it into practice with a fertility that made the company the seedbed of the entire regional industry. Noyce and Moore left in 1968 to found Intel, taking with them a young Hungarian-born engineer named Andrew Grove. Jerry Sanders left to found Advanced Micro Devices. Charlie Sporck departed to run National Semiconductor. Eugene Kleiner, one of the original Traitorous Eight, left finance altogether in a sense, or rather moved fully into it, co-founding the venture capital partnership Kleiner Perkins. These descendant firms became known, with a nice botanical wit, as the Fairchildren, and the pattern they established, brilliant engineers leaving a company that had failed to reward them properly in order to found a new one that would, repeated itself so often across the region over the following decades that it stopped being a scandal and became, essentially, the regional business model. A story from a semiconductor industry gathering in Sunnyvale in 1969, repeated ever since because it is too good to leave out, holds that of the several hundred men in the room, fewer than two dozen had never worked for Fairchild.
The whole story of the Traitorous Eight usually reads as an uncomplicated triumph of talent over a difficult boss, which understates how badly Shockley's own trajectory ended. His semiconductor laboratory never shipped a commercially successful product and was eventually folded into another company, leaving him without the fortune his former employees went on to accumulate. His later years at Stanford, where he held a professorship, were consumed increasingly by public advocacy of eugenicist theories linking race and intelligence, a campaign that alienated most of his remaining scientific colleagues and cost him much of the standing his transistor work had earned. It is an uncomfortable footnote to a history of optimism, and it belongs anyway, because Shockley's decline is a reminder that the same instinct for backing an unproven, brilliant individual on the strength of reputation alone, the instinct this whole industry would come to run on, carries no guarantee about what that individual does with the rest of his life once the important discovery is behind him.
Arthur Rock's Method
Arthur Rock was not an engineer and never claimed to be one. He had trained in finance in New York, at Harvard Business School and then at Hayden, Stone, and his contribution to the history of technology was never a piece of hardware or a line of code. It was a method, and the method, once he had worked it out, was almost defiantly simple to state, however hard it proved to execute with any consistency. Rock believed that the single most important question to ask about a proposed new company was not what the product was, nor even how large the market for it might be, but who was going to build it. His conviction, refined out of the Fairchild experience and applied with increasing confidence over the following decade, was that a truly strong team would, given time and money, eventually find its way to a good business even if its first idea proved wrong, while a weak or fractious team could take a brilliant idea and run it into the ground. This is recognisably Doriot's own emphasis, character and discipline over the cleverness of any particular plan, arriving independently at the same conclusion from a very different professional path, which is itself a small piece of evidence that the two men had stumbled on something true rather than merely something fashionable.
Rock did not stay in New York to practise the method by correspondence. In 1961 he moved to San Francisco and formed a partnership with Tommy Davis, a lawyer turned investor who had been backing young technology companies for a land and cattle concern, and Davis & Rock became the working prototype of the structure the whole industry later adopted: a private limited partnership, raised from a small circle of wealthy individuals, several of the Fairchild founders among them, putting their buyout money back into the system that had made it, with the two general partners taking a share of the profits rather than a fee for advice. The fund, commonly put at around five million dollars, ran for seven years and returned its investors many times their money, by some accounts close to thirtyfold, carried above all by Scientific Data Systems, a computer maker Rock backed early and Xerox eventually bought, in 1969, for stock commonly valued at over nine hundred million dollars. The vehicle mattered as much as the return. Doriot had built a public company and spent twenty years apologising to its shareholders; Rock built a private partnership and answered to a handful of investors who had agreed, in writing and in advance, to exactly the kind of wager he intended to make.
Rock's second defining deal put his principle to its most famous test. In 1968, Robert Noyce and Gordon Moore decided to leave Fairchild and start a new company, which would come to be called Intel, to pursue what they believed was a much larger opportunity in semiconductor memory than their parent company was capable of chasing. They approached Rock, and the financing that resulted has become one of the most frequently repeated stories in the folklore of American business, not because the details are especially dramatic but because the speed of the thing is itself the point. Rock raised the money for Intel, two and a half million dollars of it by the figure usually given, in a single afternoon according to the legend and within a day or two by the soberest accounts, on the strength of Noyce and Moore's names and a proposal that ran to a page or so of studiedly vague typescript, a length Rock, who priced words the way a telegraph office would, appears to have found ample. Investors who had never met either man, and who were given essentially no conventional due diligence to examine, committed money anyway, because Noyce and Moore's reputations, built over the previous decade at Fairchild, functioned as collateral in a way that no balance sheet could have. Notice what the episode actually demonstrates: not merely that Noyce and Moore were credible, which they certainly were, but that the entire apparatus of trust venture financing depends on, the willingness to commit real money to people rather than to audited numbers, had by 1968 travelled a very long way in a very short decade. What had been, in 1957, an improvised solution to a recruiting problem had become, by 1968, a working financial reflex, fast enough to close in an afternoon.
