Pipeline Europe Never Built

Customer-shaped products win; report-shaped products lose.

The Conversion Mechanism

Mynaric was founded in 2009 by engineers from the German Aerospace Centre. Its laser communications terminals were designed for exactly the satellite mesh networks the Pentagon wanted. The Space Development Agency selected Mynaric for its Transport Layer constellation. It had the technology. It had the contracts.

What Europe never provided was a customer.

Over ninety per cent of Mynaric’s backlog was American defence revenue. It built terminals for the SDA. It supplied Northrop Grumman.

The European orders were prototype-scale. Rocket Lab, an American-listed launch company riding $1.3 billion in SDA contracts, bought Mynaric for $75 million in February 2025.

Rheinmetall, Germany’s largest defence company, held a fourteen per cent stake. It did not make a competing offer.

The American architecture of commercial dominance is a connected sequence of institutions: DARPA, In-Q-Tel, SBIR, the commercial market. Each takes equity, shapes the company, becomes the first customer. The connection is the point. The mechanism is investment, not subsidy.

Subsidy is one-way.

A grant funds research, files a report, walks away. Nothing in the funding instrument creates a buyer.

Investment is two-way. An equity stake aligns the funder with the company’s success. The funder has skin in the game and the leverage to shape what gets built.

And in the American model, the funder also has the network to introduce the first customer, often the funder’s own institutional family. In-Q-Tel does not just write a cheque to Palantir.

It puts the company in front of the intelligence agencies that will buy the result. Two million dollars of public money becomes ten years of guaranteed revenue.[1]

Before 1980, fewer than five per cent of the 28,000 patents held by the United States federal government were commercially licensed.

Publicly funded research produced results that sat in government archives. The Bayh-Dole Act changed the ownership structure: universities and small businesses could retain patent rights to federally funded inventions.

Commercial licensing of government-funded research increased by a factor of ten within a decade.

Bayh-Dole worked because it did not work alone.

A university researcher funded by DARPA produces a patentable result. Under Bayh-Dole, the university licenses it. A startup forms around it.

In-Q-Tel or SBIR provides the first contract. The company builds a product shaped by an operational requirement, demonstrates it works, and enters the commercial market with a customer already on the books.

The pipeline is a sequence, and each stage creates the conditions for the next.[2]

Europe tried to replicate Bayh-Dole in isolation.

Germany abolished the professor’s privilege in 2002, shifting patent ownership from individual researchers to their universities.[3] The reform produced a measurable decline in both patent quality and industry collaboration.

Researchers who had previously worked informally with companies now faced institutional technology transfer offices that added cost and delay without adding demand. The mechanism was copied.

The pipeline it connected to was not.

European research has produced the World Wide Web, Linux, ARM, Bluetooth, and the MP3, each commercialised elsewhere. The capture mechanism is examined in Paper 9.[4]

The pattern has a specific cause. European institutions excel at producing capability. What they have never built is the institutional machinery that turns capability into a customer.

That machinery is a demand-generation system disguised as a research-funding system. DARPA creates a user. SBIR guarantees a buyer. In-Q-Tel signals to thousands of private venture capital firms that revenue is coming. The pipeline’s most important output is demand. Demand is the one thing Europe’s institutional architecture systematically fails to create.

What makes the pipeline invisible is that its components appear to be separate institutions operating independently.

European policymakers study Bayh-Dole and copy it. They study SBIR and establish equivalents.

They look at In-Q-Tel and fund venture arms. Each component is replicated in isolation, and each replica fails in isolation, because the components only work as a system.

Bayh-Dole without procurement is a technology transfer office with nothing to transfer to. SBIR without defence spending is a small-business grant. An innovation agency without a customer mandate is a more expensive version of a grant committee.

Europe has copied every visible part of the American pipeline and missed the invisible part: the guaranteed buyer at the end of the sequence.

That buyer is beginning to organise. The EuroStack initiative, launched in 2024 with more than three hundred CEO signatories, structures European demand into three commitments: Buy European, Sell European, Fund European.[5] It is the first serious attempt to build the demand side of the pipeline from the private sector up, rather than waiting for Brussels to legislate it into existence.

