big tech

The Silicon Valley Loop

How the dot-com crash created Palo Alto’s clueless investor class.

Photo-Illustration: Intelligencer; Photos: Getty Images
Photo-Illustration: Intelligencer; Photos: Getty Images

It’s difficult to narrativize Silicon Valley history. Narratives usually have a rising and falling action. Conflicts build, peak, resolve. The end. How, then, to tell the story of Palo Alto, where the conflicts swell but never seem to crest? Here, Icarus dusts himself off and pivots to zeppelins.

Once again, the tech industry finds itself in a contrite low-altitude phase. Having benefited from a generally hot stock market and a specifically hot sector amid a pandemic turn to online spending, tech has suffered from the Fed’s rate hikes and a reversion to in-person goods and services. Leading public firms — from hard-drive and microchip manufacturers and phone-makers and software writers to payment processors and gig platforms and dating-app developers — took huge share-price hits, and waves of layoffs rippled through the techie labor pool. As Mark Zuckerberg noted in a tightly worded recent announcement that he was sacking 11,000 Meta workers, many people predicted that COVID-19 inaugurated “a permanent acceleration” of online growth “that would continue even after the pandemic ended.” Zuckerberg continued, “I did too, so I made the decision to significantly increase our investments. Unfortunately, this did not play out the way I expected.”

Promised savior technologies such as distributed ledgers, virtual reality, and automated driving failed to deliver, and the crypto wunderkind Sam Bankman-Fried has been confined to his family home in Palo Alto. Meanwhile, without the loose funding environment they expected, start-ups drowned in bunches like bags of puppies thrown over a bridge.

Tech’s go-to cautionary tales are about hubris, bad business plans, and overvalued, undercollateralized start-ups, yet these remain three of the region’s biggest exports. It’s one thing to fall for Bankman-Fried’s magic money box; it’s another thing to do so by the flickering light of tell-all TV shows about Uber, Theranos, and WeWork. Even during this latest downward readjustment, there’s no good reason to believe there will be structural changes or even a lot of individual consequences in Silicon Valley. Microsoft is simultaneously executing 10,000 layoffs and a $10 billion investment in the fledgling software company OpenAI, valuing the money-burning start-up at something like $60 million per employee. The emperor’s nakedness revealed, he shrugs his shoulders and gets back to ruling.

Silicon Valley’s favored downturn metaphor is the bubble. In a land of exciting inventions and instant fortunes, enthusiasm tends to get out of hand, pop, and return to manageable levels. But if the bubbles keep happening, then there must be a bubble machine of some sort, a structural reason why the tech industry tends toward effervescence. The right metaphor is not a bubble, actually; it’s a loop. To find out what’s going to result from tech’s current downturn, we should take a close look at the history, despite it being — in fact, because it is — difficult to narrativize. Let us then turn our attention to the so-called dot-com bubble.

In the second half of the 1990s, newly deregulated financial markets dropped piles of cash on fast-growing internet companies. After Silicon Valley tech stocks such as Apple, Genentech, Sun, Cisco, and especially Netscape returned on their boundless potential, savvy investors felt they had no choice but to jump on the next new thing as fast as possible, even when the new thing didn’t have an obvious road to making profits or revenue. The early web was a world of enthusiasts: Literary Kicks archived Beat literature. San Francisco’s fog got its own site, as did the band Megadeth. Legacy media institutions put up their sites only to find themselves sitting alongside and guys named Justin and Glenn and Jerry posting links to sites they liked.

It was unclear at the time where the line between the web’s overlords and users was going to go. Were these user sites the web’s content or its infrastructure? The internet already traveled through a series of bottlenecks from the operating system to the router and the modem to the servers and the browser. Each one of those steps was worth a lot of money. What was one more? Investors looked for new layers to sit on top of the web. One of those link sites from 1994 was called Jerry and David’s Guide to the World Wide Web after its curators, Jerry Yang and David Filo, Stanford electrical-engineering grad students and web enthusiasts. Their directory was getting a lot of traffic. Once people got online, they had to figure out where to go next, and that was Jerry and David’s layer. They renamed the site Yahoo and locked down a few million dollars from Sequoia Capital in the spring of ’95. A year later, the site’s IPO made them one of the first true dot-com success stories.

