Fellow Signal is an intelligence layer on Activate's fellowship, 224 hard-tech ventures and 292 scientist-fellows (2015–2025), built entirely from public data. Four things the data says:
01
The picks win, independently. $226.3M in federal non-dilutive funding across 46% of ventures, DOE-led, before any equity round.
02
The frontier is moving. the 2025 cohort pivoted into Climate (2% → 32% of the cohort), discovery should track the acceleration.
03
Discovery isn't selection. a founder's research footprint finds scientists years early, but doesn't predict outcomes, use it to find, not to rank.
04
The whitespace is visible. emerging science with fresh federal money and no Activate fellow yet (Gut microbiota and health), the earliest sourcing leads.
Whitespace (bottom-right) = fields whose research is accelerating but where Activate has few ventures yet, an opportunity gap worth investigating. Real data: research momentum from OpenAlex (a large open scholarly index), as each field's growing share of publications; bubble size is federal funding momentum from USAspending.gov; presence is Activate's portfolio. Fields are keyword-defined and point-in-time, so trust the ranking over exact magnitudes.
Insightswhat the data says, and the so-what
Frontier shift
The 2025 cohort pivoted hard
Climate 2% → 32% of the cohortElectronics & Connectivity 8% → 24% of the cohortcohorts of 50 and 38, so directional
So what: Where Activate sources is moving year to year, discovery should track these accelerations, not last cohort's mix.
Whitespace
Hot science, hot money, Activate light
Transportation & Mobility: research ×5.2, federal $ ×99+, presence 7%Energy Storage & Batteries: research ×4.5, federal $ ×15, presence 11%
So what: The clearest sourcing-opportunity gaps: fields the science and the funders are racing into, where Activate is under-represented.
Hub thesis
New York is a climate & carbon cluster
Carbon Management +33pp vs the portfolioClimate +24pp vs the portfolio
So what: Hubs aren't interchangeable, each is a distinct bet, so sourcing and partner cultivation should be hub-specific.
Outcomes
The selections win money on their own
$226.3M in federal non-dilutive funding across 46% of ventures
So what: An independent, public validation that pick-quality converts to real traction, before any equity round.
Model
An equity-free model that 7.2×'d its budget
Revenue $3.7M → $26.7M (2019–2023), funded by philanthropy + government, not VC
So what: A scalable, founder-first alternative to the accelerator model, and the growth shows it's resonating with funders.
Discovery signal
The science predicts the venture, years early
Founders' pre-founding publications map onto their companies, e.g. Ryan DuChanois's membrane / ion-separation research → Solidec's electrochemical reactors.
So what: Research footprints are a leading indicator for founder discovery, surfacing scientists before they've incorporated.
Founder Discovery
The core function: where to source next, and the founder research profile to look for
Spaces ranked by opportunity = research × federal-funding momentum, discounted by how present Activate already is. Top = hot and under-sourced.
1Transportation & Mobilityres ×5.2 · fed ×99+7% in
2Energy Generation & Deliveryres ×3.6 · fed ×978% in
3Built Environmentres ×2.9 · fed ×99+9% in
4Carbon Managementres ×4.1 · fed ×99+17% in
5Computingres ×14.4 · fed ×818% in
6Earth Resourcesres ×3.5 · fed ×259% in
7Climateres ×4.1 · fed ×3726% in
8Energy Storage & Batteriesres ×4.5 · fed ×1511% in
How to read a row
res ×N: how fast the scienceis growing: this field's share of the world's research papers now (2021-24) vs a decade ago (2013-16), from OpenAlex.
fed ×N: how fast federal funding into the space is growing, recent years vs a prior window, from USAspending. "99+" just means very large.
N% in: how much of Activate's recent ventures (the 2021-25 cohorts of companies, not fellows) are already in this space. Amber = under 12%, the thinnest.
Plain version: science booming + money booming + Activate barely there = source here next. The multiples are directional, so trust the ranking over the exact number.
The typical founder, as a filter
Descriptive baselines from 187resolved founders' own research output, a starting screen for candidates, not a bar to clear.
pre-founding research: Rheology and Fluid Dynamics Studies · Geomagnetism and Paleomagnetism Studies · Photovoltaic Systems and Sustainability
Matthew AndersonỌsọ Semiconductor
Steadman Clinic · 12,207 citations · h-index 23
pre-founding research: Hermeneutics and Narrative Identity · Robotics and Automated Systems · Combustion and flame dynamics
Zhiao YuFeon Energy
Devon Energy (United States) · 11,837 citations · h-index 43
pre-founding research: Advancements in Semiconductor Devices and Circuit Design · Advanced Surface Polishing Techniques · Crystallization and Solubility Studies
Haodong LiuTyfast
City University of Hong Kong · 7,911 citations · h-index 36
pre-founding research: Geomechanics and Mining Engineering · Advancements in Battery Materials · Advanced SAR Imaging Techniques
Sourcing implication: in the target spaces above, look for scientists with deep, highly-cited work in the underlying discipline 2-4 years before they incorporate, the research footprint is the earliest discovery signal, ahead of any company.
Selection scorecard
The talent-engine pillar: a per-space candidate screen, and whether the signal tracks outcomes
Look for disciplinesElectrical engineering · Materials science · Biomedical engineering
Research depthin the band of resolved founders (median ~522 citations); deep, cited work that precedes the venture
Training88% of fellows hold a PhD, a baseline, not a gate
A starting screen, not a scorecard to optimize: it points sourcing at the right space and the scientist profile that has worked there.
Does the signal track outcomes? (honest answer: not yet)
Closing the loop: 151 companies with a founder research profile, split at the median citation count, on the rate of winning federal non-dilutive funding.