The Multiplier Effect

When In-Q-Tel validates a technology, private venture capital receives a signal: the intelligence community has tested this, it works, and revenue is coming.

Palantir’s $2 million In-Q-Tel seed preceded a company now valued above three hundred and seventy billion dollars. Anduril’s early defence contracts preceded $3.5 billion in venture funding.

The leverage is enormous because the signal is credible: a government customer means revenue, and revenue means investors.

The European Defence Equity Facility deploys €175 million to mobilise €500 million: roughly 2.9:1. The American pipeline routinely produces leverage ratios an order of magnitude higher, because the initial relationship is a customer, not a funder.

A purchase order sends three signals that a grant cannot.

First, revenue: investors fund companies with customers, not companies with grants. The customer is proof that someone will pay for what the company builds.

Second, credibility: a government that buys tells the market the technology works; a government that funds tells the market the technology might work.

Third, direction: a customer tells you what to build; a grant committee tells you what to propose.

The investor looking at Palantir in 2005 saw a company with a classified customer and a product shaped by operational necessity. The investor looking at a European defence startup in 2024 sees a company with a grant, a deliverable, and a search for someone willing to buy the result.

Helsing, a Munich-based defence AI company founded in 2021, raised €600 million in its Series D in January 2025, reaching a $5.4 billion valuation. Anduril, its closest American analogue, raised $1.48 billion in its Series F in August 2024 at a $14 billion valuation.

Helsing’s engineering capability is world-class. Its software is deployed with the Bundeswehr, the French armed forces, and the UK Ministry of Defence.

The difference between Helsing and Anduril is the institutional architecture behind each company. Anduril entered a market where the Department of Defense spends $130 billion annually on research and development and actively seeks commercial technology providers.

Helsing entered a market where European defence ministries collectively spend a tenth of that and default to grants over orders.

Helsing is the partial exception that clarifies the rule.

Software companies can survive on smaller contracts because the marginal cost of deploying code is near zero. A government AI contract worth €50 million can sustain a software company.

Hardware companies need production-scale orders. An optical terminal manufacturer cannot survive on research grants.

It needs a government to commit to buying hundreds of units at production prices. Mynaric had the technology.

What it needed was what Palantir got in 2003: a government that placed an order rather than issued a grant.

The European defence technology graveyard extends well beyond Mynaric.

Companies with world-class engineering, strong research partnerships, and European public funding that failed or were acquired because no European government committed to buying what they built. The pattern repeats in cybersecurity, in satellite manufacturing, in quantum computing.

In each case, the postmortem identifies insufficient private capital, or inadequate scaling, or failure to achieve product-market fit. In each case, the underlying cause is the same: no anchor customer.

The companies that survive are the ones that find an American buyer, or the ones whose technology is so specialised that no acquirer can absorb it without triggering regulatory intervention.

The pipeline is a filter, and the filter selects for companies with customers.

The HydRON contract proves it works in reverse.

ESA designed HydRON as its flagship programme for high-throughput optical satellite communications. In October 2024, ESA awarded the Element 1 prime contract, worth €36 million, to Kepler Communications. Kepler is headquartered in Toronto.

It entered the optical communications market by building commercial relay services for North American satellite operators, accumulating operational experience and iterating its terminal designs through actual customer deployments. By the time ESA tendered HydRON, Kepler had what no European company could demonstrate: a product shaped by years of demand-driven development.[6]

Tesat and Thales Alenia Space, both European, became subcontractors. ESA’s juste retour principle, which allocates contracts proportional to each member state’s financial contribution, had ensured that research contracts were distributed across European manufacturers. It had not ensured that any of those manufacturers served a customer at production scale.

The customer-shaped Canadian product won the European programme.

The Moat Around the Machine

Consider a European defence executive bidding on a satellite communications contract.

Her company has the engineering capability. It has the production capacity. It has security-cleared personnel and facilities certified to national standards.

The contract requires NATO interoperability, which requires compliance with technical specifications written around American equipment. Her ITAR-free design means she cannot access the classified technical data that defines the interoperability requirements.