What Intel did with that money over the following two decades illustrates a truth about venture financing that gets lost when the story stops at the cheque itself: the money is only ever the opening move. Intel's third employee was a young Hungarian engineer named Andrew Grove, who had fled Budapest after the Soviet suppression of the 1956 uprising, arrived in New York barely speaking English, and worked his way through a chemical engineering degree at City College before taking a doctorate at Berkeley. Intel began, unglamorously, as a memory chip company, and for its first years memory was the business. In 1971, an engineer named Ted Hoff, working on a contract to build calculator chips for the Japanese firm Busicom, realised that rather than designing a separate custom chip for every calculator model, a single general-purpose chip could be built to run different instructions depending on the software fed into it. The result, the Intel 4004, was the first true microprocessor, and for years Intel treated it as a curiosity alongside the memory business that still generated most of the revenue. That changed only when it had to. Through the first half of the 1980s, Japanese semiconductor manufacturers drove the price of memory chips down to levels Intel could no longer profitably match, and Grove, by then president of the company, and Moore, still chairman, faced a decision that Grove later described, in a phrase that entered the standard vocabulary of American management and has been applied since to decisions of very much smaller consequence, as a strategic inflection point: abandon the business the company had been founded to pursue and bet its future entirely on the microprocessor instead. They did it, memory manufacturing was wound down, and the microprocessor became the product on which Intel built two decades of near total dominance of the personal computer industry, underwritten all the while by the compounding arithmetic of Moore's own law.
Rock's later and, in purely financial terms, even more spectacular wager came a decade further on, when he agreed to back two young men working out of a garage in Los Altos, the setting by then practically obligatory, on a personal computer that most established electronics firms had shown no interest in. Rock joined the board of the company, Apple, and though he was by his own later account initially unconvinced that the two founders, Steve Jobs and Steve Wozniak, quite fit his template of the disciplined, credentialed engineer, he backed them anyway, with a personal investment commonly reported at a little under sixty thousand dollars, and the return, when Apple went public in December 1980 in the largest American stock offering since Ford's a generation earlier, made his earlier fortunes look almost modest by comparison. There is a nice, slightly uncomfortable detail here for anyone tempted to romanticise the method too tidily: Rock's most celebrated late bet was on a pair of founders who did not resemble the disciplined, buttoned-down engineers his own theory of investing was built around. The method worked even when its own practitioner was not entirely sure it applied.
Arthur Rock is widely credited, more than any other individual, with popularising the term venture capital in something close to its modern sense, and whether or not he coined the specific phrase, a point historians of the industry dispute, he is the figure most responsible for turning what Doriot had institutionalised on the East Coast and what he himself had improvised with Fairchild into a repeatable working method. This method did not remain the private practice of one unusually perceptive banker for very long. Within a few years of Rock's biggest triumphs, it had been codified into a standard financial architecture, with its own legal form, its own compensation structure, and its own geographic address.
Sand Hill Road
In 1972, two venture capital partnerships were founded within a short distance of one another, and of Stanford University, on a road in Menlo Park that would eventually lend its name to the entire industry: Sand Hill Road. The name records a hill of sand, and sand is, near enough, what silicon is before anyone takes it seriously. Kleiner Perkins was founded by Eugene Kleiner, the Fairchild alumnus and one of the original Traitorous Eight, together with Thomas Perkins, an engineer and executive who had run Hewlett-Packard's computer division; their first fund, commonly cited at about eight million dollars, was at the time among the largest pools ever assembled for the purpose. Sequoia Capital was founded the same year by Don Valentine, a veteran of the semiconductor trade who had run sales and marketing at Fairchild and then at National Semiconductor, a salesman's education that shaped everything he later believed about investing. Their arrival marks the moment at which venture capital stopped being the improvised practice of a handful of unusually gifted individuals, Doriot at ARDC, Rock and Davis in San Francisco, and became a proper industry, with a standard institutional shape that thousands of firms would eventually adopt with only minor variations.
