MIT's aerospace entrepreneurship paper, read from Europe

Who wrote it, and why the US lens matters before anything else

MIT ALIENS stands for Aerospace Leaders, Investors, Engineers & Students: sixteen people across MIT AeroAstro, MIT Sloan and Wellesley College, with prior time at SpaceX, Blue Origin, NASA, Boeing, Rocket Lab, Stoke Space and the SETI Institute. The paper came out of Prof. David Mindell's course 16.445, Entrepreneurship in Aerospace and Mobility Systems, and its method is the reason it is worth reading at all: more than fifty primary interviews with people the authors say have cumulatively raised over $1 billion, including acquisition officers in the US Army and Space Force. The authors are blunt that this is not a neutral survey — "we are not neutral observers" — and that Part 4 is opinion they are willing to defend.

They are equally blunt about the boundary, and we are going to hold them to it throughout: the paper is written "primarily through a US lens", the US is 55–60% of the global space economy today, and readers elsewhere "should weigh them against their own national context". Every figure below is in US dollars, as published. We deliberately do not convert any of it into euros, and we do not mix it with the euro figures from ESA's Report on the Space Economy 2026 — two different scopes, two different currencies, and mixing them is how a wrong number gets into a pitch deck.

What travels across the Atlantic unchanged is the analysis. What does not travel at all is the plumbing: SBIR, OTAs, Space Development Agency tranches, FOCI, ITAR. A European founder who reads the paper and starts looking for the European SBIR Phase III will not find one.

The number that should reframe a European funding plan

Private capital deployed over $58 billion across more than 400 companies in 2025, up 72% year-on-year. That is the headline everyone quoted. The paper's own reading of it is more useful in three ways.

First, the rebound follows a slump: private funding peaked around 2021, fell through 2022 and 2023, and is roughly flat since 2018 once the cycle is smoothed. A +72% year is not a new baseline. Second, government still accounts for roughly 70% of total space spending — about $137 billion in 2025. Private investment has grown at about 12% a year since 2009 against government's 4–5%, but in absolute terms both added something like $50–65 billion over the period; private simply started from a far smaller base. Third, and this is the line to keep: private capital now sets direction at the margin; public money still sets the scale.

For a European team that ratio is not a curiosity, it is the shape of your funding plan. Europe's public share of space spending is higher than the US figure, not lower — ESA's own 2026 report puts Europe's public space budget at €13.5B and growing while private investment on the continent contracted. So the sequence that works here is the one the numbers imply: institutional money as the anchor, private round as the follow-on from a de-risked position. If your plan says "raise a seed, then look at grants", the plan is inverted, and the MIT paper's US data says so even more strongly for Europe than for the market it describes.

One more thing worth stealing from their capital chapter: non-dilutive money works as a signal, not only as cash. Stoke Space has raised over $1 billion, and its first capital was a modest SBIR award that de-risked early technology and, just as importantly, demonstrated traction to investors. The European ladder with the same signalling effect runs through ESA Kick-Start, GSTP and InCubed, EIC Pathfinder and Accelerator, the Seal of Excellence, and national agency contracts — see our European funding instrument map for what each one actually pays and at which TRL.

Four drivers, and where a European team can credibly stand

The paper organises the opportunity around four drivers and makes one claim about them that founders under-use: the most interesting companies sit at the intersection of two or more, not inside one.

The investor chapter puts the same point more bluntly, and it is the single most useful paragraph in the paper for the kind of company most European space teams actually are: the space economy "is no longer a bet on rockets". The authors tell investors to look at what the standard ecosystem maps under-represent — spectrum management, manufacturing software, MRO and on-orbit servicing, the integration infrastructure between commercial constellations and defence customers, and workforce- and standards-adjacent businesses. "They are the layer-two and layer-three businesses, the ones that benefit from launch costs falling without having to drive that fall themselves."

On-orbit servicing: four regimes, not one market

The servicing chapter is the strongest analytical work in the paper, mostly because it argues against its own thesis. The bull case is the aviation precedent: GE Aerospace's installed base of 70,000 engines drives an aftermarket worth roughly 70% of company revenue, inside an aviation MRO market of about $90B in 2024 that should nearly double by the mid-2030s. The orbital population that could support an analogue went from about 1,400 active satellites in 2015 to more than 11,000 by 2025, with another 18,000 projected by 2030 and an estimate of 43,000 launches by 2035 representing $665B of build-and-launch activity.