She is bidding on a contract whose specifications she is not permitted to read in full.

She loses the bid.

She will encounter the same barrier on the next procurement and the one after that. The information architecture of NATO-adjacent contracts is designed around American classification systems.

To compete, she would need access to ITAR-controlled technical data that defines the interoperability requirements. To access that data, she would need to accept American export controls over her product. To accept those controls, she would need American approval for every subsequent sale of a system her company designed, her engineers built, and European taxpayers funded.

The chain of dependency begins before the first bid is submitted.

European defence industry executives report that ITAR classification has become the threshold question in any procurement discussion involving space, cyber, or electronic warfare.

If the programme touches American technology at any point in the supply chain, ITAR applies. A European satellite constellation using a single American-origin component becomes subject to American export controls across the entire system.[7]

ITAR is framed as a security measure protecting American technology.

What it operationally does is protect American market dominance by preventing European companies from developing interoperable alternatives. A European satellite manufacturer that cannot access ITAR-controlled specifications cannot build a product that meets NATO interoperability requirements.

The manufacturer must either license American technology, accept American jurisdiction over the product, or exit the market. Each path leads to the same destination.

Tesat built a factory in Florida. Mynaric went bankrupt. Both were responding to the same structural constraint.

Behind ITAR sits the clearance ecosystem.

4.3 million Americans hold active security clearances. 1.25 million hold Top Secret.

A clearance is a market credential: it determines which contracts a company can bid on, which facilities can host classified work, which engineers can participate. European clearance systems are national, non-transferable, and do not grant access to American classified programmes.

The clearance gap is a structural barrier to market entry that operates beneath procurement rules.[8]

Raj Shah ran the Defense Innovation Unit from 2016 to 2018, then co-founded Shield Capital, a venture firm that invests in dual-use technology companies. His successor at DIU, Michael Brown, joined Shield Capital’s advisory board.

The revolving door between the Department of Defense, the intelligence community, and Silicon Valley venture capital is the pipeline’s human infrastructure. Programme managers who shaped requirements become investors in companies that meet them.

The knowledge of what the government will buy next is the most valuable commodity in defence technology. European defence technology investors have no equivalent access.[9]

The moat works as a gradient. At every stage, the path of least resistance leads to an American provider, an American partner, an American owner. The European company that tries to compete discovers that compliance, classification, clearance, and capital all tilt the same way. The moat makes staying in more expensive than giving up.

The Procurement Cascade

The pipeline’s output does not stay within American borders. It cascades through NATO into European markets (Paper 7).

A European defence ministry identifies a requirement for secure tactical data links.

NATO interoperability mandates compliance with Link 16, a standard maintained by the NATO Communications and Information Agency but implemented primarily through American systems built by L3Harris and Viasat. The ministry issues a tender.

The European competitor offers a technically capable system that has never been tested in a NATO exercise. The American competitor offers a system deployed across fourteen allied nations with ten years of operational data.

The interoperability requirement is formally neutral. Operationally, it selects for the system already embedded in the alliance.

European Foreign Military Sales from the United States jumped from 27.83 per cent of equipment spending in 2019 to 2021 to 50.7 per cent in 2022 to 2024.

The surge in European defence spending after February 2022 flowed disproportionately to American contractors, through NATO interoperability requirements, through familiarity with American systems, and through the simple fact that American companies had production capacity because American government demand had already funded the factories.[10]

The cascade is administered.

The Arms Export Control Act explicitly authorises the Department of Defense to waive nonrecurring research and development costs for Foreign Military Sales. The American taxpayer funds the R&D.

The Department of Defense sells the result to European allies at marginal production cost. The European competitor funded its own R&D and competes at full cost.

The subsidy is invisible because it is laundered through the defence budget.[11]

The Institutional Void

Every institution Europe has built to address the dual-use gap funds supply.