Valentine's investing philosophy complicates the tidy doctrine, running from Doriot to Rock, that people matter more than plans. Where Rock had made his name on the conviction that a strong team would eventually find its way to a good market, Valentine argued, with equal conviction and less patience for the counterargument, that the size and shape of the market mattered more than the charm of the team pursuing it: a mediocre team in a huge, growing market would generally outperform a brilliant team chasing a market too small to matter, because a large market forgives a company's early mistakes in a way a small one never will. The industry came to shorthand the dispute as a question of whether one backed the jockey or the horse, and Valentine's answer, in its various tellings, was that he preferred to back the racetrack. His own record armed the argument well: it was Valentine who backed Atari, and Valentine who backed Apple after the young Jobs, an unwashed veteran of Atari's night shift, was passed along to him, a founder no assessment of polish would have flagged and a market, personal computing, whose eventual size forgave everything else. The two positions are not, in the end, as opposed as they sound in a pitch meeting, since the best investors on Sand Hill Road came, over time, to want both a strong team and a large market and grew uneasy backing a deal that offered only one. But the argument between them, Rock's emphasis on character against Valentine's emphasis on arithmetic, has run underneath the entire industry ever since, an unresolved and probably unresolvable tension between two things that both turn out to matter and neither of which, on its own, reliably predicts which nineteen bets will fail and which one will not.
That shape is the limited partnership, the piece of the whole story that actually does the mathematical work, and venture capital's own literature about itself tends to bury the mechanism under a fog of self-congratulation. In a venture limited partnership, the money itself comes overwhelmingly from large institutional investors, pension funds, university endowments, insurance companies, and latterly sovereign wealth funds, who commit capital as limited partners for a fund with a life of roughly ten years. The venture firm's own partners, the general partners, invest that committed capital into a portfolio of young companies over the fund's early years, manage and support those companies as they grow, and eventually return the proceeds to the limited partners at the end of the fund's life. In exchange for running the fund, the general partners typically collect an annual management fee, conventionally around two percent of committed capital, to cover the cost of actually operating the firm, and, more importantly, a share of the eventual profits, conventionally around a fifth, an arrangement generally called carried interest or simply carry.
The structure needed one further ingredient before it could grow to any size, and the ingredient came from Washington rather than from California. Through most of the 1970s, the great American pension funds were effectively barred from venture investing by the prudent man provisions of the 1974 pension law known as ERISA, which trustees read, reasonably enough, as forbidding retirement money from anything so likely to fail. In 1979 the Department of Labor clarified the rule: prudence attached to the portfolio as a whole, not to each investment within it. That change of a few sentences redirected billions. Annual commitments to venture funds, measured in the low hundreds of millions through the lean middle of the decade, rose within a few years into the billions, and the small partnerships of Sand Hill Road became stewards of the retirement savings of schoolteachers and firefighters, either the fulfilment of Doriot's dream of democratised venture investing or a quiet irony at its expense, depending on the day. A cut in the capital gains tax the year before is commonly credited with helping the same flood along.
This structure fits the underlying mathematics of venture investing with a precision that is easy to overlook because the structure has become so familiar it no longer looks designed. Venture returns follow what statisticians call a power law. This distribution differs in kind, not merely in degree, from the one that governs a portfolio of investment-grade bonds or a diversified basket of public equities, and it demands from the people managing it a discipline different in kind as well.
The phrase power law tends to glide past a reader without leaving much of an impression. Take a fund of a hundred million dollars that invests roughly five million each into twenty companies. On the arithmetic that governs most of this industry's history, perhaps eight or nine of those twenty will fail outright, returning nothing at all, which is at least simple to account for. Another seven or eight will limp along, get acquired for a modest sum, or return something close to the original stake, neither a loss nor a triumph. Two or three will do respectably well, returning three or four times the money put in, the kind of result that would make a private equity investor happy but would, on its own, leave a venture fund of this size roughly flat once fees are accounted for. The fund's actual fortune rests on whichever single company, out of the twenty, turns out to be the one in a hundred: the business that grows into a household name and returns not three or four times its five million dollars but three or four hundred times it, a single position worth more than the other nineteen combined many times over. Take that company out of the portfolio and the fund is a mediocre or poor investment. Leave it in and the fund is, by the standards of the entire financial industry, an exceptional one. This is why a venture investor's working life is organised so completely around the search for that single company rather than around the more familiar discipline of avoiding bad ones: avoiding bad ones barely changes the outcome, and chasing the very good ones too cautiously, out of a reasonable-sounding desire to protect capital, is the most reliable way to miss the one bet that would have justified the whole fund.