This “portal” layer was the most valuable spot on the web for an obvious reason: Users went through it no matter where they were going. Yahoo had to scrap it out with a handful of competitors, including the face of anticompetitive tech, Microsoft. Bill Gates wasn’t content to let competitors snap up doors between Windows users and content, so Microsoft used its operating-system revenue to support attacks on web competitors under the MSN banner. But thanks to big cash infusions from the market, some start-ups, including Yahoo, could compete with the big guys by acquiring other start-ups. That’s what Jerry and David did, buying a games company, a DIY-website company (GeoCities), a groups company, and a messenger company in quick succession. Microsoft kept pace, buying the free email site HoTMaiL in 1997 along with its user base. For start-ups like HoTMaiL — bought within 18 months of launch for an undisclosed amount, reportedly $400 million to 500 million — an acquisition could be just as good as an IPO. Investors were happy to finance the feeding frenzy by bidding stock prices up. Microsoft’s went from under $7 in 1995 to $58 by the end of the decade. Yahoo’s skyrocketed over $110, a classic example of what the Federal Reserve chair, Alan Greenspan, famously called “irrational exuberance.” Obviously something in the market wasn’t working normally, but was that a problem?

The question wasn’t whether there was a connection between internet stocks and what Greenspan called the “real economy” — the economy of production, jobs, and commodity prices — but rather what the precise nature of the relationship was. Greenspan’s worry was that a popping asset bubble could take it all down. Whether it actually would in the midst of the “portal wars” in 1996 was an open question. Maybe the hit would just go to fancy investors and their stock-optioned coders; part of the appeal to these companies was that they didn’t have many workers to lay off anyway.

“You’re either a zero or a one. Alive or dead,” says the Tim Robbins version of Bill Gates in the 2001 Palo Alto cyberthriller Antitrust, describing both the Valley’s business climate and its killing-spree competitive ethos. For investors, that meant they priced firms not based on expected returns per se but on the odds that they would become ones rather than zeros. Venture capitalists always played that way with a few big winners picking up the slack (and then some) for a bunch of losers. The Clinton administration did what it could to encourage the boom in the late 1990s, including imposing a bipartisan cut to the capital-gains tax in 1997.

Climbing stocks gave big fish the jaws to nibble start-ups the way VCs might or even to consume them whole. A lot of founders got rich off-loading what were more or less duds to incumbents. In 1999, computer manufacturer Compaq paid over $300 million for the city directory site Zip2, scoring Zip2 co-founder Elon Musk his first fortune. But Zip2 couldn’t save Compaq’s search engine and portal AltaVista, which quickly lost ground to the superior Google and was forced to cancel a planned spinoff IPO the following year. Mark Cuban sold his IPO’d but still money-losing to Yahoo in 1999 for more than $5.5 billion in stock. As the New York Times noted at the time, only speculative tech firms could afford to make bets like that. “Internet companies simply cannot be bought by established companies where stocks are valued on profits, not promise,” wrote journalists Saul Hansell and Laura M. Holson. “Thus is out of reach for such logical buyers as CBS or Walt Disney. But it is an easy bite for Yahoo, which can simply exchange its highly valued stock for that of” Yahoo could also afford to downsize a short couple of years later, taking a bath on the buy. Yahoo eventually got Zip2, too, buying the ad company Overture in 2003, only four months after that firm scooped up all of AltaVista for the bargain-basement price of $140 million.

The era’s winners defined themselves not so much by inventing the best websites — a lot of people had similar ideas — but by hedging and cashing out at the right time. Paul Graham sold his web-storefront software company Viaweb to Yahoo for around $50 million in Yahoo stock and had the sense to liquidate his award quickly. “By 1998, Yahoo was the beneficiary of a de facto Ponzi scheme,” he wrote later. “Investors were excited about the internet. One reason they were excited was Yahoo’s revenue growth. So they invested in new internet start-ups. The start-ups then used the money to buy ads on Yahoo to get traffic. Which caused yet more revenue growth for Yahoo, and further convinced investors the internet was worth investing in.” Jerry and David’s Guide to the World Wide Web was not a sustainable growth engine for the global economy.