The European Defence Fund (€7.953 billion, less than two years of DARPA alone) funds research but cannot procure. The Hub for EU Defence Innovation coordinates but cannot invest. DIANA validates technology but cannot buy it. The European Innovation Council can take equity stakes in deep-tech startups. It cannot become their customer.[12]

The gap between writing a cheque and placing an order is the gap. Every European institution sits on the wrong side of it.

In September 2024, Mario Draghi delivered a report to the European Commission calling for a European equivalent of DARPA. Eighteen months later, no legislation has been proposed, no institutional design published, no budget allocated.[13]

Every answer addresses the supply side. None creates demand.

US defence R&D spending is approximately $130 billion per year. EU member states combined spend approximately $13 billion. A ten-to-one ratio.[14] But the gap in outcomes is wider than ten to one, because the American money flows through institutions designed to generate customers, while the European money flows through institutions designed to generate reports.

The spending gap matters. The institutional architecture matters more.

Fraunhofer illustrates the paradox. Seventy-five institutes, €3.6 billion in annual revenue, approximately 32,000 employees. The world’s largest applied research organisation.

Its output is prolific. Connection to the commercial market depends on individual researchers forming industry partnerships, technology transfer offices licensing patents, and companies choosing to adopt the results.

There is no institutional customer. There is no SBIR equivalent requiring government agencies to set aside procurement for the technologies Fraunhofer develops. The institute system produces excellent research.

The question that no European institution answers is: who buys it?[15]

European spending also fails to concentrate.

The juste retour principle (Paper 17) distributes contracts proportionally to member-state contributions rather than concentrating behind capability.

In defence, it appears as workshare: the Eurofighter consortium split production across four countries based on order quantities rather than engineering efficiency, and the €100 billion FCAS programme has stalled because Dassault wants eighty per cent of the work and Germany will not accept less than fifty.

The EU operates over 170 weapons systems compared to thirty in the United States.[16]

Buy American does the opposite. It does not allocate contracts to specific states in proportion to their tax contribution. It creates a market floor: any American company can compete, the most capable wins, and an American company wins. Demand concentrates behind capability. The result is SpaceX, Palantir, Anduril.

SPRIND: The Supply Side Works

In December 2019, Germany established SPRIND, the Federal Agency for Disruptive Innovation.

Its founding director, Rafael Laguna de la Vera, modelled the agency on DARPA.[17] The constraints were real.

German state-aid law included a prohibition of betterment: public funds could not improve any recipient’s financial position beyond comparable private transactions. Budget law required line-item appropriation.

Procurement law mandated competitive tendering for every contract. Public-sector pay regulations capped what SPRIND could offer the technical talent it needed to evaluate deep-tech proposals.[18]

Each constraint was reasonable in isolation. Together, they meant SPRIND could not perform the directed-investment function In-Q-Tel performs by default. Germany passed the SPRIND Freedom Act, in force from 30 December 2023. The Act exempted SPRIND from the betterment prohibition, relaxed procurement requirements, and permitted more flexible hiring.[19]

What followed proves the supply side works when given room.

SPRIND has invested in Hypersonica’s €23.3 million Series A for hypersonic propulsion. It backed eleven projects across quantum computing, space propulsion, and advanced manufacturing.

It runs structured “challenges” rather than open calls, builds programme-manager teams that can shape what gets built, and has demonstrated that European institutions are capable of operating in the DARPA register when the legal constraints are lifted.[20]

What SPRIND still cannot do is become the customer.

The Freedom Act gave permission to fund companies. It did not give permission to buy what they build.

The German government remains constrained by the imported procurement directive (Paper 7) from mandating that public-sector buyers prefer SPRIND-funded suppliers. SPRIND can validate the supply side.

The demand side gap remains.

The recursive irony is structural.

SPRIND was inspired by DARPA and modelled on In-Q-Tel. The institutional design was copied.

The European regulatory environment, shaped by the imported constitutional order, prevented the copy from becoming a customer. Europe imported the ideology that makes the American pipeline invisible.

When it tried to build its own pipeline, the ideology it had imported blocked the construction.

The lesson SPRIND teaches is precise. Supply-side innovation institutions are buildable. Demand-side anchor customers are not, under the imported procurement directives. The fix is at the demand layer.