The practical consequence is that the craft of venture investing is organised not around avoiding failure, the assumed and budgeted default for most of any portfolio, but around securing access to the rare company that will matter enormously, because access, not skill at avoiding losers, is what determines a fund's returns. This is why venture capitalists will often compete for an allocation in a company they believe in with a ferocity, and sometimes a nakedness, that a purely rational bystander might find undignified: bidding up the price, softening the terms they would normally insist upon, courting a founder who has other, equally eager suitors, all in pursuit not of a bargain but of the mere right to participate. This is also the most useful single point of contrast with the older and larger trade of private equity, with which venture capital is often lazily confused. A private equity firm typically buys whole, established, cash-generating companies, using a substantial amount of borrowed money secured against the target's own assets and future earnings, and makes its return by improving operations, cutting costs, and eventually selling at a higher multiple than it paid. A venture capital firm buys small minority stakes in unproven companies that generally have no meaningful assets to borrow against and often no revenue at all, and its entire investment thesis rests on nothing more solid than the judgment that this particular team, with this particular idea, at this particular moment, might be one of the rare ones. Private equity negotiates hard on price because the underlying asset's value can, at least in principle, be assessed with some rigour. Venture capital often does not negotiate hard on price at all, because the whole point of a great opportunity is that its true value, if the company turns out to be one of the rare winners, will dwarf whatever premium was paid to get in. The resulting term sheet, a few pages of provisional arithmetic, is nonetheless signed with the solemnity of an armistice.
By the middle of the 1970s, this newly formalised industry had begun to prove that its method could travel well beyond the semiconductor and computing businesses that had given it birth. In 1976, a restless twenty-eight-year-old named Robert Swanson, lately of Kleiner Perkins and out of a job, telephoned the biochemist Herbert Boyer, who had pioneered with Stanley Cohen the technique of splicing genes between organisms, and asked for ten minutes of his time. By the account both men enjoyed telling afterwards, the ten minutes stretched to three hours over beer and ended with each man putting up five hundred dollars to found a company. Kleiner Perkins then backed them with an initial sum commonly cited at a hundred thousand dollars for a quarter of the venture, one of the more improbable bets of the decade: a scientist who had never run a business and a businessman in his twenties who could not do the science, proposing to manufacture human proteins inside bacteria. The resulting company, Genentech, extended the venture model for the first time out of electronics and into biology, an entirely different scientific domain, with a different regulatory environment, a different timeline to any product, and a different set of technical risks. When Genentech went public in October 1980, before it had a single product on the market, the shares were priced at thirty-five dollars and touched, by most accounts, nearly ninety within the first hour of trading, one of the fastest run-ups a new listing had ever produced, directed at a company whose entire saleable inventory consisted, at that point, of some promising bacteria; the market, it turned out, had learned to price the wager itself. Kleiner Perkins and Sequoia went on, over the following decades, to finance or refinance a very substantial share of what became the modern technology industry, including major roles in Apple and Google among a great many others, and Sand Hill Road settled into its position as the geographic centre of the whole trade, a few miles of suburban office parks that came to command an amount of global economic attention wildly disproportionate to their physical size.
What the Model Needs
Governments and investors around the world have spent the last several decades trying to reproduce the Sand Hill Road model somewhere else, and the failures are more instructive than the successes. Venture capital is not simply a pool of money looking for a home. It is an ecology, and an ecology's parts do not do much of anything on their own.
The most obvious ingredient is a research university, or a cluster of them, producing a continuous supply of technically excellent, ambitious young people, in the way Stanford did for Terman and MIT did for the engineers who built Digital Equipment. But a great university on its own produces excellent employees for other people's companies far more often than it produces founders, unless something else is also present: a cultural willingness, among truly talented people who could otherwise walk into secure, well-paid jobs at established firms, to instead take a large pay cut and an even larger risk on an unproven venture of their own. This willingness is not a universal human trait waiting patiently for the right incentive structure to call it forth. It is itself a local, learned disposition, a kind of regional optimism that has to be modelled by visible examples before it becomes a normal choice rather than an eccentric one, which is precisely why the Hewlett-Packard garage matters as a story quite apart from what the company achieved: it gave the next generation of Stanford engineers a concrete, walkable example of what leaving to start something looked like, worth more than any amount of abstract encouragement from a professor.
