Full comparison
Every structure evaluated across the same criteria — from organisational design and funding horizon to talent attraction and risk tolerance. Scroll right to see all nine. Academia and Industry are included as a baseline.
| AcademiaDecentralised PI model | IndustryCorporate R&D | Industrial LabsBell Labs model | Government LabsNational mandates | Core-funded Nonprofit InstitutesRedesigning research incentives | Focused Research OrganizationsScientific infrastructure labs | BBNsFrontier Research Contracts | ARPA-style OrgsProgram manager–led R&D | |
|---|---|---|---|---|---|---|---|---|
| Examples | Majority of university research labs. Also many nonprofit research institutes. | Pharma, tech companies; VC-backed startups; SBIR-funded small firms | Bell Labs, Xerox PARC, DuPont Experimental Station, GE Research, IBM Research, Ink and Switch | CSIRO, Battelle Memorial Institute, US DOE National Labs, Fraunhofer Society, RIKEN | Arc Institute, CZI Biohub, Francis Crick Institute, Howard Hughes Medical Institute, Broad Institute | E11 Bio, Cultivarium, Parallel Squared Technology Institute (all via Convergent Research) | BBN Technologies, SRI International, MITRE Corporation | DARPA, ARPA-E, ARPA-H, IARPA, ARIA (UK), SPRIND (Germany), Speculative Technologies, Actuate, Wellcome Leap |
| Description | Decentralised system of PI-led labs funded by project-based grants | Research embedded in companies and startups, tied to products and commercial returns | Corporate-funded labs operating with research autonomy from business units — historically the most productive research environments ever created | Publicly funded labs solving national and strategic problems with long-term mandates | Institutes that intentionally redesign incentives and funding, by replacing competitive project-based grants with no-strings-attached funding for their research. | Independent, standalone orgs built to execute a specific scientific or technical goal with a unified full-time team | Contract-based networks that assemble elite teams across institutions to tackle defined frontier problems | Agencies that fund and coordinate external research teams toward ambitious, high-risk goals via empowered programme managers |
| Org structure | Highly decentralised; PI autonomy; loose coordination between labs | Hierarchical with cross-functional teams; product-aligned divisions | Semi-centralised; research director with autonomous lab groups embedded within (but distinct from) corporate hierarchy | Hierarchical with programmatic divisions; government oversight; shared infrastructure | Hybrid: PI autonomy + strong central platforms and shared infrastructure | Centralised, in-house teams with unified leadership. Core team does all research design. Internalised research enables tight iteration loops. Ideally autonomous from funding source. | Distributed teams coordinated through contracts, often spanning multiple institutions. Core team coordinates — external orgs execute. Enables parallel tracks without internal politics. | Centralised programme managers coordinating external performers. Single PM is the "core team." External research enables parallel tracks, flexible recombination, and access to specialised expertise without hiring. |
| Funding horizon | Short–medium: 2–5 year grants; can chain via tenure | Short–medium: months to 3-year runway | Long when corporate will maintained (decades); vulnerable to ownership changes and strategy shifts | Long: 5–20+ years; stable institutional funding | ~8 years per programme | 3-7 years (org lifespan tied to mission). Deploys ~$10M in year one, ~$50M over lifespan. Shuts down on mission completion. | Short–medium: 1–5 years per contract; can chain into longer programmes | 3–5 years per programme. PARPA-style programs ramp slowly (from ~$200K over 6 months up to FRO-level funding) with go/no-go gates. |
| Outputs | Research papers, reviews, trained researchers | Products, IP, patents | Foundational technologies, patents, breakthrough papers (transistor, laser, Ethernet) | Technologies, national capabilities, shared infrastructure | Platforms, translational outputs, datasets, papers | Tools, datasets, platforms, scientific infrastructure — public goods, not products | Capabilities, prototypes, strategic outputs (e.g. biosecurity tools, datasets, frontier systems) | Prototypes, validated capabilities, proof-of-concept demonstrations; expects transition pathways to other sectors |
| Research problem type | Early-stage, exploratory, curiosity-driven science | Product-aligned, commercialisable problems | Mix of basic and strategically applied; defined by adjacency to parent domain but with latitude for pure research | National priorities (energy, defence, health, agriculture) | Important but underexplored scientific areas; platform-building in emerging fields | Clearly scoped bottlenecks requiring tight coordination. Clear goal from day one. Tight coupling between workstreams. Strong conviction about the right approach early on. | Frontier, mission-driven problems requiring flexible, high-end expertise across organisations. Often engineering-heavy and deliverable-driven. | High-risk "white space" problems. High-risk, high-reward with defined programme goals but no clear best approach. May take 6–12 months to nail down the goal. |
| Risk tolerance | Medium Constrained by grant cycles, journal incentives, career risk |
Low–Medium Risk tolerated only if tied to plausible commercial upside |
High Historically very high; corporate capture eventually reduces over time |
Medium–High Can take ambitious problems within government mission constraints |