The Customer Already Mandated

The customer is one law amendment away. Or rather, the customer is already mandated by laws Europe has already passed.

DORA, in force from January 2025, requires every European financial institution to compile vulnerability documentation.

NIS2, in force from October 2024, requires every essential and important entity to compile resilience documentation. Public procurement directives govern every government purchase above threshold.

The AI Act requires risk assessments. MiCA requires resilience reporting. GDPR records of processing reveal the data flows of every European organisation.[21]

Mandate sovereign-stack interpretation of these laws and the customer is created. Approximately five thousand DORA-regulated entities migrate (Paper 5). Tens of thousands of NIS2 entities migrate. Twenty-seven member-state public procurement budgets migrate. The aggregate is anchor customer demand on a scale Europe has not produced before.

The mechanics are not hypothetical.

They are the demand-flywheel mechanics that the seven-layer diagnostic (Paper 2) makes operational. A procurement mandate requires sovereignty. The sovereignty stack defines what sovereignty means at each layer. That definition makes the mandate enforceable.

An enforceable mandate creates guaranteed demand. Guaranteed demand makes European infrastructure investable.

The fund mechanics (how pension funds and sovereign wealth follow guaranteed utilisation into European infrastructure) are developed in Paper 26. But the fund cannot exist without the definition.

Without the stack, “sovereign infrastructure” is a phrase politicians use. With it, sovereign infrastructure is a specification investors can underwrite.

The European cloud providers exist. OVHcloud, Hetzner, STACKIT, Proximus, Outscale operate at production scale today. None currently absorbs the workload that this migration would generate. That absence is the missing demand signal.

The crisis is the catalyst.

Without the post-quantum migration deadline, sovereign-stack migration competes for political attention with cheaper short-term priorities and loses. With the deadline, the migration is a compliance obligation that has to happen anyway.

The marginal cost of building European rather than upgrading someone else’s infrastructure will never be lower than during this transition.

NeoNephos, the Linux Foundation Europe cloud sovereignty initiative announced in March 2025, is an early signal. EuroStack’s three pillars (Buy European, Sell European, Fund European) are an early signal. France’s nationalisation of Alcatel Submarine Networks in November 2024, paying a premium to prevent foreign acquisition, is an early signal.[22] The institutional shift is starting.

What is missing is the binding procurement mandate that turns the existing compliance laws into purchase orders.

Conclusion

The pattern reduces to a single failure. Europe funds but does not buy. It produces capability and then leaves that capability to find its own market, in competition with companies backed by the largest procurement machine ever built.

The response, for sixty years, has been to fund more. More grants, more frameworks, more accelerators, more agencies. Each time, the money flows through channels designed to produce reports rather than revenue. Each time, the companies that survive are the ones that found an American customer.

Mynaric had the technology. Tesat had the technology. Kepler had a customer. The company with the customer won the European contract. The company without a customer went bankrupt. The company that built a factory in Florida survived.

The pipeline Europe never built is a customer.

Europe subsidises. America invests and buys.

Europe wrote the laws to do both. Use them.

[1] Prior to the Bayh-Dole Act of 1980 (Patent and Trademark Law Amendments Act, 35 U.S.C. §§200-212), approximately 28,000 patents were held by the federal government, of which fewer than 5 per cent were commercially licensed. In-Q-Tel, the CIA’s strategic investment arm, was established in February 1999 to bridge the gap between commercial technology and intelligence community requirements.

[2] The Bayh-Dole framework operates as a sequence: federal research funding (DARPA, NIH, NSF) → patentable invention → university or small-business retention of rights under Bayh-Dole → spin-out company → first government contract (SBIR, In-Q-Tel) → commercial market. Each stage depends on the next; copying any single stage in isolation does not reproduce the system.

[3] Amendment to the German Employees’ Inventions Act (Arbeitnehmererfindungsgesetz), 2002, transferring patent ownership of publicly funded research from individual professors to their employing universities. Subsequent studies showed measurable declines in patent quality and industry collaboration in the years following the reform.