Money alone cannot manufacture that disposition, nor can it manufacture the second ingredient the ecology needs, which is a supply of lawyers, accountants, recruiters, and other professional services people willing to work partly on deferred terms, taking equity or reduced fees now against the chance of a much larger payment later, effectively betting alongside the founders rather than simply billing them by the hour regardless of outcome, which is what the profession was built to do. A region with brilliant engineers but only conventionally billed lawyers makes every young company pay full freight for its legal and accounting needs from year one, a burden a region with equity-friendly professional services does not impose, and the difference compounds, across hundreds of companies and many years, into a structural advantage.
Perhaps the hardest ingredient to manufacture on purpose, though, is a local tolerance for visible, public failure that does not permanently mark a founder or an early engineer as damaged goods. In much of the world, presiding over a company that goes bankrupt is a mark against a person's judgment and character, discussed in hushed and disapproving tones at family gatherings for a generation. In the version of Northern California that grew up around Terman's engineering park and Rock's method, a failed company was, and largely still is, treated as something closer to a credential: the person who ran it has learned, at somebody else's expense, precisely which mistakes not to make the second time, and is consequently a better bet, not a worse one, for the next venture. This is not a moral position so much as a practical one, forced into being by the underlying mathematics of the whole industry: if you are running a system that expects nine failures for every one success, you had better not treat failure as disqualifying, because doing so would eventually disqualify almost everyone worth backing, including, very often, the people who will go on to build the actual winners.
Where these elements, ambitious engineers willing to leave stable jobs, professionals willing to defer their fees, and a culture that does not brand failure as shameful, are absent, capital by itself accomplishes remarkably little, a lesson that has not discouraged a large number of governments and sovereign investors from trying anyway, usually by supplying the capital first and hoping the rest of the ecology assembles itself around it. The result has been a long and slightly comic procession of officially announced technology districts bearing some variant of the word Silicon, Silicon Fen around Cambridge, Silicon Roundabout in London, Silicon Wadi in Israel, Silicon Alley in New York, and dozens of others besides, most launched with a ribbon-cutting, a government minister, and a great deal of confident language about ecosystems, and rather fewer producing anything like the density and self-sustaining momentum of the original. This is not, on the whole, because the officials lacked intelligence or good intentions. It is because they were trying to purchase, with a line item in a budget, something that had taken Terman two decades of patient institution-building, several generations of engineers willing to bet their careers on each other, and no small amount of ordinary historical accident to produce the first time around. One or two of the transplants, Israel's technology sector being much the most complete success among them, did eventually take root, generally by getting most of the ecology right rather than merely the funding, which is itself the clearest evidence for the argument: it can be done elsewhere, but only by rebuilding the whole system, not by writing the cheque and waiting.
A related candour concerns the model's record even where the ecology exists, because the mythology of venture capital quietly generalises from a handful of triumphant funds in a handful of sectors to the industry as a whole, and the generalisation does not survive contact with the data. Studies of aggregate venture returns, among which the Kauffman Foundation's unsparing 2012 review of its own twenty years as an investor in funds remains the most quoted, have repeatedly found that the majority of venture funds fail, after fees, to beat what the same money would have earned in public equities; the average investor in the asset class has often paid handsomely for the privilege of trailing an index fund. The great returns are real, but they concentrate in a small fraction of firms, which concentrate in turn in a few sectors, software and biotechnology above all, and in a few square miles of California and Massachusetts, and access to those firms is largely closed to newcomers. Nor has the method conquered every industry it has marched into. The clean energy investment wave of the mid-2000s, when Sand Hill Road turned its confidence toward solar cells, batteries, and biofuels, ended with the industry losing, by common estimates, more than half of the billions it had committed, defeated by long development timelines, heavy capital requirements, and competitors subsidised by states rather than partnerships. The power law rewards the rare, fast, capital-light winner; it has so far shown little mercy to sectors where winning is slow and expensive. The model built the modern technology industry; taken outside its native habitat of sector as much as of geography, it has failed at least as often as the transplanted Silicons did.
France offers an instructive variant on the same lesson, since it has tried harder and spent more than most, without ever quite closing the gap. Successive French governments have poured public money into incubators, tax credits for early-stage investors, and state-backed venture funds, and Paris now has a perfectly respectable and growing technology sector to show for it. What it has never quite managed to build, at least not yet at anything like California's scale, is the third and hardest ingredient: a deep, unembarrassed cultural comfort with visible failure. A French engineer who leaves a secure position at a large, prestigious company to found a start-up that then collapses is still taking a bigger reputational risk than his Californian counterpart, in a society that has historically prized the prestige of the grandes écoles and the state bureaucracy they feed rather more than it prizes the scar tissue of an honourable defeat. None of this is a fixed national character, immovable by policy, since cultures do change, sometimes quickly, when enough visible examples accumulate to change what a young engineer considers a normal career. But it does mean that the single hardest ingredient to legislate into existence is also, on the evidence of every attempted transplant so far, the one that actually decides whether the transplant takes.
