“Proof-of-concept fund,” “gap fund,” and “translational research program” are used almost interchangeably across universities, but the programs behind those names are not interchangeable in structure. A research administrator advising a faculty inventor, or a technology transfer office (TTO) designing a new internal program, needs to know what actually varies between models — funding size, whether awards go to individuals or institutions, what strings are attached to intellectual property, and what “graduation” from the program is supposed to produce.
This guide compares three real, named program models that between them cover most of the design space university proof-of-concept funding actually occupies: MIT’s Deshpande Center for Technological Innovation, a technology-agnostic, faculty-grant model; Stanford SPARK, a biomedical-specific translational program with an international licensee network; and the broader category of campus-specific “gap funds,” which follow a common structural pattern (seen at institutions like the University of Pittsburgh) even though dollar amounts and eligibility rules differ from campus to campus. This is not a ranking or an endorsement of any one program — it is a structural comparison to help you evaluate or design a program, or advise a faculty member choosing between options.
Where these programs sit in the commercialization pipeline
All three model types sit in roughly the same place in a university’s innovation pipeline: after a technology has been invented — usually after an invention disclosure has been filed with the TTO — but before it is mature enough to license to an existing company or spin out as a new one on ordinary commercial terms. This is the same gap CASRAI’s guide to university innovation accelerator programs describes structurally: cohort-based, milestone-anchored, non-dilutive funding designed to raise a technology’s technology readiness level (TRL) and reduce commercial risk, distinct from NSF’s I-Corps customer-discovery curriculum and distinct from the standard disclosure-to-license workflow a TTO runs on its own. The dictionary term proof-of-concept funding covers the general category; this guide is about the specific, named program models research administrators most often need to compare.
What all three share structurally:
- Non-dilutive. None of these take equity in exchange for funding — awards are grants, not investments, which is the defining feature separating them from venture capital.
- Milestone-based. Funding is released against defined technical and/or commercial milestones, not as a lump sum with no accountability.
- Tied to institutional IP. Funded technology is typically already disclosed to, and owned or co-owned by, the university — funding a proof-of-concept program is not a substitute for filing an invention disclosure with the TTO.
- Time-bound. Awards run for a defined period (commonly 12-18 months) rather than indefinitely.
MIT Deshpande Center: technology-agnostic faculty grants
The Deshpande Center for Technological Innovation, founded at MIT in 2002 with a gift from Gururaj and Jaishree Deshpande, is one of the oldest and most frequently cited university proof-of-concept models in the US. Its structure is built around two sequential grant tiers awarded directly to MIT faculty-led teams:
- Ignition grants — smaller, earlier-stage awards (reported around $50,000) aimed at developing a prototype and doing initial market assessment for a novel technology idea.
- Innovation grants — larger follow-on awards (reported up to roughly $200,000) available to a subset of Ignition-funded projects that have de-risked the initial technical question and need to work through remaining commercial and technical risk before the technology is ready for licensing or spinout.
Alongside funding, Deshpande runs a Catalyst mentor program pairing funded teams with entrepreneurs, venture investors, and industry experts. The center reports cumulative program totals in the range of roughly 200+ funded projects, several dozen resulting spinout companies, and tens of millions of dollars granted since 2002 — exact current figures are published on the center’s own site and change as new cohorts are announced, so treat any specific cumulative number as a snapshot rather than a fixed fact.
Eligibility is MIT-specific: awards go to MIT faculty (or MIT-affiliated principal investigators), not to external applicants or other institutions, and the underlying technology is MIT-owned IP. This is the key structural point research administrators outside MIT need to understand: Deshpande is not a funding source another university’s faculty can apply to — it is a model to study and adapt, not a program to apply into. Several other institutions have built internal programs explicitly modeled on the Ignition/Innovation two-tier structure.
Stanford SPARK: a biomedical translational program with a global network
Stanford SPARK, launched in 2006 within Stanford Medicine, is narrower in technical scope than Deshpande but broader in institutional reach. SPARK funds only biomedical translational projects — therapeutics, diagnostics, and devices — with the explicit goal of moving an academic discovery to the point where it can be licensed to an existing company, licensed to a new startup, or advanced into a clinical trial.
Structurally, SPARK differs from Deshpande in three ways worth flagging for anyone comparing models:
- Funding is comparatively modest per project — reported at roughly $50,000 per project per year — with the program’s real value weighted more heavily toward its weekly, hands-on mentorship from a large volunteer network of industry-experienced advisors (reported at 100+ volunteers) than toward the dollar amount alone.
- It is disease/therapeutic-area focused rather than technology-agnostic, with a stated emphasis on child and maternal health, global health, and orphan (rare) diseases — areas that are frequently underfunded by conventional venture capital because the addressable markets are small.
- It has been deliberately replicated as a network, not just cited as an inspiration. The SPARK model has been formally adopted by dozens of other academic institutions worldwide (reported at 60+ programs across 20+ countries as of the program’s own published figures), each running its own locally-funded SPARK-branded program under a shared curriculum and mentorship framework coordinated with Stanford. This is a meaningfully different adoption pattern than Deshpande’s: a research administrator at another university may be able to affiliate with or launch a SPARK-network program, where affiliating with Deshpande directly is not an option.
SPARK reports strong translation outcomes from its own program history — a substantial share of graduated projects going on to licensing, follow-on grant funding, or clinical-trial entry — though exact success-rate figures are best pulled from the program’s own current published data rather than treated as a fixed benchmark, since they are cumulative and updated as new cohorts graduate.