Then they take the analogy apart. Aviation's aftermarket dominance rested on three conditions — enormous per-unit cost, ground access for servicing, and mandated inspection cycles — and none of the three transfers cleanly to orbit. What you get instead is four regimes:

Two numbers to price before anyone pitches this. A servicer typically has to serve three to five customer satellites before its own cost amortises — a 2010 NASA Goddard finding, and the reason a 2012 refuelling deal collapsed despite a willing customer. And almost nothing now in orbit was designed to be serviced: no standardised docking interfaces, no refuelling ports, no modular subsystems. Serviceability is a design revolution sitting upstream of the service business, which is why standards work (CONFERS in the paper's account) gates the market.

The commercial-timing advice is the part European teams should copy: defence is the bridge market. Commercial operators wait for pricing data, insurance models and repeated demonstrations; defence customers already value manoeuvre, on-orbit logistics and refuelling. So the first viable product "probably won't look like aviation MRO. It will look like manoeuvrability, asset recovery, or mission flexibility, sold to a customer who already values those capabilities."

The honest ceiling on AI

The paper's treatment of AI is unusual because it quantifies what AI cannot do. The problem it addresses is tacit knowledge: much aerospace manufacturing expertise lives with experienced technicians and was never documented — machine-specific parameters, sequencing decisions, informal troubleshooting — and in some cases retirees have been recalled to perform tasks understood by only a few people. Roughly 43% of aerospace-certified businesses have fewer than 25 employees, so the knowledge is spread thin across a base that never consolidated the way the primes did.

Their estimate of what AI recovers, by category:

• Procedural knowledge — step sequences, parameter settings, vendor-specific quirks, inspection rules of thumb: 30–50%, and only if the capture work happens while the people carrying it are still on the floor.
• Pattern-recognition judgment — the felt sense that a process is drifting, or that a supplier's delivery commitment is performative: 10–20%, and only in narrow domains where the pattern can be exposed to sensors.
• Social and institutional knowledge — who decides what, which contracts are flexible, the political map of a customer organisation: essentially none. It transfers human to human or not at all.

In aggregate: a quarter to a third of the problem. AI plus reshoring plus education plausibly covers two-thirds over fifteen to twenty years, and the authors call the remaining third permanent loss. This is one of four claims they say they will defend hardest — that the industrial base is a workforce and tacit-knowledge problem rather than a supply chain problem, and that instruments treating it as the latter will keep underperforming. If you are writing an impact section for a European manufacturing or workforce call, that distinction is a better framing than the one most proposals use.

Five predictions, with the counter-arguments they attached themselves

Part 4 is explicitly the authors' own bets, each "calibrated to be falsifiable", and each published with a where serious people disagree paragraph. Reproducing the bet without the counter-argument would misrepresent the paper, so here are both.

Prediction four is where a European reader has to be most careful, because it is the most instrument-specific and therefore the least portable. "Defence becomes the founder's cheapest money" is a claim about a market with SBIR, OTAs, SpaceWERX and the Defense Innovation Unit. Europe's version of that sentence is being written right now through EDF, EUDIS and national defence-space budgets, and it is not yet the same sentence. Prediction five, by contrast, is the one that most helps a European company: if fragmentation is permanent, then regulatory geography is a strategic choice, and the paper's own examples make the point that this cuts in Europe's favour — Rocket Lab built in New Zealand, Skyroot in India, ICEYE in Finland, ICON in the UAE, and "each made a strategic bet on geography that the founders defaulting to El Segundo did not."

What we would actually take into a European funding plan

1. Sequence public first, and say so out loud in the plan. At ~70% of spending in the market the paper describes, and a higher share in ours, institutional money is the anchor rather than the fallback. The corollary is that the anchor has to be chosen deliberately — see the instrument map for which line fits which TRL.
2. Pick an intersection of two drivers, not one. Their claim is that the disproportionate value over the next decade goes to companies operating across two or more of capital duration, dual-use, lifecycle economics and AI. Pick the intersection your team is genuinely credible in, and let the others shape second-act optionality.
3. Aim at layer two and three. Manufacturing software, ground segment, spectrum management, servicing, the integration layer between commercial constellations and defence customers. These are also the categories where a small European team can be world-class without a launch budget.
4. Count your miracles before you choose an instrument. One miracle, two, or three-plus decides whether you should be raising venture money at all — we go through the diagnostic and the European instrument mapping in Feasibility is not viability.
5. Treat regulation as a design constraint in year one. Standards bodies are upstream of regulators, and the committees are open. Which rooms, and what it costs to skip them, is in Regulation is a design constraint.

And the sentence we would put above a founder's desk, which is the paper's central thesis rather than any of its numbers: in aerospace the decisions that determine whether a company survives are made early, in choices that do not feel architectural at the time. "The successful aerospace founders we interviewed treat year one as strategic and year three as operational, and the inversion is consequential."

FAQ

Sources