Gravity Tax

The Escape Velocity Problem

Microsoft shipped Copilot to enterprise customers and watched most of them use it to create meeting notes that nobody reads. Meanwhile, a handful of companies restructured how work gets done and started compounding their advantage. The gap between these two groups is widening, and it has little to do with technology.

In physics, escape velocity is the speed you need to reach to escape an object’s orbit. Below it, gravity wins no matter how much energy you spend. Above it, every unit of effort moves you further away.

$$v_e=\sqrt{\frac{2GM}{r}}$$

There are only three variables (technically 2 variables and a constant): the mass you’re escaping ($M$), the gravitational constant ($G$), and your distance from the centre ($r$). You can’t change $G$. That leaves two levers: reduce mass or increase distance. Mass and distance set the threshold. Velocity is what gets you past it.

These map to AI adoption more cleanly than you’d expect. Mass is the accumulated weight of how your organisation already works: the processes, governance, headcount, and institutional habits that made you successful but now resist change. Distance is how far your AI efforts sit from that centre of mass, whether you’re bolting AI onto existing workflows or rethinking the workflows themselves. And velocity is whether all the energy you’re spending on AI is pointed somewhere specific or just keeping you in orbit. Most organisations are below the threshold, and many don’t understand what crossing it actually requires.

Distance: Most AI lives at the bottom of the gravity well

In the escape velocity equation, $r$ is the distance from the centre of mass. The closer you are to the centre, the stronger the pull, and the more speed you need to break free.

Your organisation’s centre of mass is its existing processes and practices. Applying AI to these same processes, workflows, and mental models is operating at minimum distance. You’re trying to escape gravity from its strongest point.

The bottleneck for most organisations isn’t execution anymore — so much has been sped up with reflexive AI use. The bottleneck is coordination. Review cycles, approval chains, alignment rituals: these are organisational scaffolding built for slower systems. When delivery accelerates but structure doesn’t, the structure becomes the gravity.

A software team uses AI to write, test, and review a bug fix in a single afternoon. The code is ready. But the release train runs fortnightly, so the fix sits in a staging branch for eleven days waiting for a QA cycle that assumes code takes a week to write. Then it goes through a change advisory board that meets on Tuesdays. Then it needs sign-off from someone who’s already context-switched into the next sprint. Three weeks later, it ships. The bottleneck is no longer engineering.

Creating distance means changing the scaffolding, not just the work inside it. It means compressing approval chains so the structure can absorb the speed that AI enables. It means questioning whether the alignment rituals that made sense at human pace still make sense when the constraint has shifted from production to coordination. Low $r$ optimises the existing way of work. High $r$ creates distance from it. Distance is what weakens gravity’s grip.

Mass: Success is the heaviest thing to escape

The same things that made a company successful are the things now holding it down. Scale. Process maturity. Risk management. Institutional knowledge. Governance. People. These all add mass, and the higher the mass, the higher the gravity.

A startup doesn’t have an AI advantage because it’s smarter or more innovative. It has a mass advantage. There’s nothing holding it down. No approval chains. No procurement cycles. No “we tried something like this before.”

Large organisations have to generate enormous thrust just to overcome the gravitational pull of their own success. Big things take more time to move. And the irony is that the more successful you were in the pre-AI world, the harder escape velocity becomes. The processes that scaled your business to millions in revenue are the same processes that now resist change by design. These processes needed to be reliable and repeatable to scale, and that’s what makes them heavy.

Rockets solve the excess mass problem with staging: they jettison spent fuel tanks mid-flight. The mass that got you off the launchpad becomes dead weight at altitude. Organisations rarely stage. They carry every process, every layer, every committee that powered the last phase of growth into the next one. Organisations need to identify dead weight and jettison it.

Distance and mass set the threshold. Velocity is what gets you past it.

Velocity: Most companies have motion, not velocity

Velocity is speed with a direction. Motion is movement without one. Buying tools, running pilots, hiring an AI lead, publishing an AI strategy, setting up a centre of excellence — that’s motion. Energy spent without direction. You need to actively point your AI initiatives somewhere specific and add sustained forward movement, or you’re spinning in place.

An organisation buys Copilot seats for a thousand people. People use Copilot to take meeting notes they never look at. Some people consider this a success because adoption and usage are high, but how has this helped to achieve business outcomes?

Motion is the default outcome for most AI initiatives. Velocity requires you to first know where you want to go, and then apply sustained, directed energy against it. AI adoption is a vanity metric. The question isn’t are people using AI, it’s is AI moving us toward something we couldn’t reach before.

The Tax Compounds

Once a company reaches escape velocity, there’s almost no drag. Every unit of energy converts directly to forward motion. They learn faster because they’ve already paid the structural cost of absorbing change. New capabilities layer on top of the last ones. The gap widens not because they’re spending more, but because their energy converts more efficiently.

The companies still in orbit spend the same energy and go nowhere. Worse, the tax increases over time. Each quarter spent optimising within existing structures is a quarter where the structural gap grows. The companies that escaped are ahead and they’re accelerating, while the companies in orbit are paying more and more gravity tax to maintain the same altitude.

So what does the escape velocity threshold actually look like?

Below it, AI is something the organisation does. It’s a tool bolted onto existing processes. Every AI initiative has to fight the organisation to exist: it needs a business case, a sponsor, a pilot, a steering committee. The organisational immune system treats AI as a foreign body and routes it through the same approval structures that govern everything else.

Above it, AI is something the organisation is. The processes themselves have been reshaped around what AI makes possible: how teams coordinate, how decisions get made, how work actually moves through the company. The threshold isn’t a technology question. It’s the point where AI is a part of how people work rather than treating AI as an initiative people work towards.

The difficult part of this equation is that it doesn’t reward intent. Wanting to change isn’t enough. If you can’t shed the mass or create the distance, you’ll need an impractical amount of effort to break free. And that’s the uncomfortable truth most organisations face: it’s often easier to keep spending energy in orbit than to make the structural changes needed to cross the threshold.