Unscaled

The Death of Bigness as Competitive Moat

There is a particular kind of institutional blindness that comes from having won. Procter & Gamble spent a century perfecting the machinery of scale — the distribution networks, the shelf-space negotiations, the mass-market advertising budgets — and in doing so built something that looked, for a very long time, like an impregnable fortress. Then Dollar Shave Club arrived, renting rather than owning the capabilities that P&G had spent decades accumulating, targeting a narrow slice of dissatisfied customers rather than broadcasting to everyone, and moving with a speed that a billion-dollar infrastructure could never match. The fortress turned out to be a liability in disguise. This is the central provocation of Taneja’s argument: that scale, the organizing principle of industrial capitalism for over a century, has inverted. What was once a moat is now an anchor.

The logic of scale was airtight when the world was physical and capital-intensive. If you spent a billion dollars building a factory, you desperately needed to produce a billion units to amortize that cost into something survivable. The math forced bigness. It forced standardization. It forced mass markets, because you needed enough customers to justify the fixed cost of serving them at all. This is not a failure of imagination on the part of twentieth-century industrialists — it was rational strategy given the constraints of the physical world.

The 2007 Inflection and What It Actually Meant

What changed, and when, is where Taneja’s analysis becomes most interesting. He locates the origin point with some precision: 2007, the year the iPhone launched, the year Facebook became a genuine platform, the year AWS began giving every developer access to cloud infrastructure that previously only giants could afford. More than one billion people were on the internet that year; by 2016 it was three billion. Smartphone penetration went from a rounding error to 2.5 billion people in roughly the same window. These numbers are not merely impressive — they describe the mechanism by which the old economics of scale were dismantled.

The deeper point is about what all those connected humans produced: data. And data, it turned out, was not just a resource to be managed or a metric to be reported. It was the training material for artificial intelligence that had spent decades as an expensive theoretical disappointment. The platforms that emerged around 2007 created the conditions — mobile, cloud, social — for data to accumulate at a scale and richness that finally made AI useful. The machines could now learn, and what they learned came from the behavior of billions of people who had moved substantial portions of their lives online.

This means unscaling is not simply a business strategy. It is a structural consequence of a technological transformation. Platforms absorb the fixed costs that used to require massive institutions to bear. A lone entrepreneur renting AWS, running ads targeted by Facebook’s behavioral data, shipping through Amazon’s fulfillment network, can now access capabilities that once required the organizational infrastructure of P&G. The more platforms mature, the less any individual actor needs to own in order to operate.

Personalization as the Mechanism, Healthcare as the Test Case

The diabetes story Taneja tells is worth sitting with carefully, because it illustrates the mechanism more concretely than any abstract economics argument can. Here is a disease affecting thirty million Americans, fastest-growing in the world, substantially driven by the mass-market food system — the carbohydrate-heavy diets pushed by industrial food producers optimizing for shelf life and mass appeal, the high-fructose corn syrup that crept into everything because it was cheap at scale. The mass-market machine created a mass-market pathology. And yet diabetes is manageable; the outcomes diverge enormously depending on individual behavior, individual biology, individual circumstance. The scale-era healthcare system could not address this with sufficient granularity, because it too was built for the average patient rather than the specific one.

This is where AI and unscaling converge. If you can collect rich behavioral and biological data from a specific person and run it through systems trained on millions of similar cases, you can begin to deliver care that responds to that individual rather than to a statistical abstraction of them. The scale model needed to treat everyone similarly to make the economics work. The unscaled model can treat everyone differently because the platforms absorb the costs that used to make personalization prohibitively expensive.

Adjacent Resonances: Education, Autonomy, and the Tyranny of Average

There is a digression in the book that feels almost incidental but lands with unexpected weight. Taneja describes skipping classes at university, finding them too fast or too slow, joking with a friend named Sal Khan that self-paced learning was the real answer. Khan, of course, went on to build Khan Academy on exactly that insight. It is a small anecdote, but it captures something structurally important: the mass-education system, like the mass-market food system, was optimized for the average student at the expense of every actual student. The lecture format, the semester schedule, the uniform curriculum — these were solutions to the logistical problem of delivering education at scale, not solutions to the problem of learning itself.

The same critique applies across sectors. Mass media, mass retail, mass medicine — all were efficient approximations that served no one perfectly because they were designed to serve everyone adequately. The adjacent field here is psychology, specifically the literature on individual variation and how stubbornly institutional design ignores it. Unscaling, read generously, is a technological argument for finally taking human heterogeneity seriously.

Why This Matters Beyond the Business Case

I find myself returning to the structural question underneath the entrepreneurial advice. The scale era produced not just large companies but large concentrations of power — economic, political, cultural. If the unscaling thesis is correct, that concentration becomes unstable. The question is what replaces it, and whether the platforms that enable unscaling become their own form of concentrated scale, which is not a hypothetical concern given what Amazon, Google, and Meta have become in the years since Taneja wrote this.

The argument for optimism is that the costs of reaching specific people with specific solutions have dropped toward zero, which is genuinely democratizing. The argument for skepticism is that the infrastructure enabling all of this is itself heavily concentrated. I do not think Taneja resolves this tension, and perhaps that is appropriate — we are living inside the transformation, not yet able to see its shape from the outside.