A permanent institution for open chip design infrastructure
Every government developing AI infrastructure, every startup building climate sensors, every researcher designing medical devices faces the same invisible barrier: before they can build anything, they must pay rent to gatekeepers controlling the tools to design chips. We have built a digital economy on freely accessible software infrastructure. Yet the physical hardware running this software remains locked behind proprietary walls, extracting billions in rent and concentrating power in ways that undermine technological sovereignty itself.
Open-source software became foundational infrastructure through a proven pattern: communities demonstrated technical merit, then institutions provided professional support that transformed grassroots projects into backbone infrastructure. Companies collaborate on shared layers while competing on products. Nations gain participation without proprietary gatekeepers controlling terms, pricing, and priorities.
The physical chips running all this software exist in a completely different world. The EDA tools to design chips are controlled by three companies, Synopsys, Cadence, and Siemens EDA, which together command roughly 74% of the global market and charge hundreds of thousands to millions of dollars per licence. Manufacturing equipment is monopolised by firms like ASML, with advanced lithography machines costing up to $380 million per unit.
Each layer has distinct access barriers, but all share the same fundamental problem: infrastructure that should enable broad participation instead functions as an extraction mechanism.
The infrastructure barrier does not just raise costs. It determines what can be built at all.
Climate scientists need specialised sensors for monitoring permafrost degradation, ocean acidification, and methane emissions at scales that current general-purpose chips cannot efficiently address. Precision agriculture faces similar barriers: high initial costs prevent deployment of optimised low-power sensor networks in water-stressed regions where they are most needed. Medical device researchers, disaster monitoring systems, public health infrastructure: each represents technology we know how to build but do not, because infrastructure costs do not align with social value.
Consider: a university research team develops a breakthrough chip architecture that could dramatically reduce power consumption. The design works in simulation. But when a single seat of commercial EDA tools costs $750,000 annually, several times the entire research budget, the chip never progresses beyond the paper stage.
The result is systematic underinvestment in entire categories of socially valuable technology that infrastructure gatekeepers have little commercial incentive to enable.
RISC-V demonstrates both the potential and the limitations of current approaches. As an open instruction set architecture, RISC-V enables anyone to design processors without licensing fees. But designing a RISC-V chip still requires expensive proprietary EDA tools, access to restricted foundries, and navigation of complex intellectual property regimes. The instruction set is open; the infrastructure to use it remains closed.
OpenROAD, a DARPA-funded project to create open-source EDA tools, has made impressive technical progress but struggles with the last-mile problem: getting from research demonstration to production-ready infrastructure that companies will trust for million-dollar tape-outs. The jump from technically possible to production viable requires sustained investment at scales that neither volunteer communities nor academic grants can sustain.
What is needed is not another consortium with membership fees. It is a permanent institution modelled on how infrastructure historically gets built: public investment creating commons that enable broad participation and economic activity, then transitioning to self-sustainability through service revenue rather than rent extraction.
The Open Hardware Infrastructure Works would operate like the institutions that emerged for open-source software: permanent professional operations building and maintaining infrastructure as commons. It would employ engineers as career positions with competitive salaries and long-term job security, starting with hundreds and scaling to thousands as infrastructure matures.
For hardware infrastructure, the same logic that worked for software should apply. Companies need EDA software to design chips, but these tools are not their competitive advantage; their chip designs are. Open hardware infrastructure would enable the same separation: shared tools and processes, competitive products, broadened participation in technological sovereignty.
OHIW would capitalise through public infrastructure bonds with 20-30 year maturities, following a centuries-old pattern. Governments would guarantee the bonds, treating chip design and manufacturing infrastructure like other public goods essential to economic activity and strategic sovereignty.
| Element | Detail |
|---|---|
| Instrument | Government-guaranteed infrastructure bonds |
| Maturity | 20-30 years |
| Scale | Several billion dollars over the first decade |
| Workforce | Hundreds of engineers initially, scaling to thousands |
| Access model | Core infrastructure freely accessible; revenue from enhanced services |
As the infrastructure matures, bonds are repaid through service revenue:
Revenue never gates access to core infrastructure. EDA tools remain freely downloadable. Design platforms remain openly accessible. PDK documentation stays open.
A tripartite commons governance model balances power between three constituencies:
None of this is without precedent. Patient public capital building shared infrastructure that enables broad economic activity is one of the oldest patterns in institutional design.
Dutch Water Boards
Established as early as 1323, financed dyke construction through perpetual bonds. Bonds issued in 1648 still pay interest over 375 years later.
CERN
Founded 1954. Multinational research infrastructure producing open knowledge, from the World Wide Web to particle physics instrumentation.
ESA / Airbus
European aerospace sovereignty through coordinated public investment. Nations that could not individually compete built shared industrial capacity.
Internet (DARPA/NSF)
Long-term government investment, not venture capital. TCP/IP, HTTP, DNS became commons that enabled rather than extracted rent from the digital economy.
GPS
Government-funded infrastructure made freely available globally. Economic value created vastly exceeds what licensing could have captured.
Municipal Bonds (US)
Tax-exempt bonds financing infrastructure since 1913, demonstrating how patient capital invested over decades gets repaid through economic activity enabled.
Open hardware infrastructure addresses the institutional problem: who controls the tools, who sets the terms, who can participate. But it does not address a harder question. The semiconductor industry depends on minerals from the majority world, billion-dollar fabrication, and labour exploitation. Is computation as currently conceived separable from these harms?
The tools exist to fix this. The financing model is proven across centuries of infrastructure development. The governance structures have precedent in successful open-source foundations. The question is not whether this can be done. The question is: who will act?
Open hardware needs institutions as serious as the infrastructure it depends on.