Taking not Making. Has Silicon Valley failed at Digital Innovation?
This piece has been written with Vincenzo Luise, post-doctoral researcher at the University of Milan.
The aborted IPO of the tremendously overvalued co-working company We Work was the latest in a series of series of events that, over the last year, have raised public concern about the Silicon Valley start-up/ venture capital systems ability to deliver. (‘Can uber ever make money’ as the ever cautious Financial Times entitled one of their analyses already two years ago) After over thirty years, where Silicon Valley style start-ups have seemed to be the only way to institutionalize digital innovation, maybe it is time to look at what they have delivered? Is the silicon valley start/up venture capital model living up to its own claims? And more urgently, is it likely to produce the kind of innovation that we need to confront the accelerating environmental and social crisis of the early Anthropocene?
The startup system divides into three components. First, startups, or small, newly formed companies that mostly exploit emerging high-tech niches or other potentially lucrative but fundamentally insecure future markets. Second, venture capital: in the form of initial seed funding or angel investing, or in the form of more structured venture capital funds. Venture capital is a form of private equity oriented to high risk and high returns. Its purpose, at least in theory, is to support companies as they grow large enough to enter pubic capital markets via an IPO (Initial Public Offering). Third, intermediaries like accelerators or incubators. The difference between the two is fleeting. Some are simply co-working spaces for startups. Most offer support, consulting and a plethora of business services, from accounting to legal advice and connect startups to venture capital. Many incubators or accelerators make small (‘seed’) investments in the startups that they ‘incubate’ in return for a share of these companies. This business model was pioneered by the Silicon Valley based Y –Combinator, which has incubated famous tech ‘unicorns’ like AirBnb, Dropbox and Coinbase, among others.
All of these components have developed during the post-War years, in relation to new investment opportunities in high-tech sectors like computing, data processing and medical technologies. Starting in the 70s they came together in California’s Silicon Valley (along with a number of similar areas, like Boston’s Route 128) to provide the institutional framework for supporting new ventures producing silicon based microprocessor ‘chips’ for the booming computer market, which required new forms of capital ready to take great risks for similarly great returns. The system evolved with the emergence of a computer and software industry in the 1980s and the successive dot.com boom that accompanied the commercialization of the internet in the 1990s, to become the principle way of financing ventures into the, then, uncharted terrain of the online economy. In the last decade the number of start-ups, incubators and venture capital funds, has skyrocketed globally expanding from the ‘homeland’ of Silicon Valley into Europe, Asia and Africa. Incubators and accelerators have been promoted by public actors like universities, national and city governments as ways to promote innovation, to counter youth unemployment and to boost local economies. Like ‘creativity’ a decade ago, the entrepreneurial attitude fostered by start-up incubators has come to be seen as a way to give young people, in particular, an adequate mind-set to deal with the insecurities of a precarious labor market, and to address a and solve a number of social problems, from unemployment to petty crime. How has it worked?
Not very well. While there are many success stories of venture capital translating into returns hundreds or even thousands times the initial investments- starting with Digital Equipment Corporation which gave back 1200 times the money already in 1968, overall performance has been much more modest. With the exception of investments in the dot.com boom in the mid-1990s, venture capital funds have, on average mirrored the performance of the S&P 500 index which is often used as a benchmark for the corporate economy overall. The mid 1990s also saw a peak in overall US venture capital investments and those investments, going into companies like Amazon, eBay and –even if they are older- Apple and Cisco, have generated returns of double or even triple the S&P average. However, with the exception of this period, the startup venture capital system has not been able to beat the market. Even from a purely financial point of view, it has not been able to make new tech significantly more profitable than old tech.
