Yes, we were partly thinking about the FTC complaint, as well as the broader political insistence by liberals like Elizabeth Warren that the way of going after Amazon and supporting workers is through measures to “break up” the company and “restore competitiveness.” These approaches actually rely on Friedmanite assumptions that competition is somehow inherently a good thing for workers — maximizing efficiency, employment, and consumer satisfaction through the magic of the market. In fact, such ideas also have deep roots in American progressivism, going back to the “trust busters” of the early twentieth century. The problem, as our analysis of Amazon shows, is that this is absolutely not the case: competitiveness benefits and empowers capital, reinforcing class discipline and the exploitation of labor.

Seeing how the forces of competition compel Amazon to continuously restructure, to innovate technologically, and to intensify the exploitation of labor is the only way to make sense of what the company is doing — and points to the necessity for workers to develop the disruptive capacities, at the necessary scale, to be able to exert real leverage and fight it. But competition is integral to capitalism as a system. Thus our analysis also critically highlights that the problem for workers is not just Amazon — one bad boss, or a single evil monopoly — but capitalism itself. And it is this system that workers ultimately have to build the capacity to take on — which of course implies a broader project than collective bargaining by one particular group of workers at one particular firm. Rather, this is inherently a class project.

The claim that AWS is more important than, or separable from, Amazon’s core logistics operations basically comes from two quarters. First, an element of the business press has been making the case that, since the bulk of Amazon’s profits seem to come from AWS (about 68 percent), Amazon should be split in two, allowing the more profitable business to stand alone rather than being dragged down by the significantly lower-margin retail, delivery, and warehousing side of the business. Such investors would like to get their hands on the higher profits of AWS, rather than these being used to support Amazon’s other operations.

The second place you see this argument made is among some on the Left (including Cédric Durand and Yanis Varoufakis) who are now claiming that Amazon’s involvement in the cloud sector marks a shift to a new “techno-feudal” mode of production that is distinct from capitalism. According to this view, Amazon is the paradigmatic example of a monopolistic firm that extracts rents from consumers for using space on the “cloud.” The nexus of exploitation in this techno-feudal mode of production is not production or the labor process, but rather consumption: we are all paying “rents” to these techno-feudal firms for using “intangible assets” over which they have monopoly control.

The techno-feudal argument is basically an extreme reiteration of the “monopoly capital” thesis that we take on in the paper. As we said before, this really doesn’t help us understand why Amazon does what it does. Only if we account for the intensely competitive nature of Amazon does its strategy make any sense. As we argue in the paper, competitive disciplines compel Amazon to continuously strive to compress circulation time and reduce circulation costs, as well as to produce surplus value, by intensifying the labor process and increasing labor productivity. As a result, Amazon undertakes extremely high levels of investment and R&D [research and development] spending. These dynamics in no way resemble feudalism but are absolutely typical of capitalist competition.

Moreover, these arguments miss just how integrated AWS is with Amazon’s logistics operations. For one thing, AWS emerged from the competitive drive to enhance the efficiency of commodity circulation. The firm then commodified and sold these capacities as standalone services, and the high returns and low marginal costs involved incentivized them to further expand these capacities. But you can’t just sever AWS from the rest of the firm without potentially damaging its overall competitiveness. AWS still helps coordinate Amazon’s highly advanced and flexible logistics network. And of course, Amazon leverages AWS profits to support its extremely high levels of investment in its retail and logistics operations, which make up nearly 85 percent of its net sales.

In fact, this integration is becoming even more important with time. Far from being the basis for a secure monopoly, Amazon is facing growing competition in the cloud sector from Microsoft, Google, and others. As a result, one would expect that margins in this sector will start to come down with time. Amazon’s strategy for dealing with this has been to leverage its advanced logistics to maintain its leadership in the cloud sector, enticing firms using its cloud services to use its logistics as well: If you like the cloud, why don’t you also let Amazon manage the back-end of your business? Indeed, Amazon is currently positioning itself to offer supply-chain management and third-party logistics services to other large firms. This means that Amazon’s strategy for maintaining its edge in the cloud sector is completely bound up with its ability to sustain the competitiveness of its logistics.

All this matters because it allows us to understand what exactly Amazon is doing and why it is doing it. But also, and more importantly, it points to the centrality of labor exploitation to Amazon’s success — and therefore to the capacity for the workers whose labor is chiefly responsible for Amazon’s profits and competitiveness to exert power. If AWS can’t be separated from the logistics and warehousing operations, then the firm can’t just divest from the latter as a result of workers’ struggles for better wages, working conditions, and shop floor power. Workers could therefore potentially exert real leverage.

What economists call the cash conversion cycle, or CCC, is an important part of Amazon’s market strategy. Basically, the CCC measures the difference between the time, in days, that it takes a firm to sell inventory and receive payment, on the one hand, and to pay its bills to suppliers, on the other. The lower the CCC, the longer a firm holds on to cash after selling its inventory and receiving payment, but before paying its bills.

