There is a lot of talk about productivity, how to improve it and measure it, but ultimately little is said about how the business actually benefits from it or even how it eludes it. However, it is not the production of value that counts, but its capture, and on this point, things are clear: those who supply the essential components or technical infrastructure become the main beneficiaries of the value created by others.
This logic, which was theorized in the late 1990s by Carl Shapiro and Hal Varian, simply describes an economic fact: in a chain, value doesn’t stop where it’s produced but flows back to those who are difficult to bypass and, downstream, spreads to customers so that in the end, the business that created value by improving its productivity captures only part of it.
In short:
- Productivity does not guarantee the capture of created value: it is often the suppliers of critical infrastructure or components who reap the greatest benefits, upstream in the chain.
- Shapiro’s Law of Value Capture illustrates that strategic position in the value chain determines the ability to capture value, rather than efficiency or innovation.
- The ‘value chain squeeze’ describes a phenomenon where intermediate businesses see their margins reduced, caught between powerful suppliers and demanding customers.
- In the digital world, standardized infrastructure (such as that provided by Nvidia, Microsoft, and Amazon) reinforces businesses’ dependence and captures structural rents.
- To remain in control of their profitability, businesses must choose a position that allows them to capture the value produced, either by internalizing key elements or by leveraging specific expertise.
Those who dig and those who sell the shovels
There is a well-known but rarely clearly articulated phenomenon: in a value chain, particularly in the digital economy where the “winner takes it all” phenomenon creates de facto monopolies, it is not necessarily those who produce the most value who capture the largest share of it.
This idea was theorized by Carl Shapiro and Hal Varian in the late 1990s, then reformulated in subsequent works under the name Shapiro’s Law of Value Capture. It has also been summarized much more directly in this quote attributed to Jim Barksdale, former CEO of Netscape: “The only thing more profitable than selling software is selling the shovel in a gold rush“.
What this means is that money does not necessarily go to those who improve, but to those who make improvement possible by providing the infrastructure, tools, or setting de facto standards. And that these players, often invisible to the end customer, capture a growing share of the value as others work to create it.
A value chain that closes in on itself
The logic is simple. The more replaceable a business is in a value chain, the less value it can capture. Conversely, the more indispensable it is, the more leeway it has to impose its conditions.
It is not a question of efficiency but of position.
It is not the players who produce innovation or create performance who occupy the dominant positions, but those who hold what innovation and performance depend on: the lower layers, the basic tools, the constrained environments.
For example, in hardware, it is the manufacturers of strategic components; in the cloud, it is the hyperscalers; in AI, it is those who control the models or APIs. What they have in common is that they are difficult to circumvent and replace, and they charge upstream, well before the value is captured downstream.
The rest of the chain, including those who “create” products or services for the end customer, depends on this foundation. And as this foundation becomes increasingly free to impose its terms and prices over time, value capture shifts mechanically.
This phenomenon, known as the value chain squeeze, is not new. It can be observed in all industries where standardization has created strong dependencies: automotive, electronics, telecommunications, and pharmaceuticals. But digital technology is giving it a new dimension because the infrastructure in question, although heavily material, is masked by a layer of software abstraction that conceals its real economic weight and, once adopted, is very difficult to abandon.
The mirage of productivity gains
AI is often presented as a massive productivity lever, but what we see in practice is that when productivity increases, it rarely benefits those who produce. The chain is structured in such a way that the gains are redirected elsewhere.
On the one hand, some of the efficiency gains are lost internally due to poorly aligned processes or costs being transferred to other departments. On the other hand, a growing share is captured upstream: by those who sell the model, access to the API, the cloud service and, above all, by those who manufacture the critical components without which nothing works. Finally, as we have seen previously, in certain sectors and depending on the business model, you can also end up with customers who pay less, production capacity that becomes underutilized, and revenues that decline (For freelancers, productivity does not always pay off).
A business can therefore find itself caught between two opposing forces. Upstream, suppliers whose power to set market prices increases as they become indispensable, and downstream, customers who are increasingly demanding in terms of price and delivery times, with a shrinking margin in between.
It’s a value chain squeeze: a situation where players on either side of the chain grow (in size, power, and ability to extract value) faster than those who carry out the transformation in the middle. It’s not that they work better, it’s that they capture more.
In this configuration, businesses no longer really determine their own performance. Even if they improve their processes, automate their flows, or reduce their costs, a growing share of the value evaporates, captured by the ends of the chain where positions are structurally more defensible.
In a recent context that will resonate with everyone, we can only talk about Nvidia, which in just a few years has become the main indirect beneficiary of the race for generative AI. Not because the business develops agents or assistants, but because it supplies the processors, the famous GPUs that everyone else, including OpenAI, Microsoft, Google, and startups, must use. Its margin is mechanical, almost structural: it sells the shovels, and it doesn’t matter who finds the gold. And, at least as I write this, everyone dreams of being an AI gold digger.
In this configuration, every promise of productivity gains reinforces the rents of hardware infrastructure providers. The more businesses want to automate, the more computing power they consume, and the more they consume, the more Nvidia cashes in, regardless of whether this promise of productivity is fulfilled.
This mechanism is not unique to AI. It is in fact a continuation of an old phenomenon that many large businesses already know as the “Microsoft tax” or “SAP tax,” namely a recurring, structural budget item that rewards dependence rather than performance. We pay not to improve, but to continue to access what has become an implicit standard: operating systems, office suites, employee tools, the cloud, etc. Today, as AI becomes more prevalent in workflows, the same logic applies to other players such as Nvidia, OpenAI, Hugging Face, Amazon, Google, and many others who, each in their own way, are rebuilding a technical rent within a value chain where most businesses are customers, rarely partners.
This is exactly what Nicholas Carr described in Does IT Matter?: as technology becomes standardized and indispensable, it ceases to be a differentiator for those who use it, but strengthens the position of those who sell it (A software that helps to streamline processes ? Run away !).
Being at the end of the chain or choosing your place
This logic does not condemn all businesses, but it does require them to ask themselves a simple question: Am I at a point in the chain where I can capture what I produce?
To avoid ending up in the wrong place, there are several options.
The first is to move up the chain by internalizing certain key components or creating your own tools. This is costly, but sometimes necessary.
The second, more common option is to focus on mastering a business or operational context that generalists cannot easily replicate.
In other words, it means promoting not the tool itself, but the intelligent application of the tool in a given environment. This is not a technological advantage, but a knowledge advantage.
Finally, the last option is to do nothing and discover, too late, that you are producing value for others.
Bottom Line
For a long time, it was believed that performance naturally went hand in hand with profitability, that producing better, faster, and with less was enough to generate results. But in today’s value chains, the question is no longer one of efficiency, but of position.
Productivity has become a fluid, transferable resource that can be immediately exploited by those in the right place. This does not mean that all attempts at optimization are futile, but it does mean that you have to choose carefully where you position yourself in the chain. Because between those who dig and those who sell the shovels, economic history shows quite clearly which side the money is on.
Image credit: Image generated by artificial intelligence via ChatGPT (OpenAI)
Related articles:
External social networks are not only for marketing purposes
10 things I believe about the ROI of Enterprise 2.0
A socialnomics Manifesto
Cost reduction : a false good idea for organizations in tough times ?
On the place of social media in corporate strategies
Experience breaks down the walls between sectors