What sort of companies has the venture capital system generated? In theory, startups are small companies with above average potential for profitability. The so called Californian Ideology that accompanied the development of the Silicon Valley startup system in the 1990s emphasized how digital technologies would generate a multitude of small companies operating on markets that were egalitarian and transparent. Mark Andreessen, partner to the successful venture capital fund Andreessen & Horowitz, predicted, as late as 2011, that the new economy resulting form –software (startups) ‘eating the world’- would be marked by ‘real, high-growth, high-margin, highly defensible businesses’. However, in retrospect the start-up system has instead generated a small number of disproportionately large and powerful actors who, despite this, struggle to make a sustainable profit. Instead of a large number of small profitable companies, it has generated a small number of large unprofitable companies.
Why is this the case? First of all the start-up system uses a ‘spray and pray’ approach to investment, investing in a large number of essentially similar start-ups hoping that a few will make it. This means that the ones who do make it need to make it big to compensate for the losses endured by the others. This way, venture capital tends to privilege growth and scaling-up even at the expense of profitability and market sustainability. As one ‘serial tech entrepreneur’ put it in his TechCruch contribution, ‘the only realistic way for a fund to get acceptable returns is to try to find only the companies that could be the next Ubers, Facebooks and Airbnbs. [..]. There’s just no place for “average” companies looking to be worth and sell less than $500 million.’ The recent tendency to postpone IPOs, and the public scrutiny of plans for profitability that this entails, has tended to exacerbate the trend. To put it simply, it has become less important to achieve market sustainability, even in the medium run, and more important to maintain a fiction of potentially endless expansion even while incurring substantial losses. As we- work CEO Adam Neumann famously claimed in a recent interview: ‘No one is investing in a co-working company worth $20 billion. That doesn’t exist. Our valuation and size today are much more based on our energy and spirituality than it is on a multiple of revenue.’
The recent boom and bust of bike sharing companies like Ofo (and the present similar trajectory of eScooter sharing enterprises) illustrate this well. These companies have come to exemplify a new business model sometimes called ‘V2C’ or Venture Capital to Consumer, where venture capital is used to subsidize services, like bike or scooter sharing, that nobody wants to pay a full price for, in order to ensure market expansion and thereby underpin growing financial valuations. While Ofo provides a good example of the irrationality of this strategy, eventually going bankrupt with hundreds of thousands of bikes ending up in Chinese landfills, Uber and Lyft have pursued a similar strategy, enduring real losses in the billions in order to support artificial market expansion that can translate into more or less fictional increases in valuation. Whether aiming for an IPO or, more commonly to be bought up by a large company, venture capital does not primarily strive to secure sustainable profits, but rather seeks to enhance valuations by sustaining a myth of future profitability through, among other things, rapid market expansion. Consequently start-ups are not primarily seen as young companies headed for as sustainable market position, but as financial investment vehicles to be hyped up and then passed along to the next stage. Indeed it is telling how much easier it is to obtain information on investment rounds, market share or even social media popularity for start-ups, than data on their real economic performance, The later does not matter much in overall investment decisions.
As a consequence of the tendency to postpone actual profits almost indefinitely, valuable start-ups tend to project a fictional future of world domination. Uber will eventually dominate the market not simply of taxi rides, but for ‘transportation’ overall. This tendency is reinforced by the abundance of capital available, which tends to be concentrated to the actors that have already shown above-average growth potential, if only in terms of the potential to attract capital. In other words, since the only thing that will eventually pay-off is a Facebook style monopoly over a particular market, and since the only ones that are able to realize such a monopoly are the ones that have already received the greatest investments, new investments tend to go to start-ups that already have attracted large investments in the past. This self-reinforcing tendency leads to a quest for ‘unicorns’ (or ‘pluricorns’) who are able to attract investment in the range of billions of dollars, on the promise of them eventually being able to realize a worldwide monopoly over a particular sector of human activity. Some have been successful in doing this: Amazon, the most significant start-up to survive the do.com era of the 1990s, has managed to virtually monopolize online retailing and is making a modest, if stable profit from that. Facebook, has, for now, managed to fence in a substantial part of the online advertising market and is realizing modest earnings from that.
We write ‘modest’ because if you take away the ‘energy and spirituality’, even the unicorn companies that do make it to the IPO are not particularly profitable, at least not by old-style, industrial society standards. And the more recent they are, the less profitable they become.