A firm can thus reduce its CCC by getting payment faster, selling inventory faster, delaying payment to its suppliers, or some combination of these. Typically, a low CCC is interpreted as a sign of a company’s strength and efficiency, since this translates into a strong cash position that helps it to pay its obligations, fund its operations, and pay down debt. This is also an indication of a firm’s power over its suppliers and sellers, since it means it can delay or demand payment.

Amazon has among the lowest CCCs of any major corporation. It has achieved this by maximizing the efficiency of its logistics system, selling inventory and collecting payment extremely quickly. This rapid turnover has helped attract a large number of third-party sellers to sell on Amazon, which make up a large and growing part of its business. Amazon’s extremely efficient logistics and the reach of its e-commerce platform has allowed these sellers to benefit from selling on Amazon even though it delays payments it owes them for a considerable period of time (and charges fees). Amazon must therefore sell products very quickly, which allows it to delay payment to its sellers for a significant period while still offering these firms a competitive advantage by selling on Amazon.

In fact, at year-end 2023, Amazon posted a CCC of -14.7, as compared to the retail sector average of +44.9. In the same period, Amazon’s main competitor, Walmart, posted a CCC of +8.7. This allows Amazon to acquire as much as $30 billion in interest-free cash annually, maintaining a very strong cash position while displacing the debt and interest load on to smaller and weaker supplier firms. This supports Amazon’s high share price, as one of five $2 trillion firms, without having ever paid a dividend nor conducting hardly any stock buybacks. However, there are signs that this significant advantage is coming under increasing competitive pressure, driving Amazon to further intensify work discipline and maintain high levels of investment to continuously improve productivity through automation and the de-skilling of labor. At the same time, it points to Amazon’s possible vulnerability to organized job actions such as slowdowns, if these could be undertaken at a sufficient scale.

That’s a great question. Without having researched Walmart in depth, a few things come to mind. Of course, Walmart has an advanced logistics system, as it must in order to survive amid the cut-throat competition in these sectors. But it is set up a bit differently than Amazon. Walmart negotiates high-volume contracts with a smaller number of larger suppliers, receiving discounts via bulk purchases. Amazon does this too in relation to its “first-party” suppliers, but what is different about Amazon is its substantial reliance on a large number of smaller “third-party” sellers.

Rather than negotiating favorable wholesale prices, Amazon charges these sellers fees to use its system — somewhat akin to a toll road, although with the very important caveat that Amazon also adds value to commodities it purchases. Amazon has thus deeply commodified its logistics operation in that it makes its system and platform available to any firm willing to pay the fees it charges for these services.

Thus instead of a scale-based model in which it gets discounts from large sellers for large sales volume, Amazon relies much more on the extraction of fees by virtue of its control and competitive operation of an e-commerce logistics system. Because its third-party sellers are fragmented, while Amazon is centralized, Amazon also gains more leverage over these firms and is able to capture a larger proportion of their total revenues in the form of seller fees — sometimes as much as 50 percent — as well as to gain interest-free cash by delaying payment to them.

As we argue in the paper, these third-party sellers are a distinct type of business that specializes solely in selling on Amazon. Amazon therefore effectively centralizes these capitals by absorbing them within its overall structure, continually funneling value into Amazon, while sellers benefit from their ability to profit “independently” from using Amazon’s highly competitive system. In fact, the fees and conditions Amazon imposes on these sellers is nearly sufficient to finance all new investment as well as its day-to-day operations.

Exactly. Like all firms, Amazon is competing with other firms in various sectors across the economy over market share, but it is also locked in a competitive struggle with its customers and suppliers to capture the maximum quantity of surplus value within value chains. It wants to acquire the largest possible amount of surplus by forcing suppliers to accept the lowest possible prices (and other conditions) while charging customers the highest prices it can get away with. But its ability to raise prices is limited by the intensely competitive nature of these markets. Amazon has to attract supplier firms by offering them the opportunity to reduce their costs, compress turnover time, and reach the widest markets. Meanwhile, it has to attract customers by offering them fast delivery times and low prices. This means Amazon is under incredible competitive discipline to maximize efficiency, intensify work, and exploit labor.

In order to pull this off, Amazon undertakes extremely high levels of investment in its warehousing and logistics system. The competitiveness of Amazon, including AWS, hinges on the efficiency of its warehouses, where thousands of workers are employed laboring together under conditions of bitter exploitation, close work discipline, and intense surveillance. In many respects, it looks like a scene out of Capital, Volume I.

As you mention, these warehouses are hubs within Amazon’s regionalized logistics network: 76 percent of orders are shipped from an FC in the customer’s region. This means that although Amazon operates on a national scale, its operations are segmented at the regional level. While the intense competitive pressure Amazon is under makes it vulnerable to disruption, the regional structure of its system potentially allows workers to exert significant leverage by organizing regionally. And this points to the centrality of the warehouses, which serve as regional hubs.

Of course, the scale of these facilities, and the level of surveillance and discipline workers are under, as well as the high level of turnover and other factors, mean that organizing there is definitely very challenging. As others have pointed out, smaller facilities such as delivery stations and the “last mile” may be easier to organize with the relatively small resources available to us. And certainly this could play a role. But questions of strategy have to come back to how the company will react.

