Platform Capitalism

interviewed by Norbert Morvan

edited by Matthew Phillips

On April 28th, the International Strategy Center and Progressives for All of the Justice Party invited Nick Srnicek, lecturer at King’s College London and author of Platform Capitalism to its monthly Progressive Forum to talk about the political-economic of platforms. The interview was conducted by Norbert Morvan and edited for clarity and brevity by Matthew Phillips.

Your book provides the historical trajectory that brought us to today’s platform capitalism. In particular, you state that the 1970s saw a shift towards lean production models and precarious work due to increased competition in manufacturing; the 1990s saw a dot com boom and bust that laid today’s technological infrastructure and loose monetary policy; with the post 2008 financial crisis continuing this loose monetary policy. Up to recently, governments focused on austerity on the one hand and “asset price Keynesianism” on the other to create wealth through stock bubbles, especially that of platform companies. What made this “asset-price Keynesianism” the preferred method of funding recovery after economic crises?  

It's part of the long rise of neoliberalism and particularly the increasing dominance of the financial sector over the manufacturing sector, which had dominated in the postwar period. This dominance […starts] to shape policy-making and government actions. So there are efforts to keep stock prices and other financial assets rising, all of this to keep the financial sector happy. And it was often at the expense, the competitiveness, particularly global competitiveness, of manufacturing. Now, it also came about because the alternative response to a crisis, which would be things like government stimulus, more spending, [all these have] been deemed ideologically impossible until very recently. So there's been consistent fear-mongering over government deficits throughout the neoliberal period [i.e. over the past 40-50 years] and this has led to governments preferring other options rather than spending money. This is particularly the case after 2008 [with] the US, the UK and especially the EU. They all united in a consensus about using monetary policy to respond to crises instead of [using] fiscal policy, despite many economists basically saying what was needed was active fiscal policy.

So what are the long term impacts of this approach to dealing with crises in the economy?

Okay so, basically [these responses have resulted in the fueling of] speculative bubbles. So this effort, this [type of] monetary policy, has brought about cheap money through low interest rates, and through things like quantitative easing has meant a surplus of capital which has been seeking to find a decent return on investment. So you can't just park these things in government bonds: you have to actively go out and search [for] more risky ventures, which has fueled these bubbles. I think there is an open question now about how many of the advanced economies are actually dependent and reliant upon this easy monetary policy. So the option seems to be either we accept secular stagnation and the near zero growth that comes with it, or we accept the repeated inflation of asset bubbles. And that seems to be the situation that a lot of advanced economies are in right now.

Recently the Fed increased interest rates to stave off inflation. And furthermore, US President Biden appears to be trying to onshore manufacturing. How do both of these trends affect the future of asset price Keynesianism?

It's a potentially momentous change, this move in America to higher interest rates, into much more active fiscal interventions through things like the IRA [the Inflation Reduction Act of 2022] and the chip subsidies and all these sorts of things – things which haven't been really all that significant in US policy for at least 30 years. So it reverses this [focus on] asset-price Keynesianism, the focus on monetary policy, to a new focus on fiscal policy as a response to economic challenges. 

[As regards] reshoring: I don't think it's going to change all that much because the way in which manufacturing is coming back to America and coming back to, you know, the rich world, it's only coming back in a highly automated form. It tends to benefit the owners of capital rather than having any benefits broadly. And crucially also, it doesn't increase their power with respect to finance. Now I think there are still open questions about how successful the new US approach is going to be in terms of generating growth, especially in the medium term. 

In platform capitalism you mentioned that the expansion and extraction of data from platform users is a fundamental driving force of platform capitalism. In other words, Facebook doesn't violate your privacy because of Zuckerberg, but because of capitalism. What impact will recent breakthroughs in AI such as Chat GPT have on these tendencies [ie of fragmentation and monopolization]?

So I think data is still crucially important for the platform economy, but you have the situation where the upper echelons, the platform economy are reaching a stage where all of these companies have immense and competitive quantities of data. So you can look at, for instance, recent AI training, whether it be from Microsoft, whether it be from Google or whether it be from any other company. It's a huge amount of data which is being thrown into these models.

So I think AI certainly drives the search for more data. And one of the key principles that has been found in research and in practice in recent years is that more data makes better AI models. And that's particularly when it comes to the new generative AI models. So we see increasingly large datasets being thrown into these systems.

