Why The Future Favors Systems Over Products
Dr. Aditya V Kashyap, AI and Innovation Leader, driving enterprise transformations through trusted strategy, governance and bold leadership.
Visa does not issue most credit cards. It does not lend money, set interest rates or generally carry consumer credit risk. Banks and other card issuers do. Yet Visa earned $40 billion in net revenue on $14.2 trillion in payments volume, connecting 4.9 billion payment credentials to more than 175 million merchant locations. Visa keeps only a sliver of each transaction, but it designed the network architecture, the standards and the rules under which the participants transact, and that coordinating position has proved more durable than almost any product moving across it.
This asymmetry, in which the designer of a system often captures disproportionate value relative to the makers of the products inside it, is becoming one of the defining patterns of modern competition. Most executives sense it, but fewer have named it, and it reframes nearly every strategic question.
For most of the twentieth century, the product was the strategy. Henry Ford drove the price of a Model T from $850 in 1908 to $260 by 1924, and his advantage, notably, was already partly systemic: the moving assembly line and interchangeable parts formed a production system competitors spent years copying. But it was an internal system, owned and operated inside one firm to make one product cheaper. Value creation and value capture still lived in the same place, inside the product and the factory that built it.
Software loosened that unity. Once products became programmable and connected, they stopped being endpoints and became components. A smartphone is a slab of glass until an application ecosystem animates it. A graphics processor is an expensive calculator until software gives it something worth calculating. The decisive systems are increasingly external rather than internal: shared platforms, standards and infrastructure that thousands of other firms voluntarily build upon. The economically important question shifted from "how good is the product?" to "whose architecture does the product depend on, and who governs it?"
Consider how that dependence compounds. Nvidia introduced CUDA, its parallel computing platform, in 2006, years before the AI boom vindicated the investment. Nvidia's own annual report counts more than 5.9 million developers worldwide using CUDA and its related software tools. Every optimized library, every dissertation and every university course adds to a stock of complementary investment made by Nvidia's customers, not by Nvidia. That is the mechanism behind the switching costs. A rival can design a faster chip; it cannot quickly replicate two decades of other people's accumulated code and expertise. The chip is a product and can be benchmarked. The architecture around it is far harder to displace.
Apple illustrates a different mechanism: governance over participation. The company reported that its App Store ecosystem facilitated over $1.4 trillion in developer billings and sales in 2025, with no commission paid to Apple on more than 90% of that figure. Critics fairly note that "facilitated" is a generous word, since the total includes groceries ordered and rides hailed through apps. But the accounting choice is itself instructive. Apple does not need a share of every transaction. It sets the terms of entry, review and distribution for an economy of other firms' products, and profits from owning the gate rather than the goods.
Amazon shows a third mechanism: asymmetric profit capture through infrastructure. The retail operation is the famous product. Yet in 2025, Amazon Web Services generated $128.7 billion in sales, roughly 18% of the company's $716.9 billion total, while producing $45.6 billion of its approximately $80 billion in operating income. The store sells products at retail margins. The cloud rents the infrastructure on which hundreds of thousands of other companies run their products at infrastructure margins. Guess which one carries the enterprise.
I have come to think of this as the architectural advantage. Products compete within markets. Architectures shape how markets operate: what interoperates with what, who may participate and on what terms, where data accumulates and how incentives steer every participant's behavior. Network effects, switching costs and bargaining power in these businesses are less features that clever managers bolt on than properties of the architecture itself. The Model T's edge depreciated with every competitor's factory upgrade. Visa's tends to appreciate with every new participant, because each one raises the cost, for all the others, of leaving.
An important caveat: this is a structural tendency, not an iron law. Excellent products remain necessary; they are increasingly insufficient on their own. Every system described here has a superb product at its core, and the causality often runs from product to system: the App Store exists because the iPhone succeeded, not the reverse.
Nor does the system owner capture everything; most of the value in these ecosystems accrues to participants, which is precisely why they stay. And architectural power invites architectural scrutiny. Apple regularly publishes ecosystem studies as regulators and courts continue to scrutinize the App Store's rules. As private architectures begin to resemble essential infrastructure, they tend to attract greater regulatory and judicial scrutiny over how they govern participation.
Still, the pattern recurs across too many industries to dismiss, and it clarifies a familiar puzzle: why do firms with excellent products keep losing ground to firms with merely good ones? Often they were competing on the wrong layer. They optimized the product while someone else designed the arena, and the arena shapes how the game is scored.
There is precedent. Standard Oil combined efficient refining with control of pipelines and favorable railroad arrangements. Separately, standardized railroad gauge transformed isolated rail lines into a continental transportation system. In both cases, control over the architecture of interconnection proved as important as the underlying product.
Products will keep creating value, as they always have. But the question that increasingly decides twenty-first-century competition is not only who builds the best product. It is who designs the system that shapes where the value of every product ultimately flows.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?