Figure II. Price/earnings (p/e) ratios for major tech unicorns, using NASDAQ data.
p/e (S&P average, 21.5)
Apple
17,5
Microsoft
31,5
Alphabet
34,3
43,4
Amazon
227,5
Uber
68 billion valuation, $404 million loss
Only Apple has a price/earnings ratio that is better than the S&P 500 average. Alphapet (Google) and Facebook are about half as profitable as an average multinational company; Amazon is ten times less profitable and Uber (and Lyft) keep operating at substantial losses. As the title of a recent editorial in The Economist read: ‘Tech’s new start have it all- except paths to high profits’. So once again, the impression is that the start-up system is unable to realize the potential value that lies in the new relations of production that come with digital technologies.
We seem to have fallen back into what, back in the 80s, Nobel laureate economist Robert Solow famously called the ‘productivity paradox’, the fact that ‘we can see computers virtually everywhere except in the productivity statistics’. Similarly smartphones, always on internet connectivity, advanced algorithms and massive data gathering has transformed our lives, but these technologies seem unable to generate economic growth, or even average profitability for the companies that exploit them. Why is this the case?
At the very simplest, but also the most profound level, it is because of a lack of imagination. Essentially digital technologies have been used to conserve, marginally improve on, and above all monopolize a number of market sectors that already have been established within 20th century consumer society. There used to be an advertising market, composed of newspapers, television stations advertising agencies and, later, media bureaus. Now most of the advertising revenues go to Facebook and Alphabet. There used to be a retail sector, made up of shopping malls, department stores, supermarkets, and lately, Walmart. Now amazon is taking over that sector. There used to be a multitude of taxi business, now Uber and Lyft hope to ‘eat’ them as well. And so on. It can be debated whether software platforms thus ‘eating the world’ has improved the customer experience. A growing number of people now claim that it has not. The point here is that the process of ‘eating the world’ has not generated much in terms of new market revenue, but simply concentrated existing revenues to a small number of platform unicorns who now control transactions across whole markets. (While online advertising budgets have skyrocketed since 2000, press and television advertising budgets have contracted. Overall the proportion of ‘Adspend’ to US GDP ha actually shrunk from over 2 per cent in 2000 to 1.2 per cent in 2017).
The platformization of these markets has realized one central tenant of the Californian ideology. It has broken down barriers and destroyed entrenched privileges. This has meant that more actors have entered these markets. But since overall revenues have not expanded much, these new actors tend to operate at very low margins. Indeed their productivity is so low that their pay tends to cover only marginal costs, not the costs of reproducing capital. AirBnb has vastly expanded the amount of people offering hospitality experiences. This has lowered the cost of accommodation. But it has also lowered the revenue of hosts to the point that it is generally impossible to finance a mortgage through AirBnB revenue alone. It only works if you already have a house, it will not allow you to buy one. (Or rather high earners on AirBnb are generally professional -hosts- who manage multiple properties). Similarly Deliveroo ‘riders’ do not earn enough by riding for Deliveroo to buy the smartphone or the bike that they need for work. It only works if they already have these things and are ready to make additional revenue at the margin. Working in the knowledge worker sweatshops where Siri is trained certainly will not enable you to finance the MA in linguistics that is required for employment. In this sense, even the meagre revenues of platform unicorns are based on their ability to exploit productive resources that are already there and that they do not contribute to reproducing. The same logic applies at a more general level: the tax avoidance strategies of these companies means that they actively avoid contributing to reproducing the infrastructure that they depend on: From schooling to general literacy via medical care for damaged riders and Amazon warehouse workers (who generally last about three years on the job) to the basic infrastructure of digital connectivity. This way when start-ups are successful they are primarily extractive. They are in the business of taking, rather than making, to use Jason Moore’s characterization of ‘neoliberalism’ overall. They seek to extract as much revenue as possible through their control of particular markets, but they do not invest in making those markets more efficient, much less in opening up new markets able to meet different and more urgent kinds of needs. However even as extractive businesses they are not particularly efficient. Paradoxically the large unicorns that have come out of the start-up system are labor intensive and capital intensive at the same time. They employ large armies of workers with low productivity and meagre pay, and vast amounts of capital to generate revenues that are overall rather meagre, without much space for future improvement in profitability. Spotify already has 100 million paying subscribers but still makes quarterly losses of around $ 50 million. Facebook has exhausted the potential of the online advertising market and is already causing attention fatigue among its most attractive users, not to mention mounting privacy concerns. As the Libra initiative illustrates the company’s only hope is to try to operate as a bank instead. Google’s meagre margins is under threat from the European web tax along with growing privacy concerns and antitrust sentiment.