Now at the same time, to train AI requires immense amounts of computing power. So it's not [only] a matter of data, it's also this hardware aspect. So GPT-4, for instance, is said to have cost more than $100 million to build, and a vast amount of that is because of the computing hardware that's been necessary for it. Between 2012 and 2019, there was a 300,000 times increase in the amount of computing power needed to train the largest models, which is a doubling of the computing power needed, every 3 to 4 months.

All of this has made ownership of and access to computing power increasingly important. And I think it's cloud computing companies which are positioned to be able to benefit from this reliance upon hardware. The shift to cloud computing is an expression of this change. It's datacenter-scale computing that's required to do AI today and GPUs [Graphics Processing Units, a vital component to much AI work in 2022-2023] that you own at one point.

So computing resources are in turn an expression of financial resources. Datacentres cost tens or hundreds of millions of dollars and all of this is only available to the top tier technology companies. So their control and ownership over hardware, not just data, but hardware, means they're therefore capable of setting the terms of access and development of AI and the future of the digital economy.

So a huge amount of power is being concentrated in their hands and I think this focus on hardware is one of the big shifts I've made in my own thinking since Platform Capitalism.

In the book you state that overcoming the tendencies towards a monopoly of these platforms isn't possible within a capitalist framework, possibly implying the need for Platform Socialism. Also, you mention the potential of governments to harness their vast resources to create platforms that serve the public good. Do you have any examples of such platforms, and if not, what might they look like?

It's surprisingly difficult to find examples. [But] I think one of the best examples though is a sort of a ride sharing platform called Ride Austin, based out of Austin, Texas. And it's interesting because it's an Uber competitor. And effectively what happened was Uber and Lyft, major companies, were forced out of the city after they failed to meet local regulations. And this left a sort of a hole in the ride sharing platform business. And what happened in that city was that there was a number of nonprofits which rose up to replace the old companies.

So one of the nice things about right Austin was they also made completely public the process of doing this so you can go and find information on exactly what they did and how they did it. They built an app in four weeks with only six people. They aimed to have a fast rollout, so they got a product out to the market as soon as possible and then they constantly updated it. So within eight months they had something like 60 releases of all the different versions of the software.

Now in terms of financing this, they simply use donations from the local tech community. So it wasn't like some billionaire came along and gave them a ton of cash. They could just get community funding. They used third-party services: So Amazon again. But also services for payments and things like that. And these costs tended to decrease in price as the app grew larger, but other costs did go up. So dealing with fraud as it became a bigger target, things like that started increasing their costs.

Now the benefits of this approach was that they managed to provide a cheap service, so it was pretty competitive with Uber and Lyft, but they also paid workers much better than what Uber and Lyft were paying. They also donated money to local charities, so the money stayed within the local economy rather than being siphoned off to Silicon Valley.

The challenge for Ride Austin was that Uber and Lyft were eventually let back in the city. Uber and Lyft lobbied state regulators to change laws to get back into the city. Now, as a result, the use of these nonprofit apps took a significant hit. So one basically immediately dropped out. Another saw a 20% decrease and another app saw a 50% decrease in one week.
Basically, Uber and Lyft came and used their venture capital funding to immensely subsidize the costs. So really cheap rides and they sort of temporarily jacked up the wages for drivers as well. So Ride Austin ended up getting driven out of business by these large companies.

And my final question: what prevents the creation of municipally run versions of platforms like Airbnb all over the world?

There's a sort of belief that what Silicon Valley does can't be replicated by a government and the idea that you could simply build a platform seems absurd to a lot of politicians, like “go and build an Uber?! No, that would be too difficult!” The thing I love about Ride Austin is that it shows that it's entirely possible six people in four months with community funding did it. Why could London not do it? Why could New York not do it? Why could Berlin not do it? You know, it's entirely possible.

But the other reason why is that there's a lack of political strength to take on these platform monopolies [as represented by] these large companies. So Ride Austin shows the challenges that face these alternatives: if they have to compete with these massive companies, which are subsidized in the billions by venture capital, they stand no chance.

There's no way that a locally-funded alternative is going to succeed against a $1,000,000,000 company. So if a government really wants to build an alternative, they also need to put significant restrictions on the large platforms. And I don't think most governments are willing to do that.