The Code

by Margaret O'Mara

  Zschau might have been a rookie backbencher, but his tireless advocacy made an impression. “If he has done nothing else,” one GOP colleague observed, “he now has Members of Congress pronouncing it ‘Silicon’ instead of ‘Silicone’ Valley, and I think many of you will recognize that as quite an improvement.”34

  A new face joined the crowd too: Burt McMurtry. By the time Ronald Reagan entered the White House, McMurtry commanded one of the most valuable portfolios in the Valley, surging to more than $200 million by 1983. The Steiger Amendment victory had changed his mind about politics, so much so that he spent a term as the NVCA’s president and became one of the most plugged-in Valley financiers in 1980s Washington.

  Shortly after Reagan took office, McMurtry started shuttling groups of legislators and executive-branch officials out to the Valley, where he’d lead them on tours of his favorite portfolio companies. He was particularly fond of taking them to a start-up called Sun Microsystems, founded by three Stanford graduates to build a powerful mini-meets-micro hybrid called a workstation. Co-founder Scott McNealy—“such a showman,” remembered McMurtry—never failed to wow his audience. “Who told you that you could do this?” one Senator asked the entrepreneur in wonderment after he learned about Sun’s products. “How did you get permission?” McMurtry shook his head ruefully. “It was a perfect example of the disconnect between D.C. and Silicon Valley.”35

  Despite this, the Texan remained impressed by the political world. He hadn’t been a big fan of Reagan at first, but when he attended his first small gathering of business leaders at the White House, the president was charming, engaged, and “ran that meeting as effectively as any CEO I had ever seen.” With access to the White House and energetic advocates in Congress, venture guys like McMurtry and Morgenthaler and right-leaning CEOs like Young and Noyce won another capital gains tax cut, further tax credits for R&D, and more. Steve Jobs and Regis McKenna might be playing footsie with liberal Democrats, but the money men and the chipmakers were mostly sticking with the GOP, and it was reaping dividends.

  AMERICA’S MITI

  While outside observers (then and now) tend to think of Silicon Valley as an undifferentiated unit, and “the tech industry” as a singular noun, the Japanese challenge and the politics that spilled out from it made it clear that there were multiple Valleys, and multiple tech industries, existing side by side. And in the early 1980s, the business interests of the different parts of Silicon Valley were not aligned. The semiconductor crowd wanted trade restrictions so that their more expensive chips could regain market share. The personal-computer makers like Apple relied on cheap chips to keep their prices competitive.

  All the Silicon Valleys were united in their desire to pay as little in taxes as possible, but their tax-cutting priorities were different. Capital gains remained the chief preoccupation of the older generation, but you’d rarely find anyone from the younger crowd adding their voices. The Bob Noyces and Burt McMurtrys lobbied lawmakers in the road-tested manner practiced by nearly every other industry: visits to Congressional offices, blizzards of issue briefs, hours of testimony, sitting on presidential commissions. The Steves and Bills, only a few years removed from phone phreaking and stealing computer time, had neither interest in nor need for that kind of politicking.

  Regis McKenna—consigliere of the new generation, a lonely Democrat in a strongly Republican place—tried to get the personal-computer crowd to engage in good old retail politics. He threw fundraisers for Gary Hart’s 1984 presidential bid; he played matchmaker between his industry friends and the new generation of tech-friendly politicians. But it was hard to get traction in the middle of the Reagan Revolution, no matter how many Atari Democrats popped up on Capitol Hill. Jobs’s misadventures with the Apple Bill hadn’t helped. And having “Silicon Valley’s Congressman” hard at work in D.C. made little difference to McKenna. He and Ed Zschau had very little patience for one another, both being strong partisans with very different ideas about what Washington needed to do to support the tech industry.

  * * *

  —

  So whatever happened to industrial policy? Jerry Brown pushed it; Atari Democrats couldn’t stop yapping about it. Republicans like Ed Zschau could barely stand the idea, even while agreeing that something had to change. Japan was eating the United States’ lunch because of MITI. Yet the idea of another MITI made members of both parties uneasy. MITI was winning the war for chips and refrigerators and Walkmans because it closely controlled the national economy, subsidizing markets, choosing favored producers, and most definitely picking winners. Japanese “capitalist developmentalism,” as economist Chalmers Johnson dubbed it, just wasn’t America’s style, and certainly not in the Reagan era.