This inability to guarantee ‘healthy and sustainable’ profits is paralleled by the low profitability of incubators. Accelerators are now diversifying away from their original business- start-up acceleration- that shows less profitable than expected. Even Y-combinator has started to launch online courses in order to expand its base of retail investors. In 2016 only 77 out of 579 Acccelerator programs world-wide reported that they had an exit. In total, there were 178 exits worldwide, out of 11.305 start-ups. In fact, in 2015, with the exception of Africa and the Middle East, the traditional ‘cash-for-equity’ model, which involves investing a small amount of seed money in a start-up — around $25,000 on average — in exchange for equity (usually between 5% and 10%) is being abandoned. To quote the 2016 Global Accelerator Report, by no means a critical forum:
Increasingly, this model is becoming rare, as more accelerators reconsider their general outlook. Most likely, the small number of exits — 178 reported in 2016 — has proven insufficient in funding their operations. Consequently, many accelerators around the world no longer rely on generating revenue from exits.’ Accelerators have relied on, and continue to explore, new models of revenue generation. 90.4% of accelerators plan to increase their revenue in the medium to long term by incorporating alternative revenue models in addition to exits. 52.1% of accelerators are at least partially funded by a corporation, and 67.2% aim to generate future revenue from services sold to corporations.
So not even the taking is going very well. Or, rather, it is increasingly concentrated to a small number of wealthy insiders. The growing tendency to postpone IPOs, and the public scrutiny that they entail means that an ever-larger share of the financial value generated by start-ups remain within a circle of tightly connected inside investors. As Mark Andreessen laments ‘that just creates wealth inequality, it’s terrible’ (Waters 2019). If the start-up system might have been efficient in developing new markets for personal computers and software, and to some extent also for online advertising, now it is increasingly becoming an extractive system that concentrates profits in the hands of a few ultra-wealthy, and generates little or no value for the public at large.
Apart from not making us any better off in economic terms, have start-ups improved our lives in other ways? Some things have come out of the system: we are now able to make phone calls on the internet, to stalk old girlfriends on social media and to order pizzas via apps instead of having to call up the local pizzeria. Sectors like hospitality, travel and transportation have become cheaper for consumers, while at the same time reducing revenue for producers. But in relation to what we might have imagined at the inception of the digital economy it is overall quite unimpressive. The film Blade Runner made in 1982 imagines the future year of 2019 as ripe with space travel, androids and back alley genetic engineering (but no smartphones). Facebook and Uber pale in comparison. Why has thirty years of digital innovation produced such unimpressive results?
