The short book falls into three parts. There is a context-setting chapter on ‘The long downturn,’ which will surprise no-one who has read, for example, Wolfgang Streeck’s book and articles on the end of capitalism. The second chapter analyses different types of platforms and their strengths and weaknesses. A final chapter, ‘Great platform wars’, looks at the prospects.

Official futures

One of the strengths of this approach is that it provides some valuable context. One early observation is that the tech sector remains small by value and even more by employment. In the US it represents 7% of the value added by private companies, and 2.5% of the employment. However, it is also a pervasive technology, both systemically and ideologically.

The ‘official future’ (my phrase) of the digital economy, as he observes, is that

cities are to become smart, businesses must be disrupted, workers are to become flexible, and governments are to become lean and intelligent. (p5)

The critical view, in contrast:

[W]ith a long decline in manufacturing profitability, capitalism has turned to data as one way to maintain economic growth and vitality in the face of a struggling production sector. (p6]

Platforms as data engines

This is where platforms come in. If data has become a massive new raw material for capitalism, then platforms are the engines that allow it to process this data. Srnicek identifies five types of platform (p49), described below. For me this is a valuable piece of analysis, since it allows greater clarity about their differences and their strategies, rather than bundling them together as if they are a single phenomenon. What they do have in common, though, is that “collecting massive amounts of data is central to the business model.” (p89)

• Advertising platforms (e.g. Google, Facebook), which extract information from on users, analyse it, and use the outcome of that process to sell advetising
• Cloud platforms (e.g. AWS, Salesforce), which own the hardware and software used by digitally-dependent businesses and rent them as needed
• Industrial platforms (e.g. GE, Siemens), which build the hardware and software necessary to transfer traditional manufacturing into internet-connected processes
• Product platforms (e.g. Rolls-Royce, Spotify), which generate revenue by transferring a traditional good into a service
• Lean platforms (e.g. Airbnb, Uber), which seek to connect buyers and sellers of a service while maintaining a minimum of assets. (I have adapted Srnicek’s definition slightly here to emphasise the *market-making* nature of such platforms.)

Questions of scale

For the detail of these you’ll need to turn to the book, but pulling them apart like this helps understand their differences. Advertising platforms have been hugely profitable, to the point where they don’t know what to do with their cash, but are running out of road in terms of growth and monetisation. Cloud platforms generate good returns (AWS is by far the most profitable part of Amazon) but require significant scale and investment, as well as robustness. Industrial platforms are specialist versions of cloud platforms; they need scale, and to be able to handle prodigious amounts of data. (Srnicek reports that GE’s natural gas platform alone is collecting as much data as Facebook.)

Product platforms, perhaps better thought of a service platforms, own the asset that they lease to the end user (this is the difference between Zipcar and Uber), although in the case of Spotify that “ownership” takes the form of a licence from the underlying rights holders, which is a problem for its business model. There’s a useful discussion here of the aero-engine market, now moved more or less completely to a service model: effectively this allows the engine companies to integrate higher margin maintenance with the lower margin product by connecting them through their data.

Venture capitalist welfare

Finally, lean platforms need to own the relationship with both buyers and sellers. They are lean because they outsource everything else. “It’s hard,” writes Srnicek, “not to regard the new lean platforms as a retrogression to the earliest stages of the internet-enabled economy.” Or further, I’d say, to the piecework of early industrial capitalism (pdf). As Srnicek notes, “the most notorious part of these firms is their outsourcing of workers.” (p75,76).

It’s worth taking a small detour here on lean platforms, party because Srnicek suggests that despite the hype about them, not to mention that tedious conference trope, which was apparently created by IBM, they are the most minor of these five.

The profit-making capacity of most lean models likewise appears to be minimal and limited to a few specialised tasks. And even there the most successful of the lean models has been supported by VC [venture capitalist] welfare rather than any meaningful revenue generation…. these models seem likely to fall apart in the coming years. (p87-88)

Building on this, Airbnb appears sustainable, because it can charge a larger price per unit than the other two, and therefore take a larger cut itself, the units have lower running costs, and often do represent genuinely marginal capacity, and the balance of power between seller and platform is more equal. Uber, on the other hand, has relied completely on “VC welfare” to undercut conventional taxi prices—to the point where the scale of its price advantage matches almost exactly the losses per trip funded from its reserves. I’m always surprised that competition authorities haven’t paid closer attention to this aspect of Uber’s business model.

Platforms need to expand

In the final chapter, he uses an analysis of the ‘tendencies’ within platform capitalism to identify some of the long-run trends. Since platforms are based on the extraction of data and the creation of network effects, they also have certain dynamics: “expansion of extraction, positioning as a gatekeeper, convergence of markets, and enclosure of ecosystems.” (p98)

Taking these in turn, platforms need to expand because they need more data. Because they need to expand, they tend to encroach on the same markets and the same types of data. One can see this most clearly in the competition between Google and Facebook, for example. Srnicek asks whether this tendency goes beyond competitors around similar platform spaces (e.g. Amazon and Alibaba) to platforms in general. I suspect this is unlikely. The skills and competencies needed to run an industrial platform like GE’s Predix, not to mention the customers and sales channels, seem to me to be fundamentally different to a consumer platform like Spotify or an advertising platform like Google.

Locking in customers

The final tendency, about the enclosure of ecosystems, is because the acquisition of data does not, finally, ensure a relationship with the customer, and so they have to be locked in, as much as this is possible without irritating them. Facebook has been effective as this; Apple too. In the industrial space, it is even more acute; it is possible to migrate your sales and customer data from Salesforce, but it is not something you would do lightly. Apps reinforce this tendency towards closed systems.

There are challenges here. Srnicek mentions the limits of outsourcing, a particular problem for lean platforms, the dependence on loose finance capital, and, for advertising platforms, a dependence on advertising growth. Platforms do continue to grow, but to do this, they have “to extract rents by providing services.” On this basis, “Amazon is more the future than Google, Facebook, or Uber.” (p126)

What is to be done? Srnicek suggests that in the short term national governments (and by extension, cities) might take more interest in platforms and their anti-social side effects:

Antitrust cases can break up monopolies, local regulations can impede or even ban exploitative lean platforms, government agencies can impose new privacy controls, and co-ordinated actions on tax avoidance can draw capital back into public hands. (p127)

Beyond that, there might be scope for publicly owned and publicly controlled platforms, creating new public utilities. There are co-operative platforms out there already.

Reaching limits

My own take on this is that while there should certainly be more controls on what platforms do, they might, nonetheless, be reaching their limits. Given that they need data to fuel their expansion, this comes from two places: growth in the number of customers, and second, growth in how much you can get those customers to pay for (monetisation).

In most markets, growth in customer numbers is plateauing, or close to it, in both consumer and B2B markets. And in a world of long-run slow economic growth, the prospects for further monetisation are slowing too. This may lead to some unpleasant effects as they try to maintain growth by gouging customers or playing faster or looser with their data. This also suggests that platforms that do not already have a sustainable business model are unlikely to get to one.

This means, in turn, that the loose finance capital (e.g. venture capital money) that has been keeping such platforms afloat are likely to cut their losses and move on to other areas of investment, probably not even in the digital sector.

Platform Capitalism is published by Polity Press. The image at the top of the post is by Andrew Curry, and is published here under a Creative Commons licence.