Amazon has a relatively large amount of flexibility to respond to disruption in the smaller delivery stations through uberization and subcontracting delivery to other carriers, rerouting flows of commodities through other stations, or even moving delivery stations. If organizing efforts are going to focus on delivery, we need a plan to respond to such moves. Disruption at the warehouses, on the other hand, is much harder to get around. The assumption often seems to be that organizing delivery can spill over to the warehouses, but this faces significant challenges that would also need to be addressed.

As we argue, “organizing Amazon” seems to ultimately mean organizing the warehouses that are the hubs of its system. But this demands a strategy that combines disruption and scale. Given the size of the warehouses, building power there requires broad and deep organizing involving a relatively large number of workers. And to be effective, these efforts cannot just be limited to a single warehouse but need to take place at a regional scale. But scale alone is not enough. Amazon’s investment strategy affords it substantial surplus capacity, and therefore flexibility in the face of challenges or blockages. This means that the traditional business unionism model, whereby union strategy is limited to collective bargaining over contracts at regular intervals, is unlikely to be sufficient for exerting real leverage over the firm. The flexibility of Amazon’s model, along with its vulnerability in the context of the intense competition it faces, means that building power requires that workers develop equally dynamic capacities for disruption.

Amazon definitely has lots of flexibility in its system, especially during nonpeak times. Yet Amazon’s surplus capacity is not infinite — there are limits to what the system can accommodate. This means it is important to identify where and when they have the least ability to circumvent and undermine any disruptive power. For starters, Amazon’s logistics system has both centralized and dispersed components. The FCs (and IXDs [inbound cross docks]) are relatively centralized. They are hubs from which commodities are dispersed as they flow out to a larger number of delivery stations. Thus, FCs offer the opportunity to exert leverage over the network of more fragmented facilities they supply. Focusing on the latter, smaller facilities would mean that a larger number of independent facilities would have to be organized in order to exert the power that could result from shutting down warehouses.

The surplus capacity Amazon maintains points to the need for a strategy oriented around both disruption and scale. In terms of scale, it is clear for the reasons you suggest that organizing a single warehouse is not enough. As we mentioned earlier, while Amazon’s system is national, it is regionally segmented. This points to the possibility of building leverage within particular regional segments of the system — but it also means that this organizing has to take place on at least a regional scale. At the same time, there needs to be a focus on building disruptive capacity. The typical business unionism model won’t work because it gives Amazon time to prepare, and thus to undermine any actions. Taking on Amazon therefore requires a fundamentally different model than that of the big unions.

If the question is how to maximize workers’ leverage over Amazon, then we are talking about how to impose costs on it. This requires an analysis of how its operations are organized, which gives us a sense of the potential impact of various strategies. The regionalization of Amazon’s logistics network suggests that maximizing power requires worker action on at least a regional level. Forcing the firm to “forgo” this structure, as you say, would come at a significant cost, particularly given the fiercely competitive nature of this sector, as we point out in the paper. Such restructuring would impose near-term costs and impact Amazon’s ability to accelerate commodity circulation. Indeed, this competitive drive to compress space and time is at the core of its competitiveness and led to the development of the regionalized structure in the first place. To us, this points to the strategic importance of the FCs, as the central nodes in each region.

In terms of the Teamsters efforts, it is of course great to see the labor movement getting involved in organizing Amazon, which is rapidly becoming the largest private employer in the US. Organizing Amazon is a generational project that presents the opportunity to actually impact the entire working class, as well as to revive the labor movement. But as we pointed out earlier, the traditional “business unionism” model is unlikely to succeed in this case, particularly given Amazon’s incredible flexibility.

Thus, while taking on Amazon offers the chance to revitalize the labor movement, which is currently in a deep crisis, at the same time it challenges the labor movement to change. This is unlikely to come from the top down, but will require rank-and-file activists to challenge existing strategic orientations and institutional hierarchies. In addition to flexible disruptive capacity, the competitive imperative to accelerate circulation, for example, suggests that building workers’ power at Amazon demands a focus not only on wage bargaining, but also incorporating demands for greater worker control at the shop-floor level.

Getting into this fight in a serious way requires the unions to go all in and calls for the participation of the labor movement as a whole. It is not enough to focus on delivery stations, which, as we noted before, are small and relatively easy to route around, relocate, or even subcontract or uberize. Organizing Amazon demands a far larger commitment of time and resources to build power where the workers are — especially in the warehouses. Frankly, the recent Teamsters “strike” against Amazon, in which probably under 1 percent of Amazon’s US blue-collar workforce participated, wasn’t all that encouraging; it came from the top, was premature, and in general did not indicate the commitment and willingness to change that is needed. A regional strategy, initiated from the bottom, via a series of rotating actions in places where the union was strong (for example, starting in California and moving on to another region shortly thereafter) may have been more effective.

It seems to us that there is a real need for sober reflection on these shortcomings and issues, as this is the only way we can move forward with the critical project for rebuilding the power of the working class. If we want the labor movement to advance, we simply can’t shy away from the hard questions.