The Y2K bubble was overdetermined; it had more causes than it needed. One was that technology hedge funds bid up stock prices with a plan to jump out at the high and leave less sophisticated capital holding the heavy bag. That strategy worked well enough, and the funds mostly came out of the experience surprisingly whole. But if the failure of brought investors to their senses and tanked the NASDAQ, then perhaps the coked-up sock-puppy mascot did some good, stopping people before they threw more money down the dog-food-delivery garbage chute. In the period after the pop, many in Silicon Valley talked like dreamers awakened. Analysts wrote off whole e-commerce sectors as folly. In a 2001 interview about his case study of the grocery-delivery site Webvan, Harvard Business School professor John Deighton said that direct-to-consumer internet advertising was here to stay, but “home-delivered groceries? Never.” There was no way to compete with supermarket efficiencies. Deighton’s prediction was — not to belabor the point — wrong. But whereas the conventional narrative was that dot-coms drowned themselves in wasteful spending, Deighton did see that it was the relation between the internet and the real economy that needed to change. As it turned out, it wasn’t the internet that had to do most of the changing.

The real problem with the delivery dot-coms was that the people running them didn’t understand their historical context. In 2013, Peter Relan, the founding head of technology at Webvan, published a post on TechCrunch discussing why the company failed and how the next round of delivery start-ups could avoid the same fate. Webvan’s strategy, he wrote, was to offer “the quality and selection of Whole Foods, the pricing of Safeway, and the convenience of home delivery.” But according to Relan, the company shouldn’t have invested in so much infrastructure. Webvan built high-tech distribution systems from scratch: giant networks of new algorithms, miles of conveyor belts, fleets of custom trucks with PalmPilot-wielding delivery drivers. At its short peak, Webvan had a billion-dollar contract with Bechtel to build new distribution facilities around the country. This was the utopian vision of e-commerce, one in which the web’s efficiencies generated gains for everyone involved: investors, workers, and customers alike, all financed by forward-thinking financiers who would reap their rewards in the better world soon to come.

Webvan’s idealism now seems less quaint than alien. In a 2000 report to the Securities and Exchange Commission, Webvan bragged that all its couriers “are Webvan employees … The courier training lasts two weeks and includes 36 hours of classroom training, 12 hours of driving training and 28 hours of on the job training … Webvan’s couriers receive a competitive compensation package, including cash and stock options.” Commentators pegged Webvan’s delivery-labor costs at $30 an hour, or over $50 in 2023 money. Of the company’s 4,476 reported employees on January 1, 2001, 3,705 worked out of the “real” operating facilities spread over nearly 1.5 million square feet of rented urban-warehouse space across seven metropolitan regions. The company filed for Chapter 11 bankruptcy in the summer of 2001 after losing hundreds of millions of dollars the year before.

Webvan successfully lured workers out of grocery and meatpacking unions with stock options, and it used the same overvalued stock to devour the Amazon-funded competitor HomeGrocer. The firm’s model was haphazard, though not in the way that seems obvious. Big investments in fixed capital and labor training were the key to decades of American prosperity, but that was a different era. Just as capitalists came to understand that they weren’t going to win the Cold War with high wages, no dot-com was going to win its particular race by handing out stock to delivery drivers. was making a similar push to expand its in-house warehousing-and-distribution system when the firm’s bubble burst. Industry leaders had to put the win-win-win tech-economy fantasies aside — that’s not what the internet was for. After their capitalist predecessors spilled blood around the world to get them out from under obligations to anyone except their shareholders, these fluffy web companies were being too responsible. It was all fine and good to issue press releases about how the web was going to improve life for everyone, but you weren’t supposed to actually spend billions of dollars investing in that image. “Move fast and build things” wasn’t going to cut it; what Silicon Valley needed was a reversion to Reagan.

