A company slows down as it grows, and every founder experiences this as a moral failure — of nerve, of culture, of hiring. It is mostly none of those. It is closer to a law of physics, the same one that explains why an elephant's heart beats thirty times a minute and a mouse's beats six hundred. Once you see the mechanism, you stop trying to will your way back to being fast and start doing the one thing that works: redesigning the network that is dragging your metabolism down.
The biology is exact, so let me state it exactly before I map it onto anything.
An elephant is not a large mouse
In 1932 Max Kleiber measured how metabolic rate scales with body size across animals and found something that should be strange and mostly isn't. Metabolic rate does not scale in proportion to mass. It scales to roughly the three-quarter power of mass. Double an animal's mass and its energy consumption rises by a factor of about 1.68, not 2. Put differently: per unit of body mass, bigger animals burn less energy, and they do it on a precise curve — each doubling in size buys a 16% cut in metabolic rate per gram.
Run the numbers to the extreme and the effect turns violent. An elephant outweighs a mouse by a factor of about 100,000. Kleiber's law says its total metabolic rate is only about 5,600 times higher (100,000 to the 0.75 power). So per gram, elephant tissue runs at roughly one-eighteenth the metabolic intensity of mouse tissue. This is why the elephant's heart is slow and the mouse's is frantic, why the elephant lives seventy years and the mouse lives two, and why — this is the eerie part — both hearts beat roughly the same number of times in a lifetime, somewhere around a billion. Size does not buy you more life measured in beats. It buys you the same allotment, spent slower.
The mouse is not a small elephant running behind on efficiency. The elephant is a slow mouse, and it is slow because it is large, on a curve you can write down.
The slowdown is in the plumbing, not the cells
Here is the part founders need, because it is the part that transfers. The three-quarter exponent is not a property of the cells. An elephant's cells and a mouse's cells are made of the same machinery and, in a dish, respire at similar rates. The slowdown is imposed from outside the cell, by the network that has to reach every cell.
The leading explanation, worked out by West, Brown, and Enquist in the 1990s, is that metabolic scaling is set by the geometry of distribution networks — the vasculature. To feed an organism you need a branching network that delivers resources from one source to every cell, fills the whole volume of the body, and ends in capillaries that are the same size in a mouse and a whale. As a body gets larger, the network that services it has to get disproportionately more elaborate, and the physics of pushing blood through that branching tree — minimizing the energy lost to the plumbing itself — forces the three-quarter law. The cells could go faster. The network won't let them, because past a certain size, more and more of the organism's energy goes to running the infrastructure that keeps the rest of the organism supplied.
That sentence is the whole essay. Past a certain size, a growing fraction of the system's energy goes to servicing the network rather than doing the work. Biology solved this as well as it can be solved — the fractal vasculature is a genuinely optimized answer — and even the optimal solution still bends the curve down to three-quarters. There is no version where the whale keeps the mouse's per-gram metabolism. The network tax is not a defect. It is the cost of being large.
Your company obeys a cruder version of the same law
Now the analogy, and I want to be honest that it is an analogy — a company is not literally an organism and there is no measured three-quarter exponent for org charts. But the mechanism transfers cleanly, and it transfers in the wrong direction: a naive company scales worse than a body, not better.
The distribution network in a company is not blood. It is coordination — the communication paths, shared context, approval chains, and alignment that let people work on the same thing without colliding. The raw number of possible communication links between people is not linear. It is the handshake formula, n(n−1)/2. Watch it move:
| People | Communication links | Ratio to a 5-person team |
|---|---|---|
| 5 | 10 | 1x |
| 50 | 1,225 | ~122x |
| 500 | 124,750 | ~12,475x |
A 500-person company has 100 times the headcount of a 5-person company and roughly 12,000 times the potential coordination links. Nobody maintains all of them, which is exactly the point — the org spends enormous energy deciding which links to maintain, routing around the ones it drops, and paying for the misfires when a link that should have existed didn't. That expenditure is your organizational metabolism being consumed by the network instead of the work. It is why a 500-person company is not 100 times the throughput of a 5-person one, and why per-person output falls with size as reliably as heart rate does.
