The Towel Problem
Think, for a moment, the sheer range of things humanity has managed to not screw up. We split atoms. We put rovers on Mars that outlived their warranty by fourteen years, which is more than can be said for most marriages or mid‑range appliances. We built vaccines for a novel virus in under twelve months. We engineered a telescope that photographs galaxies as they appeared thirteen billion years ago, which means we can literally see the past but apparently cannot see our own companies dying in front of us.
And yet. When a company starts to decline, the most common explanation, delivered with the practised solemnity of a funeral director who also does motivational speaking, is: “Well, every company has a lifecycle.”
We put defibrillators in shopping malls. We train strangers to restart stopped hearts in airport terminals. But when a billion‑dollar enterprise begins to flatline, we fold our hands, nod gravely, and call it the natural order. We teach surgeons to restart hearts and teach MBAs that corporate death is a season, something dignified and cyclical, like autumn leaves falling from a tree nobody remembered to water. This is, and I want to be precise here, insane.
Nobody calls it what it is: a surrender wearing a good suit.
So here is a different question, and it changes the shape of everything that follows. What if, instead of managing decline with grace and a tasteful press release, you designed a system capable of redesigning itself? What if the cosy bedtime story about corporate lifecycles was nothing more than a narrative lazy minds constructed to avoid the much harder, much less comfortable work of perpetual creation? Because if even one company on this planet has cracked that, and several have, then the lifecycle theory is not a law. It is an excuse. And excuses, unlike thermodynamic principles, can be fired.
The Architects Who Forgot They Could Build
Before looking at who solved this, it is worth understanding how the forgetting happens, because the pattern is remarkably consistent and almost comically predictable.
The standard version goes like this: Kodak failed to adapt. Nokia missed the smartphone. Blockbuster ignored streaming. These stories have been chewed over so thoroughly in business schools and keynote speeches that they now have the nutritional value of sawdust. The retelling always frames the failure as a failure of foresight. But that is not what happened. Not even close. Kodak invented the digital camera in 1975. Nokia had smartphone prototypes before Apple had a keynote. Blockbuster had a deal to launch a streaming service in the year 2000, with Enron of all partners, which in hindsight explains quite a lot. These were not blind organisations. They saw the future with perfect clarity. They simply could not bring themselves to build it, because building it meant dismantling what already existed, and dismantling is an act of architecture, not maintenance. It requires people who know how to pick up a hammer. They had hired people who only knew how to hold a clipboard.
The more revealing cases are the ones still warm. General Electric did not collapse because the industrial age ended. It collapsed because two decades of financial engineering, under leadership that mistook the balance sheet for the building, hollowed out the actual engineering that once produced jet engines, MRI machines, and power turbines. By the time anyone thought to check, GE was not a technology company. It was a leveraged spreadsheet wearing a hard hat. Between 2016 and 2018, GE lost over $200 billion in market value. Not because of disruption. Because the termites had been eating the frame for twenty years and everyone was too busy polishing the front door to notice.
Intel tells the same joke from a different angle. For decades, Intel was the undisputed sovereign of chipmaking. Then it spent roughly ten years optimising what existed rather than questioning whether what existed was still worth optimising. TSMC did the questioning. By 2024, TSMC was manufacturing the most advanced chips on Earth while Intel was announcing layoffs and restructuring plans with the grim enthusiasm of someone rearranging furniture on a sinking ship. Intel had every resource, every patent, every institutional advantage. What it did not have was a single person willing to say: tear it down, we’re building something better.
The thread connecting all of these is not disruption. Disruption is what happens to you. This is abdication. This is what you do to yourself. These companies replaced architects with custodians, people who could maintain a building but had forgotten, or never learned, how to build one. And that gap, between maintaining and building, is the crack through which everything leaks out. So if those were the ones who forgot, the question that matters is: who remembered?
The Paranoid and the Recursive
Forget the usual Silicon Valley saints. Everyone has already written that article. The interesting proof is in the places nobody thinks to look.
Start with ASML. A Dutch company. Makes machines. Specifically, it makes the only machines on Earth capable of printing the most advanced semiconductor chips in existence, using a process called extreme ultraviolet lithography. One machine costs roughly $380 million, contains over 100,000 parts, and includes a laser system that fires 50,000 droplets of molten tin per second to generate the specific wavelength of light needed to etch circuits smaller than a virus. No other company on the planet can build one. Not one. ASML did not get here by protecting what it had. It got here by destroying what it had, deliberately, repeatedly, and on schedule. Every generation of lithography technology at ASML began development while the current generation was still profitable, still unchallenged, still the industry standard. They spent over twenty years and billions of euros developing EUV while their existing machines were printing money. Most companies cannot bring themselves to cannibalise a product line that is still profitable. ASML made it a religion.