[4] World Wide Web: Tim Berners-Lee, CERN, 1989. Linux: Linus Torvalds, University of Helsinki, 1991. ARM: Acorn Computers, Cambridge, 1985. Bluetooth: Ericsson, Lund, 1994. MP3: Fraunhofer Institute for Integrated Circuits, Erlangen, 1987-1993. Each was commercialised primarily by American corporations.

[5] EuroStack initiative, launched at the European Parliament conference ‘Toward European Digital Independence: Building the EuroStack’ on 24 September 2024 in Brussels, organised by Cristina Caffarra, Francesca Bria and Meredith Whittaker. Over 80 organisations signed the supporting letter published January 2025. Core pillars: Buy European, Sell European, Fund European.

[6] European Space Agency, HydRON Element 1 prime contract awarded to Kepler Communications Inc. (Toronto) in October 2024. Value approximately €36 million. Tesat-Spacecom and Thales Alenia Space serve as European subcontractors.

[7] International Traffic in Arms Regulations (ITAR), 22 C.F.R. §§120-130, administered by the Directorate of Defense Trade Controls, US Department of State.

[8] Office of the Director of National Intelligence, Annual Statistical Transparency Report. Approximately 4.3 million individuals held active security clearances; 1.25 million held Top Secret clearances.

[9] Raj Shah served as Managing Director of the Defense Innovation Unit (DIU) from 2016 to 2018. He subsequently co-founded Shield Capital, a venture firm investing in dual-use technology. Michael Brown, who succeeded Shah as DIU Director (2018-2022), joined Shield Capital in October 2022.

[10] Bruegel Policy Brief, “Europe’s dependence on US foreign military sales and what to do about it,” 2025. European equipment procurement through US Foreign Military Sales rose from 27.83 per cent (2019-2021) to 50.7 per cent (2022-2024).

[11] Arms Export Control Act, 22 U.S.C. §2761 et seq. The Act authorises the Department of Defense to waive nonrecurring research and development cost recoupment charges for Foreign Military Sales, effectively subsidising allied procurement.

[12] European Defence Fund, Regulation (EU) 2021/697, total budget of €7.953 billion for the 2021-2027 period. DARPA’s annual budget was approximately $4.1 billion in FY2024.

[13] Mario Draghi, “The future of European competitiveness,” report to the European Commission, presented 9 September 2024.

[14] European Defence Agency, Defence Data 2024-2025. US defence RDT&E approximately €129 billion (2023); EU member states combined approximately €13 billion (2024).

[15] Fraunhofer-Gesellschaft, Annual Report. 75 institutes and independent research units, €3.6 billion annual budget, approximately 32,000 employees.

[16] European Defence Agency. The EU operates over 170 different weapons systems compared to approximately 30 in the United States.

[17] Bundesagentur für Sprunginnovationen (SPRIND), established December 2019 in Leipzig. Founding director: Rafael Laguna de la Vera.

[18] The Besserstellungsverbot (prohibition of betterment) under German federal budget law prevented publicly funded entities from offering terms that would improve a recipient’s financial position beyond what comparable private transactions would provide.

[19] SPRIND-Freiheitsgesetz (SPRIND Freedom Act), entered into force 30 December 2023. The Act exempted SPRIND from the betterment prohibition, relaxed procurement requirements, and permitted more flexible hiring.

[20] SPRIND public reporting on portfolio investments. Hypersonica €23.3 million Series A announcement, 2024. SPRIND has run structured technology challenges across quantum computing, space propulsion, and advanced manufacturing.

[21] DORA: Regulation (EU) 2022/2554, applicable from 17 January 2025. NIS2: Directive (EU) 2022/2555, applicable from 14 October 2024. AI Act: Regulation (EU) 2024/1689. MiCA: Regulation (EU) 2023/1114. GDPR: Regulation (EU) 2016/679.

[22] NeoNephos, the Linux Foundation Europe cloud sovereignty foundation, announced March 2025. France’s nationalisation of Alcatel Submarine Networks (Calais, world’s leading submarine cable manufacturer) was completed in November 2024 to prevent acquisition by a non-European buyer.