Exporting the Wager
Not every attempt to transplant the model failed, and the exceptions show what it actually takes to move a financial technology from one country to another rather than simply announcing that it has been moved. The clearest case is British, and it belongs to a man named Ronald Cohen, whose own biography reads like a smaller, later echo of Doriot's: a foreigner who arrived with almost nothing, absorbed an American institutional idea at close range, and then spent a career adapting it to a country that did not yet have a word for what he was proposing to do.
Cohen was born in Egypt in 1945 to a Jewish family that had lived in Cairo for generations, and his childhood there ended abruptly in 1957, in the aftermath of the Suez Crisis, when the family, along with much of Egypt's Jewish community, was forced to leave the country, arriving in England as refugees with very little beyond what they could carry. Cohen did not yet speak English. He learned it, thoroughly enough to win a scholarship to Oxford, where he read philosophy, politics, and economics, and thoroughly enough after that to complete his business education at Harvard Business School, arriving on the same campus, a generation on, where Doriot had spent his own career explaining to sceptical students why character mattered more than cleverness. It would be tidy, and probably too tidy, to claim Cohen sat in Doriot's own classroom, but he certainly absorbed, in the Boston of the late 1960s, the same institutional idea that Doriot had spent two decades trying to install in American finance: that a properly structured pool of patient capital, willing to back unproven people, could be a serious and repeatable way to build value rather than an eccentric sideline for wealthy hobbyists.
Britain in the early 1970s had essentially no equivalent institution. Its capital markets were built around established firms, its banks lent against collateral in the traditional way, and the whole idea of an investor putting money into a company with no assets, no revenue, and no security beyond the founders' own talent struck most of the City of London as a faintly American eccentricity, more suited to a country with a taste for gambling on the frontier than to a nation that had spent the previous century perfecting the art of the prudent, well-secured loan. In 1972, the same year Kleiner Perkins and Sequoia were founding themselves on Sand Hill Road, Cohen, then in his late twenties, and two Harvard Business School classmates founded the firm that became Apax Partners, one of the earliest venture capital firms in Britain, built deliberately on the American model he had studied at close range. It was not, at first, an easy sell. Cohen spent a good part of the 1970s and 1980s persuading British pension trustees, a profession not naturally disposed toward exuberance, that backing unproven entrepreneurs was a legitimate use of retirement savings rather than a breach of fiduciary duty, an argument that required roughly the same case Doriot had made to Boston's institutions a generation earlier, translated into a rather more risk-averse idiom. Apax went on to become one of the largest private equity and venture houses in Europe, and Cohen's standing as the father of British venture capital is not seriously disputed, a description that ages rather well given how few claimants to fatherhood in this industry survive close inspection.
What makes Cohen's story instructive, beyond the pleasing symmetry with Doriot, is what he did with the method once he had spent three decades proving it could work. From the early 2000s, Cohen turned his attention away from conventional venture and private equity returns and toward a harder question: could the same tools, patient capital, rigorous measurement of results, incentives aligned with actual outcomes rather than good intentions, be pointed at social problems that markets and governments had both struggled to solve. He founded Bridges Ventures to invest in businesses operating in Britain's most deprived areas, later chaired Big Society Capital, a British institution set up to fund social investment, and led the G8's Social Impact Investment Taskforce, work generally credited with helping popularise what is now called social impact investing, a phrase with exactly the kind of self-congratulatory ring that ought to make a careful reader narrow their eyes, but one covering a serious attempt to ask whether venture capital's central discipline, expecting most bets to fail while structuring the whole enterprise around finding and rewarding the ones that do not, could be pointed usefully at problems that had nothing to do with semiconductors or software. Whether that attempt has succeeded on its own ambitious terms is a separate and still open question. That a refugee from Cairo spent one career importing an American financial invention into Britain and a second trying to redirect its logic toward the common good is, at the very least, a reminder that the underlying idea, patient money structured to survive its own frequent failure in pursuit of a rare and disproportionate success, has proved far more portable than any single industry or any single country's claim upon it.