Then there is Danaher, the conglomerate most people have never heard of despite the fact that it is worth more than Goldman Sachs. Danaher built something that, if you describe it at a dinner party, will make people slowly put down their forks: a self‑improving operating system for acquiring and improving companies. It is called the Danaher Business System, or DBS, and it works like this. Danaher acquires a company. Applies DBS. Improves the company. Then feeds what it learned back into DBS itself, which improves, and then gets applied to the next acquisition with greater precision. The system literally upgrades itself using its own output. Since the early 2000s, Danaher has consistently outperformed the S&P 500, not because it picks winners, but because it built a machine that manufactures improvement at a structural level. It compounds intelligence, not just capital. Most investors still have not figured out what they are looking at.
And then Mercado Libre, which most people outside Latin America have never seriously considered, mostly because underestimating Latin America is something of a professional sport in the northern hemisphere. Mercado Libre did not copy Amazon for a different geography. It built payments, logistics, lending, and commerce simultaneously, not as separate bets but as interdependent layers, because the infrastructure those services required did not exist. Their payments platform, Mercado Pago, now processes more transactions than many traditional banks in the region. Their logistics arm delivers to areas that global carriers had written off as commercially pointless. Each layer created the conditions for the next layer to emerge. This is not a company that does several things. This is a system where each component manufactures the preconditions for the others to improve, a recursive loop that feeds itself.
Three industries. Three continents. One principle: they did not build companies that operate, they built companies that recurse. And recursion, it turns out, makes entropy a non‑issue. Which brings us to why.
Entropy Delusion
Here is the part where the people who quote the second law of thermodynamics in board meetings should sit down and pay close attention, because they have been misquoting it for decades and nobody has had the nerve to correct them.
Entropy applies to closed systems. That’s it. A sealed box of gas molecules will trend toward disorder because nothing enters, nothing is reorganised, and no energy is introduced. The molecules do not choose chaos. Chaos is simply what happens when nothing and nobody intervenes. But, and this is the part that gets conveniently left in the car, a company is not a sealed box of gas molecules. It is not a closed system. It is not governed by the same constraint. Unless you make it one.
And that is exactly what most organisations spend their entire existence doing. They seal themselves shut and then act baffled when the air runs out. They do it with bureaucracy that punishes initiative and rewards compliance. They do it with quarterly thinking that treats twelve‑week intervals as the outermost boundary of the knowable universe. They do it with layers of approval designed to prevent mistakes that also, with beautiful efficiency, prevent adaptation. Every process built for consistency eventually becomes the thing that makes change impossible. The antibodies that were supposed to protect the organism end up attacking it. And then, when the inevitable slide begins, someone in an expensive suit stands at a podium and says, with a face so straight it could have been drawn with a ruler, that entropy is simply the way of things.
It is not the way of things. It is the result of enclosure. And enclosure is a choice, not a physical law.
ASML, Danaher, Mercado Libre, they did not survive because they were lucky or because they had once‑in‑a‑generation leadership. They survived because they refused to seal the box. They kept the system open, kept injecting contradiction and reinvention and the productive discomfort of questioning everything they had already built. An open system with continuous energy input does not decay. It transforms. That is not inspiration. That is thermodynamics, correctly applied for once.
A Choice That Isn’t a Choice
The universe is under no obligation to keep your company alive. It did not sign your articles of incorporation. It does not subscribe to your investor newsletter. It has no opinion about your five‑year plan. But here is what gets missed in the rush to accept that cold, fashionably nihilistic fact: the universe never once asked for your company to die, either. That part was always a human decision, made by human beings who found it more comfortable to accept a narrative about natural cycles than to do the genuinely exhausting work of building something that rebuilds itself.
The corporate graveyard is not a memorial to cosmic inevitability. It is a museum of choices. Every exhibit was placed there by people who chose custodianship over architecture, maintenance over creation, the slow familiar comfort of decline over the fast unfamiliar discomfort of reinvention. They chose to manage a system instead of designing one that could design itself, and then they borrowed the language of physics to make that choice feel like fate. It was never fate. It was a preference, and not a particularly brave one.
The companies that endure are not the ones that fight entropy, because fighting entropy still accepts the premise that entropy applies. The companies that endure are the ones that made entropy irrelevant by building the redesign into the design, by treating every finished product as the rough draft of the next version, by embedding recursion so deep into the structure that self‑improvement is not a project or an initiative or a quarterly objective but the thing the organism does the way lungs breathe.
So the choice, if you can call it that, is between two futures. In one, you design a system that designs itself, and you accept that this means never being comfortable, never being finished, never standing still long enough to admire what you built, because by the time you turn around it should already be half‑replaced by something better. In the other, you manage, with increasing elegance and decreasing relevance, the long quiet funeral of something that was never built to last.
One of those paths demands relentless, ungrateful, sometimes brutal thinking. The other requires only a comfortable chair and a calendar long enough to watch the ending arrive.
The universe, characteristically, does not care which one you pick. But you probably should.