University-specific gap funds: the generalized, adaptable model
Outside these two named, long-running programs, most research universities that run any proof-of-concept funding at all use what is generically called a “gap fund” or “commercialization fund” — smaller, campus-specific programs that follow a recognizably common pattern without being formally branded or networked like SPARK. A representative structure, drawn from a real, publicly documented example (the University of Pittsburgh’s Commercialization Gap Fund, run through its Office of Innovation and Entrepreneurship) illustrates the pattern:
- A primary award tier (commonly in the tens of thousands of dollars) plus a smaller microgrant tier for earlier-stage or lower-cost validation work.
- An annual or semi-annual RFP cycle rather than rolling admissions.
- A short window to start work after award (often within a month) and a defined completion window (often around 12 months).
- Funding tied explicitly to 1-3 technical milestones plus 1-2 business-development milestones, reviewed at defined checkpoints rather than released as a lump sum.
- A strong preference or requirement that applicants have already completed a regional NSF I-Corps cohort before applying, treating I-Corps and the gap fund as sequential rather than redundant.
- A hard eligibility requirement that the underlying technology already have a filed invention disclosure and be unencumbered university-owned IP — i.e., the gap fund sits downstream of the TTO’s disclosure intake, not upstream of it or in place of it.
The specific dollar amounts, cycle timing, and milestone counts vary meaningfully by institution — treat any single campus’s figures as illustrative of the pattern, not as a number to quote for a different university. What’s structurally stable across most campus gap funds is the shape: RFP-cycle, dual-tier award sizes, milestone-gated disbursement, and an I-Corps-then-gap-fund sequencing.
Comparing the three models
| Dimension | MIT Deshpande Center | Stanford SPARK | Typical campus gap fund |
|---|---|---|---|
| Technology scope | Technology-agnostic (any field) | Biomedical only (therapeutics, diagnostics, devices) | Usually technology-agnostic, campus-defined |
| Award structure | Two tiers: Ignition (~$50K) then Innovation (~$200K) | Single tier, ~$50K/year, renewable while active | Usually two tiers: primary award + smaller microgrant |
| Who can apply | MIT faculty/PI only | Stanford Medicine-affiliated faculty; independently, a SPARK-network program at another institution has its own local eligibility rules | Faculty/researchers at the sponsoring institution only |
| Support beyond funding | Catalyst mentor network (entrepreneurs, VCs, industry/legal experts) | Weekly coaching from a large volunteer industry-expert network | Varies; often lighter-touch than Deshpande/SPARK |
| Replicable elsewhere? | Model widely studied/adapted, but the program itself is MIT-only | Formally networked — other institutions can run an affiliated SPARK program | Not networked; each campus builds and names its own independently |
| Typical pre-requisite | Not formally required | Not formally required | Often requires or strongly recommends completing NSF I-Corps first |
What research administrators and TTOs should take from this
If you are advising a faculty inventor at MIT or Stanford, the answer to “which program” is usually determined by institution, not by choice — apply to the program your own university runs. The comparison matters more for research administrators at other institutions evaluating how to build or improve an internal program: Deshpande demonstrates a technology-agnostic, two-tier, faculty-direct model with a strong mentor network; SPARK demonstrates a narrower, disease-focused model that has been deliberately built into a replicable, joinable network rather than a single-campus program; and the generic campus gap fund shows the minimum viable structure most institutions can stand up without either program’s scale — milestone-gated, dual-tier, sequenced after I-Corps and after invention disclosure.
In every case, funding from one of these programs does not change the underlying intellectual-property allocation: the university’s rights and obligations under the Bayh-Dole Act (for federally funded inventions) and the terms of any eventual license are governed separately from the proof-of-concept award itself. A proof-of-concept grant is evidence-building funding, not a licensing decision — see CASRAI’s guide to Bayh-Dole march-in rights for how those underlying IP obligations work, and the guide to funding options for a university spinout for how gap-fund money compares to SBIR/STTR, licensing revenue, and venture capital as a startup moves further along the pipeline.
Frequently asked questions
Can a researcher at another university apply directly to MIT’s Deshpande Center or Stanford SPARK?
Not to Deshpande — its grants go to MIT-affiliated faculty only. SPARK is different: because the SPARK model has been formally replicated as a network, faculty at one of the dozens of institutions running an affiliated SPARK-branded program can apply to their own campus’s SPARK program. Neither program accepts direct applications from researchers with no institutional affiliation to the program.
Is a gap fund the same thing as NSF I-Corps?
No. NSF I-Corps is a structured customer-discovery training curriculum (interviews, hypothesis testing) funded by NSF, typically covering travel and time rather than product development. A gap fund provides money to actually build and de-risk a prototype or dataset. Many campus gap funds require or strongly recommend I-Corps completion first, treating the two as sequential steps rather than substitutes for each other.
Does receiving proof-of-concept funding mean the university automatically licenses the technology to the funded team’s startup?
No. Proof-of-concept funding is evidence-building money aimed at reducing technical and commercial risk; it does not itself grant or guarantee a license. Licensing terms — exclusive or non-exclusive, royalty structure, equity — are negotiated separately with the TTO, informed by, but not decided by, the outcome of the proof-of-concept work.
Do these programs take equity in exchange for funding?
No. Deshpande, SPARK, and the great majority of campus gap funds are structured as non-dilutive grants, not investments — this is one of their defining structural features relative to venture capital or accelerator programs that do take equity.