Left alone, coordination grows quadratically while output grows, at best, linearly. Biology's plumbing bends the curve down to a sublinear three-quarters. A poorly designed company bends it the wrong way, up toward n-squared. That is the default you are fighting.
Which slowdown is lawful and which is scar tissue
The value of the mechanism is that it draws a line most founders can't see, between the slowdown you must pay and the slowdown you invented.
The lawful slowdown is irreducible coordination cost. Aligning fifty people on a strategy genuinely takes more energy than aligning five, because there are more perspectives to reconcile and more context to keep synchronized. A decision that touches four real dependencies genuinely cannot move as fast as one that touches zero. You cannot hustle your way out of this any more than a whale can will its heart to beat like a hummingbird's. Trying to — demanding startup-speed decisions from a large org through sheer intensity — mostly produces exhausted people making the same slow decisions while feeling guilty about it.
The avoidable slowdown is everything the metabolic math does not require. Approval chains added after one incident and never removed once the incident was forgotten. Meetings that exist only to manage the consequences of a bad interface between two teams. And the one I have paid for personally: too many simultaneous bets, each one multiplying the coordination surface, because most startups die of indigestion rather than starvation — they take on more parallel initiatives than they can metabolize, and every extra bet doesn't just add work, it adds links, and links are quadratic. Five half-finished projects don't cost five times a focused one. They cost the coordination surface of five projects overlapping, which is far worse.
The test is simple. Measure how far your decision velocity has actually fallen and compare it to what the coordination math alone predicts. The gap between them is scar tissue, and scar tissue is removable.
Build better vasculature
Biology already showed you the escape hatch, and it isn't "get smaller." The fractal vasculature is the existence proof that clever network design can bend a brutal scaling law into a survivable one. The organizational equivalent is precise, and it is the most important thing you can do to keep a growing company fast.
The move is modularity: autonomous units with clean interfaces. Here is why it works, in the same handshake math. Instead of one undifferentiated blob of n people, you build m teams of k people each, where each team presents a single clean interface to the rest of the org — one API contract, one clear owner, one documented boundary. Internal coordination is now k(k−1)/2 inside each team, and the teams coordinate with each other, not everyone with everyone. For a fixed small team size k, total internal links grow linearly in n, not quadratically. Modularity is the transformation that converts n-squared coordination into roughly n. It is the fractal-vasculature trick, done with org design.
This is what Amazon's two-pizza teams and clean service APIs actually buy — not culture, arithmetic. A clean interface is valuable for one measurable reason: it caps the number of links a unit exposes to the rest of the system. A team of eight with one well-defined contract to the outside world carries about 28 internal links and a handful of external ones, whether the company has 50 people or 5,000. Without that interface, those same eight people are eight more nodes in a company-wide n-squared mesh.
The corollary that should scare you into acting: slowness is not just lost throughput, it is decay. Advantages don't hold still while you coordinate — your competitive advantage has a half-life, and a slower organization refreshes its advantages at a lower rate. A company whose metabolism has fallen isn't merely producing less; it is losing ground against its own moat while standing still, because the moat erodes on its own clock and the company can no longer dig fast enough to keep up. Metabolic rate isn't a vanity stat. It's the rate at which you can outrun decay.
What to do Monday
Stop counting headcount and start counting links. Take your last genuinely consequential decision — a pricing change, a launch, an architectural commitment — and list every person or team whose sign-off, context, or coordination was actually required before it could ship. That count, tracked over time, is your metabolic load. If it grows faster than your headcount, your network is degrading and your per-person output is already falling, whatever the org chart says.
Then act on the two categories separately. Pay the lawful slowdown without guilt; it is the cost of being large, and fighting it with intensity just burns your people. Attack the avoidable slowdown surgically: kill the approval chains that outlived their incident, collapse the meetings that exist only to patch bad interfaces, cut the parallel bets that multiply coordination surface. And architect for autonomy on purpose — small units, clear owners, clean contracts — because that is the only lever that changes the exponent instead of just working harder against it.
The founder who says "we just need to move faster" is asking the whale to have a mouse's heart. The founder who redesigns the vasculature is asking the right question: not how do I make my people run faster, but how do I stop making them carry the network.