When the dot-com bubble popped, it left a layer of winners: the middlemen at the big financial institutions, the bottom-feeders and big firms that cleaned up after, and the founders who happened to sell at the right time. Caught up in the frenzy, most large firms were stuck writing off at least one overvalued internet acquisition, effectively giving away millions of dollars to a cohort of web entrepreneurs based on the misperception that those dudes came up with important, useful stuff. But when their buyers wrote off the purchases, these men (they were mostly men) didn’t adjust their self-image downward. After all, they still had the money.

As the internet bubble reinflated faster than the first time, these men became semi-professional and then fully professional mentors to the next generation. Ramesh Balwani, for example, joined the web-auction start-up CommerceBid in 1999, right before it sold to bankruptcy-bound Commerce One for a couple hundred million dollars, of which Balwani was entitled to a substantial chunk. He went on to mentor (and date) a young Stanford student named Elizabeth Holmes as she dropped out of school to launch her paradigm-shifting diagnostics company, Theranos. Early Yahoo investor Masayoshi Son took WeWork co-founder Adam Neumann under his wing, pumping billions into the office-leasing concept and convincing Neumann to always think bigger and faster.

Dot-com winners often found ways to spread their next bets, institutionalizing their dubious wisdom. After his Yahoo play, Son’s holding company, SoftBank, became one of the tech industry’s key early-investment vehicles. Netscape vets Marc Andreessen and Ben Horowitz formalized their angel investing into the venture-capital firm Andreessen Horowitz, which raises and deploys capital in multibillion-dollar rounds. Some even pioneered new financing models in which hanging out was as important as paying up. In 2005, Graham of Viaweb and some colleagues from that Yahoo deal opened Y Combinator, an accelerator that traded advice, connections, and a little cash to start-ups in return for shares. Theirs was the best-organized institution — angel investing as a for-profit university — but there were more casual setups, too. The fictional prototype is Erlich Bachman, the extroverted bullshit artist at the center of Mike Judge’s HBO clown-era Palo Alto satire, Silicon Valley. Bachman sold his web start-up of unclear utility to Frontier Airlines and spun his winnings into his own accelerator: a moderately sized Palo Alto house in which he rents rooms in exchange for equity in start-ups. In the real Silicon Valley, having let the market convince them of their own brilliance, these medium-fry millionaires looked for the next big thing, the one that would bump them up by a zero or three. Some of them found it.

Chris Sacca was a lawyer at Google — Forbes described his early work as “going undercover to scout locations with low taxes and cheap electricity for Google’s new data centers and then creating nondescript holding companies to buy up the land” — when, in 2007, he quit and found a more important job: proprietor of the “Jam Tub.” Attached to his new house in the California skiing town of Truckee, Sacca’s hot tub was a semi-formal gathering place for tech founders to chill, work on their next brilliant ideas, and maybe rustle up some seed money. Sacca’s $25,000 check to his former co-worker Evan Williams for an early investment in his microblogging start-up Twittr gave him the juice, and his early support for successful companies such as Kickstarter, Twilio, and Instagram turned him into one of the era’s top venture capitalists. For a while, you could catch Sacca in his trademark cowboy shirts next to founder Mark Cuban on Shark Tank, where Sacca was a recurring guest “shark,” the two of them helping define what a 21st-century capitalist looks like. Still, Sacca counts the Jam Tub as his best investment: “I borrow the cash for a three-bedroom house and get a lifetime’s worth of pals and a hugely successful business in return? Best trade ever.”

Like many in the Jam Tub crew, Garrett Camp and Travis Kalanick suddenly had rich-guy problems. Both sold their web companies for millions in 2007 — Camp’s random-content portal StumbleUpon to eBay and Kalanick’s P2P firm Red Swoosh to Akamai — and the young dudes turned their attention to having a good time. Kalanick was the Jam king, spending full days submerged in Truckee, according to Tub lore. But as newly rich entrepreneurs, they ran up against novel frustrations. Camp couldn’t figure out a good transportation solution for hitting the town in San Francisco, for example, and after paying $800 for a black car on New Year’s Eve, he figured there had to be a better way.

