Our report, published in September, has a chapter on the public instruments that built deep tech industries elsewhere and the ones already running across Latin America. The question we get asked most about it is practical: what can a finance ministry in the region appropriate next year, and what would it buy? What crosses borders is the design, and the design is short. A reimbursement ceiling, a public-private split, a co-investment ratio, a milestone trigger, a rule about who sits in the room.

The Parameters Israel Left Behind

Israel’s recovery from the crisis of the mid-1980s rests on three matching funds, and our account comes from IDB Lab’s Deep Tech: The New Wave (2023). The 1985 R&D Law reimbursed up to 50% of corporate R&D, catalysing over 1,000 projects annually and lifting private R&D spending to 4% of GDP. The 1991 Incubator Program leveraged US$600 million to underwrite 24 private incubators, fuelling 1,700 startups with a 40% survival rate and US$3.5 billion in follow-on investment. The 1993 Yozma Program injected US$100 million into ten VC funds, 40% public and 60% private, and the IDB credits it with sparking a domestic venture industry that now manages over US$10 billion and has generated more than US$80 billion in value creation. The IDB attaches no year to that US$10 billion, so quote it as undated.

Read as specifications, three of those are lines a legislature can write: a ceiling of up to 50% of eligible R&D spending, a 40/60 public-private split, ten vehicles rather than one. The pipeline underneath is a design parameter too. Our report cites the BESA Center (2023) for the fact that Unit 8200 alumni alone have launched 1,000-plus start-ups, and Startup Nation Central for the ratio of one startup for every 1,400 citizens. Yozma’s public 40% was buying into deal flow that already existed upstream, so a Latin American instrument has to name its own source. The region has one no comparator can claim: the IDB counts 865,000 STEM researchers across Latin America. A fund sourced from universities and technology transfer offices is a different design, and it is the one available here.

Chile: The Mixed Cohort as a Design Parameter

Start-Up Chile, launched in 2010, is widely recognised as Latin America’s most successful public innovation initiative, and its parameters are public: equity-free US$40k grants plus one-year visas for foreign entrepreneurs who relocated to Santiago, and enforced mixed cohorts of Chilean and foreign teams. On Start-Up Chile’s own materials, 3,000-plus startups from 85 countries have entered, collectively generating over US$1 billion in sales and raising US$1.2 billion in follow-on funding, and 68% of Chilean founders reported changing their financing strategies after engaging with international peers. Alumni include NotCo and Autofact.

The most useful evidence concerns which component did the work. Michael Leatherbee’s 2014 study Boulevard of Broken Behaviors found the US$40k grant useful but non-differentiating: only 16% of foreign founders named it their primary source of value, against 45% of domestic founders. That records what founders named, in a study now eleven years old, so read it as evidence about where each group perceived value rather than as a measured difference in outcomes. Read that way it is usable: the grant was the recruiting price paid to assemble the cohort; the rule that the cohort be mixed was the product, and a composition rule is nearly free to write.

Chile pairs the accelerator with older plumbing. CORFO, the economic development agency, has run matching funds since the early 1990s, and through FONTEC, now under the Innova umbrella, provides reimbursable grants covering 40-65% of private R&D costs, while FONDEF co-finances pre-competitive joint R&D between academia and firms. Startuplab.01, launched by CORFO with Fundación Chile, adds a shared lab and cowork space for deep tech startups focused on climate tech. On EMERGE and Cubo Itaú’s Radar Deep Tech LATAM 2025, presented on 11 September, Chile attracted US$607 million of private deep tech investment in 2024, the largest in the region.

Uruguay Built the Version a Mid-Sized Economy Can Fund

Law 20.075, enacted in 2022 and formally approved in 2023, set Uruguay’s priorities as advanced digital platforms, biotechnology and green tech. In May 2024 the government inaugurated the Uruguay Innovation Hub as a public-private nexus with laboratory space, pilot programmes and mentorship tracks. Concurrent with the launch, it allocated US$10 million for a 1:1 matching-funds programme under UIH’s management: vetted venture-capital firms and angel investors co-invest alongside UIH via convertible notes, with disbursements tied to clear technical and commercial milestones.

It is the most transferable design in the chapter, and it does three things a grant cannot. Convertible notes keep a public claim on the upside. A stated 1:1 ratio means nothing disburses until a private party has priced the risk and committed first. A milestone gate means a venture that stalls stops consuming public money without an official making a discretionary call. US$10 million is also a number most economies in the region can appropriate inside an ordinary budget cycle.