  What was in style: defense spending. The Reagan Revolution swung the military-industrial complex back into gear, waking up the beast of defense contracting that had lain dormant since the drawdown of the late Vietnam era. Austerity might have ruled the day across most of the executive branch, but not within the Pentagon, where budgets swelled with every year Reagan was in office.

  And just like in those months after Sputnik, the defense surge involved a great deal of money for university research. But this time it wasn’t for missiles and rockets: it was for supercomputers and AI and scenario modeling and cybersecurity. ARPA had added on a “D” for defense and become DARPA, yet it remained the Lady Bountiful of computer science. In 1983, as Democrats rebuilt the road to economic opportunity and Ed Zschau and David Morgenthaler urged another capital gains tax cut, DARPA began a push for a new program: one to meet the national security challenges that the Japanese economic threat represented. And it was all about computers.

  September 1982—the same month that Jerry Brown’s innovation commission issued its fifty-point plan for the economic future—the people of DARPA released “A Defense Program in Supercomputation from Microelectronics to Artificial Intelligence for the 1990s.” The title was clunky, the content dry, the public awareness nonexistent. Half a year later, the unveiling of the program the report inspired—the Strategic Computing Initiative, or SCI—barely made a ripple in the press. But the proposed budget was hefty: $650 million, more than the entirety of the DARPA budget only a couple of years before. And its pedigree was unmatched. DARPA recruited PARC’s dynamic and detail-oriented Lynn Conway, co-creator of VLSI, to help steer the program. Contracts went out on a sole-source, noncompetitive basis to the most elite departments in computer science: MIT, Caltech, Stanford, Carnegie Mellon.

  Japan was working on its “Fifth Generation” computing program, an ambitious push for supercomputing, AI, and machine learning. DARPA’s new push aimed to go one better, and to harness the power of cutting-edge computing to more ordinary defense challenges along the way. SCI’s opening agenda included projects to build autonomous vehicles, a computerized copilot for fighter jets, and AI software to aid in battlefield decision-making.

  The program’s first months were rocky, to say the least—Conway and other Class A technical talent hightailed it out of there quickly, and scholars had decidedly mixed feelings about the program’s military aims—but the computing push ultimately had a tremendous impact. Between the early 1980s and the mid-1990s, federal funding for computer science research more than tripled, to close to $1 billion per year flowing to academic laboratories. The Pentagon was the largest funding agency by far, supporting advanced research in supercomputing, chip design, and AI. All told, federal grants accounted for 70 percent of the money spent on academic research in computer science and electrical engineering between the mid-1970s and the new millennium. The majority of that federal total came from DARPA.36

  Government money turned the academic halls of computer science into the seedbeds of Internet-era innovations in search and social networking and cloud software, not to mention funding a generation of computer-science graduate students who would go on to design the future of software in Silicon Valley and beyond. Politicians contin
ued to argue about what path was the best one to take when it came to countering the 1980s challenge of Japan. But as they squabbled, the military brass and the bureaucrats decided the matter for them.37

  America’s MITI would be DARPA, and the work it performed over the next decade not only put the U.S. back in the lead in the global high-tech race. It also made Silicon Valley richer and more influential than ever.

  CHAPTER 16

  Big Brother

  While the leaders of DARPA were thinking about the computers of the future, however, the people of Silicon Valley were preoccupied with the computers of the present. For the ominous Borg that most Valley tech folk worried about in the first years of the 1980s wasn’t the Pentagon. It was IBM.

