This is the story of a company that never changed its mission. The mission statement still says the same thing it said fifteen years ago. Nobody updated it. Nobody needed to. Nobody noticed.
Year 1. Three founders leave a research institute with a sensor that can see something nobody else can see. The details don’t matter — pick your wavelength, pick your orbit. What matters is the pitch deck. Slide 7: “We’re building planetary-scale environmental intelligence.” Slide 12: the TAM is $4.2 billion. Climate adaptation. Precision agriculture. Disaster response. Insurance. The VCs nod. The cheque clears.
The roadmap has twelve milestones. All of them point toward understanding what’s happening to the planet.
Year 2. First images. They’re beautiful. The founders post them on LinkedIn. The ESA contract comes through. A pilot with a reinsurer. A conversation with a UN agency that goes well but produces no purchase order, because the UN doesn’t have a purchase process for this. The team is twenty people. Everyone believes.
Year 3. Series A. The lead investor has a climate thesis. Patient capital, long time horizon, “we’re building generational infrastructure.” The term sheet is generous. The board is supportive. The roadmap still points the same way. But in due diligence, one partner asks a question that will echo for the next twelve years: “What’s your path to $10M ARR?”
The honest answer is: nobody knows, because the commercial climate market doesn’t have budget owners yet. The answer on slide 14 is: “Land and expand with insurance and agriculture.”
Year 4. The reinsurer pilot ends. Not because the data was bad, the data was extraordinary. But the reinsurer can’t figure out which department should pay for it, how to integrate it into underwriting workflows, or how to explain to their board why they’re buying satellite data. There’s no regulatory mandate requiring it. There’s no line item in the budget. There’s no precedent.
The agriculture customers want the data but can’t pay what it costs. The UN agency comes back with a grant. Just enough to keep the lights on, not enough to scale.
Revenue at end of year 4: €1.1M. Burn rate: €3.8M.
Then a defence ministry calls. They saw a conference demo. They have a specific problem. They have a budget. They want to start in eight weeks.
The founders discuss it for one evening. The contract is €600K. “It funds six months of the climate roadmap,” they tell themselves. They sign it.
The signal that the climate market needs ten more years of patient infrastructure building — regulation, workflow integration, budget line creation — won’t arrive for another three years. By then, the company will already be structured around the answer it has today.
Year 5. The defence contract delivers on time. The ministry comes back with a second contract. Larger scope, faster timeline, clearer specs. The procurement officer knows exactly what he wants. After two years of trying to explain to agricultural cooperatives why satellite data matters, the clarity is intoxicating.
Series B conversations start. The lead investor from Series A introduces a growth-stage fund. The growth fund likes the technology. They like the team. They don’t love the revenue mix. “Your commercial pipeline is soft,” they say. “But the defence traction is real.”
The founders build two slides. One shows the “commercial roadmap”: agriculture, climate services, and insurance. The other shows “government and institutional revenue,” which now includes defence contracts but is labelled more broadly. The growth fund invests. The board gets a new member who spent fifteen years in aerospace.
Revenue at end of year 5: €3.2M. Defence is 40%.
The roadmap now has two tracks. Everyone agrees this is temporary.
What nobody maps on a whiteboard: success to the successful is the most predictable loop in any system. The side that gets resources performs. The side that performs gets more resources. The side that’s starved underperforms, which justifies starving it further. It doesn’t require malice. It just requires a quarterly review cycle.
Year 6. The company hires a VP of Sales from the defence industry and a VP of Operations from a Series D startup. Both are excellent. The VP of Sales knows procurement cycles, classification requirements, and the right language for proposals. Within six months, the pipeline triples — but only on the defence side.
The VP of Operations implements OKRs, quarterly business reviews, and a reporting cadence. The dashboards look professional. Revenue per head. Pipeline velocity. Contract win rate. Time to delivery. The defence side of the business produces beautiful metrics — clear demand signals, predictable procurement cycles, measurable everything. The climate side produces pilot results, research citations, and “promising conversations” that don’t convert to the pipeline.
In the quarterly review, the defence team presents numbers. The climate team presents narratives. The board knows which one they trust.