Uruguay surrounds it with the rest of a stack: at least seven active biotech hubs on Endeavour’s and UIH’s counts, twelve free zones supporting 64,000 direct jobs on Uruguay XXI’s count, and INEFOP, with Uruguay XXI, co-financing up to 70% of bespoke company training plans. Instruments compound when they sit beside each other.

Argentina and Brazil: Ticket Size and Counterparty Concentration

Argentina’s 2017 Entrepreneurs Law created FONDCE, a fund of funds that underwrites science-based accelerators and early-stage VCs with reimbursable, interest-bearing loans and partial operating grants. The same law let startups incorporate in under 24 hours. Jenny Lin’s 2021 article in Perspectives on Business and Economics, covering the law’s first two years, records that five FONDCE-backed deep tech accelerators had invested in 79% of Argentina’s investor-funded deep tech ventures, which dates the observation to around 2019. The denominator is ventures that had already raised, so read 79% as how central a public fund of funds became to the deals getting done, not as reach across every venture in the country. Argentina attracted US$486 million of private deep tech investment in 2024 on EMERGE’s Radar.

Brazil runs the deepest stack in the region: PIPE at FAPESP, modelled on the United States SBIR programme; Centelha at FINEP; Catalisa at SEBRAE; EMBRAPII; Mais Inovação, launched in 2024 with non-reimbursable grants and credit for R&D-intensive companies; and FINEP with BNDES offering 500 million dollars to get multinational and domestic firms setting up R&D hubs in Brazil. The parameter Brazilian founders and investors returned to in interviews was ticket size: public grants were described as low-ticket funding, short of what laboratory setup requires, and the fix they proposed was a larger amount per round rather than more rounds. EMERGE’s Radar sharpens the case: of Brazilian deep techs, 47% received no investment at all, 36% rely solely on public funds and 7% received private capital, and Brazil attracted US$216 million of private deep tech investment in 2024.

An outcome multiple is not a policy parameter. A reimbursement ceiling is.

What We Are Still Measuring

Two limits on the evidence above, both of them work we have taken on. The programme figures here come through our report’s cited sources: IDB Lab (2023) for Israel, CORFO’s and Start-Up Chile’s own materials, UIH’s description of its matching facility, Lin (2021) for FONDCE, Leatherbee (2014) for the founder split. They faithfully record what those sources say and we have not yet re-checked them against primary programme documents, so treat them as design references and confirm current terms with the agency. Beyond Leatherbee, no evaluation we have found isolates which component of a programme produced its result, which is why we recommend copying parameters and not outcomes.

Closing part of that gap is what our sixteenth recommendation, a mapping of the region’s funders, is built to do: a public register of who funds what, at which stage and at what ticket.

The Road Ahead: Where the Matching Capital Comes From

Every instrument above needs a counterparty with patient money, and the region already holds one. On OECD figures, pension assets under management across Chile, Mexico, Brazil, Peru and Uruguay total roughly US$1 trillion. Redirecting even 1% would mobilise US$10 billion, which by our own calculation is about four times the US$2.5 billion deployed into Latin American deep tech since 2018. The plumbing has precedents: under the first phase of France’s Tibi initiative, from 2020 to 2022, institutional investors committed €6.4 billion on DG Trésor’s account, and the United Kingdom’s Mansion House Accord of May 2025 pledged 10% of pension provided portfolios to private markets by 2030, with at least 5% ring-fenced for UK assets. Our twelfth recommendation is the Latin American version: an independently managed fund of funds moving pension assets into domestic deep tech through vetted specialist managers.

For anyone drafting a line item, the test is two sentences long. Name the parameter you are copying, and name the trigger that stops the money. A reimbursement ceiling, a 1:1 ratio and a milestone-gated convertible note each give you both, and each conditions public money on private action that has already happened, which is the one capability a thin administration can reliably exercise. Investors get the mirror image: a published ratio, a defined counterparty class and a milestone gate make a co-investor whose behaviour can be modelled.

Geopolitics is moving in the region’s favour, and whether Latin America converts that into a deep tech stronghold will depend less on new money being announced than on money already appropriated being conditioned well. If you are designing one of these instruments, or you have run one, we want the parameters: the ratio, the ceiling, the trigger, the counterparty class. Read the chapter, then write to us with the parameter you would change.