  Conventional wisdom long mocked IBM for being asleep at the wheel when it came to the personal-computer revolution. In reality, Big Blue had been testing minicomputer prototypes nearly as long as the Apple II had been on the market. To be sure, the antitrust suit brought in the waning hours of Lyndon Johnson’s presidency had dragged on into the Reagan era (it would be dropped by the DOJ in 1982), and the battle had sapped IBM’s energy and constrained its ability to enter new markets. But by the summer of 1980, as Apple mania seized Wall Street and VisiCalc demonstrated that personal-computer software could indeed have business applications, the IBM brass had agreed to go forward with a personal-computer product, christening it with a weighty name: The Manhattan Project. Needing to move fast and do things relatively cheaply, the managers in charge did exactly what Vannevar Bush had done forty years earlier. They outsourced, trekking out to all corners of the country in search of potential hardware and software partners.

  Up in the evergreen upper-left, they found Microsoft.

  BEYOND BASIC

  After Altair went into its tailspin, Bill Gates and Paul Allen had moved Microsoft back home to Seattle in 1979 (dropping the hyphen in its name along the way). The founders had first contemplated going to the Valley, but Allen had argued against it. Bay Area engineers job-hopped too frequently. The cost of housing was too high. Seattle might have gloomy weather, but its rain would keep everyone in the office, working like maniacs.

  Relocation northward coincided with a boom in sales, as more and more companies jostling for position in the personal-computer business started bundling Microsoft BASIC with their machines. Microsoft’s enviable sales numbers soon attracted the attention of H. Ross Perot, who offered to buy the software company for a considerable sum (Perot later asserted it was upward of $40 million; Gates remembered it as significantly less). The young mogul had entertained the possibility seriously enough to go to the trouble of getting a haircut before the meeting with Perot, but he ultimately said no.1

  By the time IBM knocked on its door that June, Microsoft was on its way to selling over $7 million in software and had forty employees. Like so many other high-tech founders, Gates and Allen had plunged everything into their company, making its success a 24/7 pursuit. Bill Gates had the same approach to vacations as Fred Terman: the only break he took from Microsoft’s total immersion was one week per year at an intensely competitive tennis camp.2

  Gates was still owl-eyed and skinny, but his self-assurance had grown along with his sales numbers. He wasn’t surprised that IBM was coming to call—his software was becoming the global standard, after all—but he was excited about the possibilities of scale that a partnership with Big Blue might bring. Only a few weeks earlier, he’d persuaded a friend of his from Harvard, an exuberant Detroiter named Steve Ballmer, to drop out of Stanford’s MBA program and take on some of the operational duties of his growing company. Microsoft was a land of sweatshirts and rumpled khakis; Ballmer was, as Gates put it, “a good suit-type guy.” Gates asked his more presentable colleague to join him at the table for the IBM pitch.3

  Ballmer may have had the suit, but Gates ruled the room. The twenty-four-year-old reveled in the technical details, and IBM didn’t awe him. So many of the early tech billionaires had grown up in families without connections or wealth. Gates was different. He was the grandson of a banker and the son of a successful lawyer. His mother, Mary, was currently chair of the board of the national United Way—alongside none other than IBM’s new CEO, John Opel. (When an IBM executive mentioned the Microsoft deal to Opel at a later stage of the negotiations, the CEO responded, “Oh, that’s run by Bill Gates, Mary Gates’s son.”)4

  Gates also remained steadfast in the conviction that had propelled him to write his famous letter to the Homebrew Computer Club several years earlier: that computer software should be a stand-alone intellectual and commercial product, as valuable as the hardware itself. The computer industry that IBM had defined and dominated for so long was a full-stack business, with hardware and software developed and sold as one. Apple had carried on that model as well. Now, in their haste to get a competitive microcomputer to market, IBM executives were willing to break up the stack, to do things another way—Bill Gates’s way.

  In that Microsoft conference room on that June day, the conversation soon expanded well beyond BASIC. Big Blue was interested in Microsoft’s whole line of computer languages, and it also wanted Gates’s help in finding the right operating system for it. In a chain of events that ultimately became legendary in computer circles (and whose disputed details became another proof point for Silicon Valley’s later dislike of all things Microsoft), the IBM team tried and failed to make a deal with California-based developer Gary Kildall, designer of the operating system, CP/M, that seemed poised to become the market standard.