The original plan for UberCab was for an elite members-only service leveraging the GPS-enabled smartphones that affluent consumers started to carry around, allowing them to summon a cab on demand. “Faster and cheaper than a limo but nicer and safer than a taxicab” was the pitch, and the membership model and high price ensured a “respectable clientele.” Camp bootstrapped the earliest work and talked Kalanick into joining. Worst-case scenario, they told potential investors, it would be a solution for tech-forward rich people in San Francisco — which is to say, them. Running a luxury service for wealthy Bay Area smartphone users was not a bad way to attract capital, and they assembled a crew of VCs and angels with investments ranging from $5,000 to just over $500,000, the biggest check from a guy who scored on StumbleUpon. Sacca threw in $300,000, cementing his reputation and fortune.

It didn’t matter that StumbleUpon and Red Swoosh weren’t ultimately worth anything to the companies that purchased them; what mattered was that the founders had made money for their investors, which made them Silicon Valley successful.

In evolutionary biology, there’s a term called carcinization. It describes the tendency of all sorts of crustaceans to evolve crablike bodies. The shelled core and spindly articulated legs are apparently an excellent way to adapt to the sea floor, and various species keep stumbling upon it, evolving to look like relatives even when they’re not. Silicon Valley’s 21st-century firms underwent their own type of rapid carcinization, flattening into “platforms” suspended on rows of contractor-pin legs. At first, the Uber guys clearly did not understand what they had, and neither did a parallel group of guys building their biggest competitor, Lyft. The two were made for disparate use cases — the Lyft founders admired the efficiency of ride sharing in Zimbabwe, which made full use of empty seats, while the Uber dudes wanted to pay less for a limo — but they converged on the same model. Like the winningest firms of the dot-com era, they tended toward monopoly plays, searching for social layers to disrupt with computers. When it comes to a given niche, that meant whoever could show the most and fastest growth could attract the most and fastest capital, which turbocharged growth, which attracted more capital, and so on. Competitive start-ups didn’t have to make profits, but they did have to scale, and immediately.

Market pressures compelled Uber to plow every bit of income (and then some) into growth over profit. The firm subsidized riders and drivers, changing the model from “Better than a cab, cheaper than a limo” to “Cheaper than a cab, so don’t take a cab ever again.” It was burning billions of investor dollars, faster than any start-up in history, but already the drivers felt betrayed. Inexorably, as the rideshare companies broke the cab cartels, they pushed down wages for working drivers. This was one of the first large battles in what was, in retrospect, a struggle over the nature of work. Uber automated as much of the recruitment, sign-up, and onboarding processes as possible — forget training. These gig workers were barely even contractors. In terms of their relation to the company, they were more like users. By cutting the ribbons holding together suites of labor laws, the lean crabs freed workers to “create demand for their labor at the expense of their incomes,” as economic historian Aaron Benanav puts it.

It’s a mistake, then, to think of Uber’s carcinized business strategy as driven by its scandal-prone leader, Kalanick, and his bad personality. Nor can we dump all the fault on the Jam Tub and the “Party like it’s 1999” brosphere it represents. There are larger forces at play. When author Brad Stone asked Kalanick why the company raised over $10 billion in the previous two years alone, the billionaire’s answer comes off more as resigned than pumped: “If you didn’t do it, it would be a strategic disadvantage, especially when you’re operating globally,” he told Stone. “It’s not my preference for how to build a company, but it’s required when that money is available.” That last part is worth repeating: It’s required when that money is available. If Uber didn’t take $3.5 billion from Mohammed bin Salman and the Saudi kingdom’s sovereign-wealth fund, the royals would have put it in Lyft, and then maybe no one would want to invest in Uber, and then it would all be over. These companies didn’t choose to become crabs — that’s not how evolution works.