A Peculiar Species of Optimism
What kind of optimism is venture capital? The word gets used loosely enough elsewhere that venture capital slides easily into the same vague category as a sunny disposition or a general faith that things tend to work out. It is neither of those things. It is not the optimism of someone who looks at a long historical record of rising living standards and reasonably concludes that the trend will probably continue, which is a kind of optimism built on completed data. Nor is it the optimism of a philosopher who argues that human knowledge is, in principle, unbounded, and that any problem not forbidden by the laws of physics can eventually be solved given enough good explanations and enough time, which is optimism as a metaphysical claim about what is possible in general. Neither has ever had to raise a fund. Venture capital's optimism is neither a reading of the past nor a claim about the future in the abstract. It is a piece of engineering: a specific, deliberately constructed set of financial rules built to keep an institution functioning, keep it writing cheques, keep its partners showing up to meet the next unproven founder with an open mind, despite the near certainty, built into the design from the very first day of the fund, that most of what it does will fail completely.
The discipline this requires is the opposite of what the word discipline usually implies in finance. A disciplined bond trader avoids risk. A disciplined venture investor does not avoid risk at all; risk of total loss on any individual position is simply accepted as the ordinary cost of participating in the game. What venture discipline actually consists of is sizing: correctly judging how many losing bets a portfolio can absorb while still leaving enough capital and attention to recognise and back the rare winner when it appears, and then having the emotional and institutional fortitude to keep making new bets after the ninth, the twelfth, the seventeenth failure in a row, on the theory, borne out often enough to count as a working law rather than mere hope, that the eighteenth or the nineteenth might be the one that matters. This is optimism with its sleeves rolled up: not a feeling to be enjoyed but a discipline to be practised, on a schedule, with other people's money, under the scrutiny of institutional investors who expect the arithmetic to work out over a ten-year fund life even if it does not work out in any given year within it.
Leaving the account there, as though the industry's own preferred self-description were the whole truth, would be cowardice. Venture capital's optimism has curdled and been abused too often for that. The years from roughly 2015 to 2021 produced, across a significant slice of the industry, a version of the model that had drifted a long way from Doriot's insistence on character and discipline, rewarding a founder's talent for telling a compelling growth story over almost any evidence of a durable business underneath it, a pattern serious people within the industry criticised at the time, sometimes bitterly: money chasing narrative rather than substance, inflating valuations with only the thinnest connection to anything a company had built or earned. The collapse of several prominent, richly funded companies in the years that followed, WeWork chief among them, a landlord sustained by the unshakeable conviction that it was a technology company, valued by its largest backer at forty-seven billion dollars at the start of 2019 and unable by that autumn to complete a public offering at any price its bankers dared print, was not a betrayal of the venture model so much as a demonstration of what happens when its tolerance for visionary storytelling is allowed to operate without the counterweight Doriot always insisted on: a hard, unsentimental look at whether the people in front of you actually have the discipline to execute, not merely the charisma to raise the next round.
There is a second and, in some ways, more uncomfortable honesty required here, about who has historically been trusted with this kind of speculative faith and who has not. Arthur Rock's founding instinct, back the people rather than the plan, is, on its own terms, a sound piece of investing wisdom, and it has been vindicated often enough to deserve the respect it gets. But deciding who counts as a person worth backing is an act of pattern recognition, and pattern recognition, exercised by a small and homogeneous group of investors over several decades, has a well-documented tendency to reproduce itself: to see credibility most readily in people who look, sound, and carry themselves rather like the investor doing the recognising. Across the industry's history, that has disproportionately meant men, disproportionately meant graduates of a small handful of universities, and disproportionately meant people who arrived at the pitch meeting already fluent in the unstated social codes of Sand Hill Road, which are nowhere written down, that being what makes them useful. None of this invalidates the underlying mathematics of the power law, which is real and which does reward the rare, spectacular success wherever it is found. But it is a serious qualification of the industry's more romantic self-image as a pure meritocracy of ideas, and any account of venture capital's optimism that leaves this out has not told the whole story, only the flattering half of it.
