A public reference

Alternative Structures for Research Institutions

When most people think about where research happens, they picture universities or industry. But a growing ecosystem of alternative structures — from ARPA-style agencies to Focused Research Organizations — is filling the gaps neither can reach. This site maps them.

Background

Academia and industry are both essential — and both incomplete

The world's research enterprise has long been anchored by two dominant institutions. Universities produce the majority of basic science: curiosity-driven, peer-reviewed, and publicly funded. Industry — from large corporate R&D to VC-backed startups — drives applied research and translates discoveries into products. Both have produced extraordinary results. Both are also structurally limited in ways that create significant blind spots.

Academia

Strengths

  • Long tradition of basic, curiosity-driven research
  • Strong talent pipeline and training
  • Freedom from commercial pressure
  • Wide collaborative networks across disciplines
  • Oriented toward public goods and open knowledge

Structural weaknesses

  • Fragmentation — PI autonomy limits coordination
  • Short grant cycles (2–5 years) misaligned with long problems
  • Publication incentives crowd out tool and infrastructure building
  • Career incentives discourage high-risk, long-horizon bets
  • Poorly suited to engineering-heavy, execution-focused work

Industry & Startups

Strengths

  • Fast-moving, cross-functional execution
  • Strong engineering and product talent
  • Commercially viable research gets resourced quickly
  • Clear accountability and milestone culture
  • Access to proprietary data and infrastructure

Structural weaknesses

  • Short-termism — months to 3-year runway constraints
  • Commercialisation bias: non-profitable research starved
  • Speculative or basic science rarely funded
  • Outputs are IP, not public goods
  • Market logic incompatible with public-good research

The gap

What falls through

Plotting research institutions on two axes — basic vs. applied research, and short vs. long time horizons — reveals a structural gap. Academia occupies the upper-left: basic, relatively long-term. Industry occupies the lower-right: applied, short-term. The upper-middle and upper-right — applied-but-not-commercial, long-horizon, requiring coordination — are chronically underserved.

This is where frontier research lives: building tools and datasets that accelerate entire fields, tackling engineering problems too large for a single lab and too public-good-oriented for a startup, and pursuing high-risk bets that require years of coordinated effort.

ALTERNATIVE STRUCTURES OPERATE HERE Basic Research Applied Research TYPE OF RESEARCH Short-term Long-term TIME HORIZON Academia Decentralised PI model Industry / Startups Corporate R&D Industrial Labs Bell Labs, Xerox PARC Govt / Mission Labs CSIRO, DOE National Labs Nonprofit Research Institutes Max Planck, Broad Institute Next-Gen Institutes Arc Institute, CZI Biohub FROs Convergent Research BBNs / FRCs Contract frontier labs ARPA-style Orgs DARPA, ARIA, SPRIND

Positions are approximate and reflect typical operating mode. Industrial Labs are shown as a historical reference — the model is rare today.

The landscape

Seven alternative structures

The following structures represent distinct approaches to organising, funding, and governing research outside of conventional academia and industry. They differ along a key dimension: the degree of coordination they enable — from loosely federated institutes to tightly integrated mission-driven teams executing toward a single defined goal.

↑ Coordination increases from top to bottom.

Industrial LabsCorporate-funded, research-first
Corporate-funded research labs with significant autonomy from business units. Historically among the most productive research environments ever created — Bell Labs alone produced the transistor, Unix, C, and information theory. The model is rare today, eroded by corporate short-termism, but remains a critical archetype for what sustained, well-resourced research freedom can produce. Examples: Bell Labs, Xerox PARC, DuPont Experimental Station, GE Research, IBM Research, Ink and Switch.
Government & Mission-Oriented LabsPublicly funded, national mandate
Publicly funded institutions with long-term mandates tied to national missions — defence, energy, health, agriculture. Medium-to-high coordination, decades-long time horizons, strategic applied research. Own and operate shared infrastructure (accelerators, supercomputers, containment facilities) that no individual institution could sustain. Examples: CSIRO, Battelle Memorial Institute, US DOE National Labs, Fraunhofer Society (Germany), RIKEN (Japan).
Nonprofit Research InstitutesMission-driven, grant-funded
Independent institutes housing multiple PI-led labs. Operate like highly coordinated, very large university research departments — without the teaching mandate. Strong within-domain scientific culture. Philanthropy can enable risk, but funding is often still conservative and PI silos remain. Examples: Max Planck Society, Scripps Research, Francis Crick Institute, Howard Hughes Medical Institute, Broad Institute.
Next-Gen InstitutesPhilanthropic, redesigned incentives
A newer generation of institutes that intentionally redesign incentives, funding structures, and shared infrastructure while retaining PI-like autonomy. Funded at ~8-year horizons, with strong shared platforms enabling more coordination than traditional institutes — but not yet fully integrated teams. Built on the premise that excellent scientists need better conditions, not different people. Examples: Arc Institute, CZI Biohub.
Focused Research OrganisationsMission-driven, time-bound, in-house
Standalone organisations built to execute a specific technical goal with a unified team. CEO-led, full-time cross-functional staff, milestone-driven — pursuing public goods rather than products. Time-bound (~10 years): once the mission is complete, the FRO shuts down. FRO-shaped problems need a clear goal from day one, tight coupling between workstreams, and strong conviction about the right approach. Examples: E11 Bio, Cultivarium, Parallel Squared Technology Institute (all via Convergent Research).
BBNs / Frontier Research ContractsContract-based, distributed elite teams
Contract-based networks assembling elite teams across institutions to tackle defined frontier problems. Named after the original Bolt Beranek and Newman lab, which used government contracts to build foundational internet infrastructure while pursuing bolder agendas on the margins. More financially precarious than FROs but more flexible. Ideal for engineering-heavy work too complex for academia and too public-good-oriented for industry. Examples: BBN Technologies, SRI International, MITRE Corporation.
ARPA-style OrganisationsHigh-risk, programme-managed, external
Agencies that fund and coordinate external research teams toward ambitious goals. Defined by empowered programme managers, milestone-driven portfolio management, and explicit tolerance for failure. Suited to problems where the right approach is unknown — ARPA programs run parallel tracks to test multiple approaches simultaneously, pruning and recombining without organisational disruption. Programmes typically run 3–5 years. Examples: DARPA, ARPA-E, ARPA-H, ARIA (UK), SPRIND (Germany), Speculative Technologies, Actuate, Wellcome Leap.
Compare all structures → Browse the directory