Staying in the game was much more important than any imminent prospect of profitability, and platforms courted big bucks from Russian oligarchs, Emirati sheikhs, and cosmopolitan capitalists of every stripe. Even Canada’s Public Sector Pension Investment Board put $500 million on Lyft. Early investors were rewarded as new investors inflated the stock value and made them look smart. The region sprouted a whole crop of paper billionaires. Yet the Uber IPO flopped, and though the stock price has fluctuated in the years since, the public has not been as enthusiastic as the VCs were. The firm has continued losing money, still slugging it out for a monopoly spot that might turn the numbers around. But Uber has a market capitalization of over $40 billion at the time of this writing, and investors brag about when they bought in. Not bad for a company that hasn’t come close to making a dollar in profit.

Compared with past cohorts of successful Silicon Valley tech founders, the crab-platform leaders made Steve Jobs look like Steve Wozniak. Not only did they not build anything substantial — most of them didn’t have the technical expertise to know where to begin — they also didn’t even come up with anything new. Still, investors pumped novel magnitudes of value through these platforms, allowing them to pursue money-losing strategies indefinitely and hold out for monopoly positions. Since the start-ups were little more than fantasies before their first six- or seven-figure infusions, early investors in the top crabs got extraordinary hauls. VCs couldn’t afford not to take chances on hare-brained schemes. “Airbnb for X” and “Uber for Y” pitches proliferated. What is the lesson there? Whatever it was, capitalists took it.

Despite how narratively convenient it would have been, Silicon Valley didn’t sober up in the years following the bust, and neither did the investors who inflated the bubble in the first place. Post-pop heavyweights like Amazon and Google picked up the broken pieces and made the best of them. The “dot-com bubble” sounds like the name of a great cautionary tale, but any capitalist who kept taking their investment cues from the collapse of missed out on a lot of money. The growth that leading firms have accomplished since then has almost fully obscured the turn of the century on the stock chart, reducing it to mere first-night jitters. We are now a couple of cycles past, and today’s youngest tech founders weren’t even born yet when tanked. As for who’s a success, that’s measured in money, and the money can’t remember how it got there either.

One of the firms in the inaugural class at Viaweb founder Graham’s Y Combinator was a location-sharing app called Loopt. Founded by an archetypal Stanford-sophomore dropout, Loopt was quick out of the gate, nabbing a $5 million Series A led by VC big shots Sequoia Capital. In 2006, Graham called Loopt “probably the most promising of all the start-ups we’ve funded so far” — a list that included Reddit and Scribd. But despite racking up tens of millions in investments, Loopt was overtaken by Foursquare and, in 2012, sold itself to the prepaid-debit-card company Green Dot for over $40 million. Green Dot, which had no need for Loopt, promptly scrapped the project and put the team to work on a banking app. This is what Silicon Valley calls a success, though industry observer Erick Schonfeld described the deal as “Sequoia taking care of its own,” since the VC firm owned a large stake in both companies.

Y Combinator took care of its own, too, and after the acquisition, Graham approached the 27-year-old Loopt founder about becoming his successor at YC. “I decided I was going to partly take a mid-career sabbatical, race cars, fighter planes, travel the world, all that kind of stuff,” Sam Altman said later about his thinking at the time, “but I didn’t want to totally disengage from working, and I would try to invest for a while.” Altman took over for Graham in 2014, and he ran the accelerator to great heights before leaving in 2015 to focus on the artificial intelligence project he co-founded with Zip2’s Elon Musk: OpenAI. Loopt, indeed.

The fundamental moralistic assumption that, over the medium term, the market punishes hubris and rewards prudence is simply not true. Yesterday’s lucky fools are now wise billionaires not because they learned responsibility but because, having sussed out the general shape of the world that capital is accumulating into, they doubled down. The true moral of the dot-com bubble was “Go big fast, keep labor and fixed capital costs low, time your exit.” For this cohort of wealthy tech guys, the same one that’s responsible for a disproportionate share of early-stage private-capital allocation today, the inescapable structural-historic lesson of the dot-com bubble was “Do the dot-com bubble again.”

The Silicon Valley Loop