What gets measured gets managed. What gets managed gets resources. What gets resources gets measured better. The loop tightens. Meanwhile, the unmeasured slowly becomes unmeasurable, then unreal. Within two years, the climate work will be described in board meetings as “harder to track” rather than “strategically important.” The language shift happens first. The resource shift follows.
The climate team asks for two more engineers to build the agricultural analytics platform. The VP of Sales asks for three more engineers to meet the defence delivery schedule. Both requests go to the board. The defence engineers are approved immediately. The climate engineers are approved “next quarter.”
Next quarter, the defence pipeline needs four more.
Year 7. A new analyst joins. Smart, young, excited about the mission. On her first day, she asks which project she’ll be working on. She’s assigned to a defence contract. “Just for onboarding,” they say. “The climate work needs experienced people.”
She’s still on defence contracts eighteen months later. She’s good at it. She’s been promoted.
The company is eighty people now. At twenty, engineers talked directly to customers. Decisions happened in real time. The founders’ obsessive knowledge of the problem permeated every conversation. People self-organised around what mattered because they could see what mattered.
At eighty, there are three management layers between the engineer and the customer. Decisions travel up for approval and back down for execution. The weekly all-hands has a slide deck. The founders no longer eat lunch with the team. They eat lunch with investors and procurement officers.
The three founders now spend their time differently. The CEO talks to defence procurement officers and VCs. The CTO manages the classified development environment. The Chief Scientist, who built the original sensor, still leads the climate research group. It’s the smallest team in the company.
Revenue at end of year 7: €8.4M. Defence is 58%.
Year 8. Series C. The term sheet requires a path to €50M ARR within three years. The board deck is forty slides of metrics — growth rate, pipeline coverage, net revenue retention, CAC payback period. The defence business produces all of these cleanly. The climate business produces research impact, potential market size, and the word “emerging.”
The only realistic path to €50M runs through defence and national security. Everyone at the board table knows this. Nobody says it. They are all making reasonable decisions based on the information in front of them. The information in front of them is this quarter’s pipeline and next quarter’s forecast. The climate trajectory is a fifteen-year story. The board deck has a three-year horizon.
The commercial climate roadmap is in appendix B of the board deck. It used to be slide 7.
New investors join the board. Their portfolio includes three other defence tech companies. They’re helpful. They make introductions. The introductions are to other defence ministries.
Year 9. The Chief Scientist presents a breakthrough: a new processing pipeline that could halve the cost of agricultural monitoring. It would make the product accessible to farming cooperatives in East Africa. She needs €400K and two engineers for nine months.
The board asks what the revenue projection is. She says it’s hard to quantify because the market doesn’t exist yet, but this could be the thing that creates it.
The room is polite. But the language of the room is ARR and pipeline coverage and CAC payback. She’s speaking a different language. The one the company used to speak.
The board approves €200K and one engineer. “Prove the concept first.”
A half-commitment is worse than no commitment. It creates the appearance of support while ensuring failure. When it fails — and with half the resources and a third the timeline, it will fail — the failure will justify the original scepticism. Fixes that fail. The loop closes.
That same meeting, they approve €2.1M for a new classified processing facility.
The Chief Scientist doesn’t argue. There’s nothing to argue with. The logic is airtight. If quantity forms the goals of your feedback loops, quantity will be the result.
Year 10. The company wins a framework contract with a European defence agency. Multi-year. Eight figures. The press release calls it “a milestone for European sovereign capability.” LinkedIn goes wild. The stock options are suddenly worth something.
The founding team celebrates. This is success. This is what they worked for.
Isn’t it?
Year 11. Three climate team engineers resign within two months. They don’t leave angry. They leave tired. One joins an NGO. One joins a climate startup that just raised a seed round. One goes to academia.
The company hires replacements. The job description says Earth observation analyst. The candidates all have defence backgrounds. This is who applies now.
At twenty people, the company could pivot to anything: new sensor, new market, new application. That was resilience. Not a strategy document. An actual structural property of a system that hadn’t yet optimised itself into a single shape. Resilience isn’t static. It’s the ability to bounce back, to reorganise, to evolve. By Year 11, that ability is gone. The company is optimised. It’s also brittle. It just doesn’t know it yet, because nothing has pushed back.