  As the whole plan teetered, Gates swooped in, adapting an OS from another Seattle-based company into something he called MS-DOS. Kildall called it a clone of CP/M. Many years and lawsuits later, Kildall struck his own deal with IBM for his operating system, but by that time Microsoft had cemented its lock on the IBM personal-computer universe. The river of money to be made in software flowed northward to Seattle.5

  SILICON VALLEY NORTH

  Even as Microsoft and IBM continued their negotiations, investors had started vying for Bill Gates’s attention. Microsoft was still a partnership between Gates and Allen; it hadn’t even incorporated. As Burt McMurtry remembered it, by the autumn of 1980 Bill Gates was “having his door beat down by investment bankers” and others who hoped to get an early stake in the company. By that time, McMurtry and Reid Dennis had parted ways, and the Texan had a new firm—Technology Venture Investors, or TVI—and a new set of partners. It was the youngest of them, a thirty-year-old rookie VC named David Marquardt, who managed to squeeze through Microsoft’s door.

  Gates wasn’t dazzled by the wing-tipped Wall Street types with their term sheets and promises of spectacular stock valuations. Getting rich didn’t interest him; he had an indifference to net worth that was a privilege of people who had grown up in affluent families. He cared about the technology, and he liked people who were “smart”—and by that he meant “technically smart.” Plenty of his parents’ friends were eager to invest, but hometown money came from timber and airplanes—not tech. Gates wanted technical smarts along with that cash, and that only could come through connections into the Valley’s high-tech ecosystem.6

  This is where Dave Marquardt stood out. He was a tech geek too: a mechanical engineer, a former Homebrewer, who understood compiler design as well as capital markets. Over many months of courtship, which sometimes involved accompanying Gates and his parents to University of Washington football games, Marquardt and TVI put in $1 million to secure a 5 percent stake in Microsoft. Marquardt became a Microsoft board member, staying on for more than thirty years.7

  Marquardt’s investment and mentorship became only one of several important threads—both collaborative and antagonistic—connecting Microsoft and the Valley. Not too long after Marquardt showed up in Bellevue, he introduced Gates to an intense and energetic computer scientist from Xerox PARC named Charles Simonyi.

  Like Intel’s Andy Grove, Simonyi was a Hungarian
émigré, the son of a professor of electrical engineering in Budapest. He had escaped the oppressive regime of his native land as a teenager, first to Denmark and then to Berkeley and to Stanford, then to PARC. Despite Xerox’s continued difficulties in commercializing PARC’s marvels—“like one of those poor lottery winners that blew their millions,” Simonyi said ruefully—the lab had continued to be a mothership of extraordinary invention, and Simonyi had racked up a string of important accomplishments during his time there. He had developed the “killer app” for the legendary Alto that allowed the characters typed into a computer to appear on the screen just as they might on paper. Further refinements using icons and drop-down menus turned a laborious editing process into an easy point-and-click. The graphical user interface—or GUI—was a huge breakthrough, and everyone in the industry knew it. The previous year, Simonyi’s team had given a little demonstration of the software for an intensely interested Steve Jobs (who had persuaded Xerox to, in his words, “open the kimono” by allowing the photocopying giant to make a $1 million pre-IPO investment in Apple).8

  Simonyi had become restless at PARC, and he’d heard interesting things about what one friend called “this crazy guy in Seattle.” It only took one conversation for the Hungarian to realize that he’d found a high-tech soulmate, someone who could get just as animated and utterly consumed by the technological future. By Christmas, Simonyi had joined the Microsoft payroll, becoming what he later described as “the messenger RNA of the PARC virus” and bringing a whole new universe of software possibilities to Microsoft.9

  Another piece of connective tissue between north and south: Intel, whose chips IBM had chosen to power its new PC machines in another disruption of the full-stack model. Intel was far more well-established and deep-pocketed than Microsoft at the time, of course. Yet for a company then locked in fierce fraternal battles with its fellow chipmakers and buffeted by a seemingly unstoppable Japan, this stake in the future of the market was transformative. Thanks to the chips inside all those PCs, Intel’s revenue nearly doubled between 1980 and 1984, to more than $1.6 billion. By 1990, revenues were close to $4 billion, and rose from there as the “Wintel” empire of Microsoft Windows–platformed, Intel-powered machines took over the desktop market.10