Learn faster. Dig deeper. See below.
“The economic problem of society…is the problem of using knowledge that is not given to anyone in its entirety.”
—Friedrich A. Hayek, “Use of knowledge in society“
Venture capitalists and many Silicon Valley entrepreneurs espouse libertarian values. In practice, they subscribe to central planning: Rather than competing to win the market, entrepreneurs compete for funding from Silicon Valley’s equivalent of the Central Committee. The race to the top is no longer driven by who has the best product or the best business model, but by who has the blessing of the deepest-pocketed venture capitalists—the blessing that allows them to acquire the most customers the fastest. , often by providing services below cost. Reid Hoffman called this formula “blitzscaling” and claimed in the subtitle of his book of the same name that it is the “Lightning Fast Path to Building Massively Valuable Companies.”
I don’t agree. It’s a dark pattern, more a map of suboptimal outcomes than a real path to competition, innovation, and the creation of strong companies and markets. As Bill Janeway noted in his critique of the capital-fueled bubbles that resulted from the extremely low interest rate decades following the 2007-2009 financial crisis, “capital is not a strategy.”
Venture capitalists don’t have a crystal ball. To the extent that business finance is more concentrated in the hands of a few, private finance can drive markets independent of consumer preferences and supply dynamics. Market discipline is significantly delayed – until the initial public offering or later. And of course the IPO today they are belatedly, often precisely because companies can raise all the capital they need from a small number of deep-pocketed investors. Founders and employees are even able to redeem some of their shares without facing the scrutiny of the public markets, much like bettors on a horse race can take their money off the table when the horses complete the first lap. Thus, far from being an extension of the market (with lots of independent signals aggregated to ensure competition and consumer choice), capital can ignore the will of the market.
The ride-hailing business offers a classic example of a disruptive over-reliance on capital rather than consumer choice. It began with bold prophecies of ride-hailing replacing not just taxis but all private vehicles, and ended with a national duopoly of on-demand taxis at prices no better and often worse than those of the previous over-regulated local taxi market. In a well-functioning market, many startups would explore a technological innovation like on-demand transportation for a much longer period of time. In this alternate history, entrepreneurs would compete with different pricing strategies, different rate structures for drivers, and possibly completely different business models. The survivors would eventually do so because they provided the service chosen by the most customers and the most drivers. This is true product market fit.
But in Silicon Valley’s version of the Central Committee, Uber and Lyft, backed by trillions of dollars in venture capital, pushed the competition out rather than defeated it, subsidizing customer acquisition and an unsustainable business model—and, in Uber’s case, continuing to attract new capital with promises of speculative future cost savings through self-driving cars. Instead, once the market consolidated, Uber and Lyft achieved profitability only through massive price increases. What could happen if there was real competition in this market? We’ll never know.
In contrast, during the dot-com bubble, most companies consumed a tiny amount of capital by today’s standards. Funding was spread among thousands of companies, and it took a decade or more of relentless innovation and competition for the industry to become dangerously concentrated. This is a classic example of what Janeway calls a “productivity bubble.” Remarkably, most of the winning companies were profitable within just a few years and eventually became very profitable. Google only raised $36 million in venture capital on its way to dominance. Facebook raised billions, but did so only to fund faster growth of a business model that insiders told me was very close to profitability all along. They didn’t buy users at subsidized prices; they were building data centers. Even Amazon, long unprofitable, took very little venture capital, instead financing itself with a debt-backed business model that produced previously unprecedented levels of free cash flow.
Sure, sometimes companies require a lot of capital to lay the foundation for a possible future. Tesla and SpaceX are good examples. They used their finances for serious research and development, building factories, cars, batteries, rockets and satellites. That is the proper use of capital: to finance the heavy costs of creating something new until the projected unit economics lead to a self-sustaining business. It’s also worth noting that in these cases, private funding has been significantly augmented by government support: carbon credits and EV incentives for Tesla, and progress payments from NASA for SpaceX.
Such an investment was unnecessary in the case of riding a horse. Startups simply used the money to accumulate market power by subsidizing flash growth. Others have already deployed the capital to build much of the infrastructure for driving—GPS satellites and GPS-enabled smartphones. Even the innovation of using GPS to find passengers and drivers was not developed by VC-backed market leaders, but by the true market pioneer, Sidecar, which was quickly sidelined when it failed to raise enough capital to gain a leading market share. that’s what he imagined at first.
In the case of artificial intelligence, training large models is really expensive and requires a large capital investment. But these investments require proportionately high returns. Investors who pile billions of dollars into a huge bet expect not just a return, but a hundredfold return. The capital race to build the biggest models has already led to bad behavior. For example, OpenAI trained not only on publicly available data, but also allegedly on copyrighted content obtained from pirate sites. This led to lawsuits and settlements. But even these settlements are likely to be bad for the development of a healthy business ecosystem. As Mike Loukides points out, “Smaller startups…will be charged along with every open source effort. By settling, OpenAI will eliminate much of their competition.”
Meanwhile, the absorption of all content for the largest models into “Borg” AI data eliminates opportunities for owners of specialized content repositories to profit from their own work. Innovators are already finding that much can be done at lower cost with smaller, more focused open source models. They can tune these smaller models for specific problem domains and allow trusted content providers (like my own company’s O’Reilly Answers and related AI-generated services) to profit from our own expertise.
OpenAI aims to create a platform on which entrepreneurs can build vertical applications, but only if they pay tribute to a centralized business model in the form of API fees. OpenAI is also picking up the cream and quickly dominating some of the most profitable categories—image generation, video generation, speech synthesis, computer programming—that in a well-functioning market would be explored by dozens or hundreds of competing attempts until one or two find a winning combination of product and business model . If entrepreneurs discover other profitable categories, giants like OpenAI will move quickly to dominate those as well.
The capital-driven AI land grab is, of course, only one axis of premature market concentration. As Max von Thun points out in “Monopoly Power Is the Elephant in the Room in the AI Debate”, much of the investment in model training comes in the form of strategic partnerships (including cloud computing credits and potential returns). with existing industry giants Microsoft, Amazon and Google (and in the case of open source models Meta Platforms). As von Thun notes, “These partnerships appear to serve the same purpose as ‘killer acquisitions’ in the past—think Facebook’s acquisition of WhatsApp or Google’s purchase of YouTube—raising serious concerns about fair competition in the nascent AI market .” Again, the risk of these deals is that a few centrally-selected winners emerge quickly, meaning there is a shorter and less robust period of experimentation.
And at least based on The Information’s recent reports on Anthropic’s operating margins, it may be that, like Uber and Lyft, the overfunded AI market leaders may only be able to live up to investors’ excited expectations by crushing all the competition. This is not a bet on the wisdom of the market and what Hayek called “the use of knowledge that is not given to anyone in its entirety.” This is a bet on premature consolidation and the wisdom of a few big investors to choose a future in which everyone else will be forced to live.