Year 12. The agricultural analytics platform is quietly shelved. Not cancelled. Shelved. “Deprioritised pending market development.” The code sits in a repository that fewer and fewer people have access to, because the production environment is now classified.
The mission statement on the website still says “planetary-scale environmental intelligence.” The about page still shows an image of a green field. The last blog post about agriculture is fourteen months old.
Revenue at end of year 12: €34M. Defence is 73%. The remaining commercial revenue is mostly government environmental agencies who use the same contracting vehicle as defence.
Year 13. A journalist writes a profile of the company. “From Climate to Combat: How Europe’s EO Startups Found Their Market.” The CEO is unhappy with the headline but can’t dispute the numbers. He gives an interview explaining that the technology is “dual-use” and that the physics doesn’t change based on application.
This is true. The physics doesn’t change. The roadmap did. “Dual-use” is a boundary someone drew to make two diverging realities fit in one sentence. Boundaries are useful for clarity. They become dangerous when you forget you invented them.
Year 14. The company is approached about an acquisition. The buyer is a defence prime. The valuation is strong. The board discusses it seriously.
The Chief Scientist is not in the room. She left in year 11. She now runs a small research group at a university, publishing papers about agricultural monitoring using open-source satellite data. Her work gets cited often. It doesn’t get funded.
Year 15. The acquisition closes. The press release mentions “environmental monitoring” and “climate resilience” in the second paragraph. The first paragraph is about “strengthening Europe’s sovereign intelligence capability.”
The mission statement is updated for the first time in fifteen years. Nobody notices. It now reads: “Delivering trusted intelligence for a secure and resilient world.”
The roadmap is classified.
If you draw the two roadmaps side by side, the one from the pitch deck in Year 1 and the one that actually happened, they diverge at Year 4. Not dramatically. Just a small angle. Five degrees, maybe. A single rational contract signed for rational reasons.
By Year 8, the angle is thirty degrees.
By Year 15, they point in opposite directions.
No single decision was wrong. The board saw quarterly metrics. The VCs saw portfolio returns. The founders saw burn rate. The climate market saw a company drifting away. None of them saw all of it at once. And none of them were responsible for seeing all of it at once. That’s the thing about systems. They don’t require someone to be wrong.
Every quarter, the climate roadmap eroded a little. Not by a dramatic cut — by a reallocation, a deprioritisation, a “next quarter.” The target shifts so slowly that nobody experiences a decision to abandon it. Eroding goals.
The technical term for what happened is drift to low performance. The system doesn’t abandon the difficult goal through one dramatic decision. It drifts toward the easier goal through a thousand small ones. Each rational, each incremental, each invisible in the moment. The stated goal stays fixed on the slide deck. The actual goal creeps downward every quarter. Not to “become a defence contractor” but to “hit this quarter’s number.” Eventually, those are the same thing. By the time anyone notices, the drift has been happening for years.
Underneath that: loop dominance. The mission loop ran the company for three years. Then the defence revenue loop overtook it. And once a reinforcing loop dominates, it feeds itself. The relationship isn’t linear. It accelerates. By the time anyone notices which loop is in charge, it’s been in charge for years. Success to the successful.
The twenty-person company that built the original sensor was a self-organising system — adaptive, resilient, capable of evolving. Every layer of management, every dashboard, every quarterly review that made it “scalable” also stripped those properties away. Self-organisation sacrificed for short-term productivity. As it usually is.
If quantity forms the goals of your feedback loops, quantity will be the result. The system is stable. It’s just stable around the wrong goal.
The original sensor still works. The physics hasn’t changed. The wavelength is the same. The orbit is the same. The data is the same.
The question was never about the sensor. It was about who gets to use it, for what, and who decides.
Meanwhile, the planet is changing. As it always has. But at a rate that is much higher than before. A rate that doesn’t necessarily hurt the planet, but makes our ability to live and flourish on it less and less likely. The sensor that was built to see that is still in orbit. It’s just looking at something else now.
This is, of course, a work of fiction.