The industry's own defenders have a ready answer to this. Venture capital, they will point out, is a business of extreme information scarcity: an investor deciding whether to back a twenty-six-year-old with no revenue, no customers, and no track record has almost nothing objective to go on, and in the absence of hard information human beings fall back on pattern recognition, which is, most of the time, a reasonably efficient way of deciding quickly under uncertainty. The trouble with this defence is that it is true and inadequate at the same time. Pattern recognition built on a narrow historical sample will keep confirming its own narrowness indefinitely unless something deliberately interrupts it, and for most of the industry's history nothing much did. The interruption, where it has come at all, has come from outside the traditional venture establishment: a new generation of investors and limited partners, often themselves excluded from the original pattern, deliberately widening the aperture, and a slowly accumulating body of evidence, uncomfortable for the industry's self-image but hard to argue with, that funds making a serious effort to look beyond the familiar profile have often done perfectly well by it. None of this resolves the tension neatly. It simply means the tension stays, whatever admiration the rest of the record has rightly earned.
One further, quieter cost belongs in the ledger before moving on, because it follows directly from the same arithmetic that makes the industry work. A system built to expect nine failures for every success is, definitionally, a system that produces an enormous quantity of failure, and behind each of those eighteen collapsed companies in a fund of twenty sits not an abstraction but a group of people, employees rather than founders, who took a job on the strength of somebody else's optimism and lost their income when the bet did not pay off. The venture investor absorbs the loss on a spreadsheet, spreads it across a portfolio, and moves on to the next meeting; the mathematics really do work out fine for her, that is the whole design. The employee, more often than not, does not have a portfolio to spread the loss across. This is not an argument against the model, which remains, on the evidence of a century, a remarkably effective way of financing genuine invention. It is a reminder that optimism distributed unevenly, so that its costs land disproportionately on people who did not choose to make the wager in the way the fund's partners did, is still optimism, but optimism practised at somebody else's expense as much as one's own, a fact the industry's vocabulary of shared risk tends, conveniently, to leave out.
Coda: Doriot's Lesson, Three Thousand Miles West
To build a durable institution on that arithmetic means doing something specific and unusual: converting a society's willingness to be wrong, again and again, in public, with real money, into a repeatable set of rules, a legal structure, a compensation scheme, and a professional culture robust enough to survive the failures without losing its nerve. Not many places have managed it.
Georges Doriot taught generations of Harvard students that the character and discipline of the people in front of you mattered more than the cleverness of whatever they happened to be proposing, and that lesson, formed in the patient, institutional, faintly aristocratic temperament of a Harvard Business School classroom in the 1950s, at an institution whose enthusiasm for risk has always been most pronounced when the risk belonged to somebody else, made its way three thousand miles west over the following two decades and arrived, transformed, in the far faster, far more informal, and occasionally far more reckless culture of Sand Hill Road, stripped of Doriot's own caution but keeping, at its core, his essential insight about people.
Nobody in this story, individually, invented venture capital's real historical achievement, which is not, in the end, any single company, not DEC, not Fairchild, not Intel, not Apple, not Genentech, however remarkable each of those companies was. The achievement is the demonstration, repeated often enough across three-quarters of a century now to count as established fact rather than a promising theory, that a society can build an institution whose explicit purpose is to keep functioning while being wrong most of the time, and can do so not out of resignation or indifference to the failures but in the deliberate, calculated service of occasionally being spectacularly, world-alteringly right.
It is tempting, and would be a little too neat, to end there, on the arithmetic vindicated. The whaling merchants of New Bedford ran the same arithmetic for the better part of a century and then watched the entire trade evaporate within a generation, not because anyone had misjudged the risk of any given voyage but because a cheaper way of lighting a lamp came up out of a hole drilled at Titusville, Pennsylvania, in 1859, and made the whole apparatus of ships, captains, and lays obsolete. There is no particular reason to think venture capital, for all its institutional sophistication, is exempt from the same fate: a financial technology built to fund one kind of ingenuity does not come with a lifetime guarantee against a different, cheaper way of doing the same job appearing from an unexpected direction. That is not a reason for gloom. It is, if anything, a rather fitting final illustration of the whole argument. An institution built entirely on the expectation of being wrong most of the time ought, by its own logic, to hold its own permanence a little lightly too.
Whatever comes to replace it, or however it continues to mutate, as it has mutated already from Doriot's patient East Coast merchant bank into Rock's fast Californian improvisation into Sand Hill Road's standing industry into Cohen's British import into whatever the industry is quietly becoming right now in rooms no history has yet visited, venture capital's real contribution to the history of enterprise will remain the same: it took a form of optimism that most human institutions treat as a private virtue, or a dangerous vice, and built, deliberately and at scale, a working method out of it. Not a feeling. A discipline, with a balance sheet, that has, against most people's first instinct about how finance ought to behave, actually worked, and worked precisely because it never once pretended that it would not, most of the time, be wrong.