High Philanthropy enables risk-taking and scientific creativity. Long-term funding removes short-term pressure. |
Very High Purpose-built for high-risk execution; explicit tolerance for failure |
Very High Strong; flexible contracting enables unusual collaborations |
Very High Explicitly funds "no one else will fund this" work; ~85% of DARPA projects fail to meet original goals by design |
| Predominant funding source | Government grants (NIH, NSF, ARC etc.) | Revenue / VC; sometimes government (SBIR) or philanthropy (MRIs, PRIs) | Corporate (parent company) | Government appropriations | Philanthropy + endowment | Philanthropy (typically) or hybrid public-philanthropic | Government and/or industry contracts | Government |
| Key weaknesses | Fragmentation; short timelines; weak coordination across labs | Short-termism; commercialisation bias; limited public-good research | Dependent on parent company; nearly extinct as a model; corporate short-termism eventually dominates | Bureaucracy; slower iteration; constrained by political mandates | Not fully integrated; still PI-based; coordination is improved but not transformed | Fundraising is a key bottleneck ($20–100M required). High upfront cost. Difficult to set up. Requires strong leadership and careful problem selection. | Ongoing financial risk (no guaranteed funding like FROs). Coordination complexity. Reliance on external performers. Risk of becoming a pure R&D contractor (Battelle/Raytheon failure mode). | No in-house execution; depends on ecosystem quality; PM turnover risk; success requires exceptional programme managers |
| Talent attraction & collaboration | Strong within discipline. Weaker for integrated, cross-functional teams — incentives are siloed. Wide collaboration but slow coordination. | Strong for execution-focused, applied cross-functional teams. Less appeal for pure theorists. | Historically extremely strong — offered stability, resources, peer community, and academic freedom simultaneously | Strong for large-scale, multidisciplinary, infrastructure-heavy work. Less fluid than startups or FROs. | Very strong; curated and stable environment with long-term backing | Extremely strong — mission-driven, full-time, cross-functional teams. Purpose-built to overcome academic silos. | Very strong — can pull top talent across institutions; but less cohesion than in-house teams | Very strong at orchestrating talent across academia, industry, and labs. Not a home for talent but a network hub — exceptional at activating and connecting performers. |
| Ambition level | Medium Risky ideas possible but constrained by grant cycles and career incentives |
Low–Medium Truly speculative work rare; risk tied to commercial upside |
Very High Historically unconstrained; eroded over time by corporate context |
Medium–High Ambitious within government mission constraints |
High Long-term funding removes pressure; PI freedom maintained |
Very High Purpose-built diverse teams under one roof; designed to overcome silos |
High Flexible contracting enables unusual collaborations; execution incentives can be difficult |
Very High Explicitly funds "no one else will fund this" — highest ambition tolerance of any structure |
| Tech dev vs. commercialisation | Predominantly basic science. Limited commercialisation pressure; translational work often underfunded. | Translational / commercialisable only. Mid-stage tech pursued only if it de-risks product roadmap. | Long-run technology development; low immediate pressure, but corporate connection creates eventual commercialisation pull | Mix of applied and fundamental; less pressure for commercial returns, more for mission outcomes (energy, defence) | Technology development without immediate commercialisation pressure | Technology development as public goods, platforms, and datasets. No commercialisation pressure. | Capability creation rather than products; low immediate commercial pressure; oriented toward strategic outcomes | Breakthrough prototypes and proof-of-concept; no direct commercialisation pressure but expects transition pathways to military, commercial, or government sectors |
| Past research examples | Gene discovery, theoretical physics, foundational chemistry | Drug development, AI products, semiconductor manufacturing | Transistor, information theory, Unix, C language, laser, solar cell, GUI, Ethernet, laser printer | Nuclear energy, climate modelling, agricultural genomics | Cell atlases, biological platforms, single-cell sequencing | Whole-cell modelling, biological atlases, new measurement technologies | Biosecurity infrastructure, rapid response platforms, high-containment biology capabilities | Internet, GPS, stealth technology, mRNA vaccine platforms |
| Where This Structure Excels | Early-stage, exploratory science with no defined application | Productisable technologies with clear market. Profitable within 3-5 years. | Long-run basic and applied research requiring sustained resources and freedom — now rare; modern analogue may be philanthropically-funded industrial-style labs | National-scale infrastructure problems; defence technologies; strategic energy | Deep domain scientific questions requiring long-term investigator focus | Brain mapping at scale; bespoke biotech instrumentation; new measurement pipelines; research software and datasets; pilot-plant-scale work; translation of academic ideas to non-venture markets | Pandemic response systems; novel bio-defence capabilities; rapid assembly of elite cross-institutional teams around defined frontier deliverables | High-risk, cross-domain breakthroughs needing coordination across many performers; problems where the right approach is genuinely unknown at the outset |