54%
won federal funding
Above-median research depth
37%
won federal funding
Below-median
At this sample, founder research depth does not predict federal funding, the funded rates are within a couple of points. The average dollars differ ($1.5M vs $485K), but that gap is driven by a few outsized DOE awards, not a broad effect. The honest read: the loop is the right instrument, the predictive signal isn't there yet at this N, and that is worth knowing rather than dressing up.
Fellow background
Who Activate funds, from their own bios: degree level + where they trained (292 fellows)
Growth = a multiple, not year-over-year.Research growth is the field's share of global publications in 2021-24 vs its 2013-16 baseline (×14.7 ≈ 15× its decade-ago level); Federal $ growth compares FY22-25 funding to FY16-19. Research from OpenAlex, funding from USAspending (keyword-matched, directional). Click a space to scope the board.
Emerging Science
Bottom-up: fastest-rising research topics × federal funding momentum, ranked so the hottest unclaimed areas self-rank
The fastest-rising research topics (OpenAlex, incl. arXiv / bioRxiv preprints), crossed with federal funding momentum, and ranked by opportunity = research × funding. Amber = an emerging area with no Activate fellow yet, the sourcing watch list. The top amber rows are research-hot, money-hot, and unclaimed.
Emerging topicResearchFederal fundingActivate
Advanced Battery Technologies ResearchEngineering×3.0×20 · $690.6Min it
Concrete and Cement Materials ResearchEngineering×2.3new $176Min it
Advancements in Battery MaterialsEngineering×1.7×20 · $690.6Min it
Genomics and Phylogenetic StudiesBiochemistry, Genetics and Molecular Biology×4.4×11 · $1.9Bin it
Cancer Genomics and DiagnosticsBiochemistry, Genetics and Molecular Biology×2.2×11 · $1.9Bin it
Advanced Photocatalysis TechniquesEnergy×2.6×9 · $61.2Min it
Electrocatalysts for Energy ConversionEnergy×2.5×9 · $90.5Min it
Topic ModelingComputer Science×4.6no fed matchin it
Gut microbiota and healthBiochemistry, Genetics and Molecular Biology×2.9×4 · $3.7Bopen
Advanced Sensor and Energy Harvesting MaterialsEngineering×3.0×4 · $180.8Min it
Additive Manufacturing and 3D Printing TechnologiesEngineering×2.3×3 · $1.3Bin it
Perovskite Materials and ApplicationsEngineering×3.1$17.4Min it
Network Security and Intrusion DetectionComputer Science×2.9no fed matchopen
Ionosphere and magnetosphere dynamicsPhysics and Astronomy×2.6no fed matchin it
Educational Reforms and InnovationsEnvironmental Science×2.5no fed matchopen
Computational Drug Discovery MethodsComputer Science×2.4$13.7Mopen
IoT and Edge/Fog ComputingComputer Science×2.4no fed matchopen
Online Learning and AnalyticsComputer Science×2.3no fed matchopen
Methane Hydrates and Related PhenomenaEnvironmental Science×2.2no fed matchin it
Land Use and Ecosystem ServicesEnvironmental Science×2.2no fed matchopen
Neural Networks and ApplicationsComputer Science×2.1no fed matchopen
Cancer-related molecular mechanisms researchBiochemistry, Genetics and Molecular Biology×1.9no fed matchopen
Cancer, Hypoxia, and MetabolismBiochemistry, Genetics and Molecular Biology×1.8no fed matchopen
Supercapacitor Materials and FabricationMaterials Science×1.8no fed matchin it
EEG and Brain-Computer InterfacesNeuroscience×1.8no fed matchopen
CRISPR and Genetic EngineeringBiochemistry, Genetics and Molecular Biology×1.8no fed matchin it
Gas Sensing Nanomaterials and SensorsEngineering×1.7no fed matchin it
Adsorption and biosorption for pollutant removalEnvironmental Science×1.7no fed matchopen
Candidate surface for curation, not a ranking to act on blindly. Research is share-normalized (growth above ~8× filtered as a coverage artifact); federal momentum uses curated keywords against USAspending and only counts when current funding is material (≥$20M), so tiny-base ratios can't dominate. 12 of 28 topics are open.
Funder Landscape
Federal funding by agency × space · who's putting money where
Agencies:HHS = Department of Health and Human ServicesNASA = National Aeronautics and Space AdministrationEPA = Environmental Protection AgencyDOD = Department of DefenseDOE = Department of EnergyNSF = National Science FoundationDOC = Department of CommerceDOI = Department of the Interior
Federal obligations by agency × space (USAspending, FY15-25, log scale). Brighter = more money. Click a space to scope the dashboard.
Peer funders
Where The Engine (MIT's deep-tech VC) concentrates vs Activate · click a space to filter
Activate (224 ventures)The Engine (57 companies)share of each portfolio by space
The Engineis MIT's "Tough Tech" venture firm (founded 2016) and invests for equity; Activate is an equity-free nonprofit fellowship. So this compares where two different models concentrate, not their stage or structure. 57 portfolio companies from their public listing.
The Engine's taxonomy is coarser than Activate's (its catch-all "advanced engineering" tag inflates Manufacturing & Robotics), so read this as directional positioning, not a precise overlap. Source: The Engine's public portfolio page, mapped onto Activate's 16 verticals.
Funder & Model
Activate's own finances (IRS 990) · money in vs. impact out
A different model: an equity-free nonprofit, scaled by philanthropy and government rather than venture capital. That $26.7M of annual money-in backs the fellows whose ventures have, in turn, pulled in $226.3M+ of federal non-dilutive funding, before private capital.
IRS Form 990 via ProPublica Nonprofit Explorer · EIN 475502184
Portfolio
Click any venture for its funding, narrative, and founder research footprint