The main reason for this is that the start-up system regards innovation as an extraneous, almost mystical variable. Innovation is a matter of ‘serendipity’ or individual genius, and not rational deliberation and collective decision-making. Often innovation consists in applying a new technology (aps, blockchains) to a number of established problems. The idea is that just because something can be done, there will also be a market for it. This mentality is expressed by the doctrine of ‘disruption’. Because new technologies will ‘change the world’ , there is no need to consider what the world wants now. Desires and needs will anyways be radically different in the future as a consequence of new technologies. Who could have predicted Facebook? The doctrine of disruption thus tends to confer legitimacy to the technological genius of the innovator, freeing him or her from the need to confront the actual complexities of social, economic and political realities. Such solutionism- ‘solving’ poverty with a combination of soylent and guaranteed minimal income- tends to put the blame on reality. Rather than a Cambrian diversity, as the Economist predicted long ago, this lack of attention to actual needs and desires leads to a technologically driven standardisation of innovation. Certain technological solutions tend to dominate ideas of ‘innovation’ in certain periods. In the 1990s as websites, then came blogs, social media, apps, sharing economy and lately blockchains. In each of these periods the creation of start-ups consists, in most cases, in the application of the technology that is fashionable at the moment, to some problem. This standardization of innovation is further reinforced by the fact that most incubators tends to invest in start-ups that replicate proven models, possibly bringing them into new sectors, an Uber for pizza, another for Marijuana and a third for artisan breads, and so on. So, paradoxically perhaps, the ‘serendipity’ of innovation- the fact that it is supposed to be driven by individual ‘disruptive’ genius, and not by a systematic analysis of actual needs and wants- leads to a process of standardization. Innovation tends to become more of the same: a series of essentially similar marginal improvements to a social model which in itself is never put into question. AI for choosing playlist for the suburban home; blockchains to facilitate ecommerce payments; data mining to produce serials that replicate qualities and features that have already proven attractive. This is reflected in the motto of the Californian start up model, that ‘ideas are cheap’. Since most start up ideas in a given period are essentially the same, other variables, like the quality of the team or their contacts count more in determining the value and potential of a start-up.
The fundamentally conservative nature of Silicon Valley- style digital innovation comes to no small extent from the fact that the direction of innovation is not politicized. This leaves the digital economy in a suspended state of what Mark Fisher called capitalist realism, a situation where it is impossible to even imagine a future different form the present, and where the power to imagine a future is in the hands of a self-appointed tech elite whose visions rarely go beyond what young boys might daydream about. This impossibility to evolve tends to make the start up system increasingly extractive and toxic. Take Amazon for example. The company employs hordes of workers, knowledge workers and logistics and warehouse workers who all work long hours, and tend to burn out quickly; it uses sophisticated technologies like data mining and artificial intelligence along with an enormous amount of server farms and data processing centres. All of this to marginally improve on an outdated 20th century consumer society, while in the process consuming enormous amounts of energy and generating massive amounts of toxic e-waste which ends up in the company’s massive landfill sites. Or take the average early stage incubator. It attracts numbers of young educated knowledge workers who, motivated by a small seed grant, tend to invest their savings along with those of their family and friends to work long hours — nobody changed the world working 40 hours a week as Elon Musk claims. All of this to generate the kinds of hype that can make their innovation stand out from the others (and in the process justify investment and public financing in the incubator that they work for). They fail in 99 per cent of the cases and then go back to the kind of corporate bullshit job that they came from, or if things go well, see their innovation be bought up, and , in most cases closed down by some corporate giant. It is almost as if the purpose of the Start-up system was that of neutralizing, rather than stimulating the massive innovative potential that has come out of the new relations of production enabled by digital technologies. Whether intentional or not, it certainly functions that way, sociologically speaking.
There is an urgent need to re-politicize innovation; to introduce forms of collective decision making able to decide what direction it should take, what future needs should be met, and how resources (like venture capital) should be spent. The Chinese startup model works this way. There digital innovation is directed by heavy state intervention in venture capital. It is based on a long-term strategy and supported by state centred policies to implement and adapt innovation. This might not be a democratic model and it might not be oriented to the kinds of values that we share, but it has become the most genuinely disruptive one, pioneering most recent real innovations in the digital economy, from digital payment systems to surveillance algorithms. Any ‘new deal’ in the West also needs to direct digital innovation onto meeting a number of needs that derive from real concerns, whether building sustainability or resilience in the face of accelerating climate change and environmental degradation, or contrasting marginalization and social isolation. It is time to make innovation a collective concern and take its direction away from the venture capitalists who, anyways, have not proven very apt at the task.
