You Can't Subsidize Your Way Out of Gravity

Ottawa has $750 million to "fix" it — and three lobby organizations fighting over who holds the cheque.

Canada has 10 percent of the world's top AI talent and captures 2 percent of the capital. Ottawa has $750 million to "fix" it — and three lobby organizations fighting over who holds the cheque. All three are asking the government to pay them to do what the market is supposed to do.

The fight, and why it’s the wrong fight

The 2025 federal budget set aside $750 million to address “early growth-stage funding gaps.” The phrase is loose enough that three industry associations have each read it as a mandate to fund themselves. The Canadian Venture Capital and Private Equity Association (CVCA) reads it as growth-stage and later. The National Angel Capital Organization (NACO) reads it as pre-seed and seed, and proposes a $500-million matching program plus $250 million for early-stage “infrastructure.” The newer Canadian Startup Capital Association (CSCA) reads it as emerging managers and asks for a comparatively modest $75 to $150 million.

When one line of budget language produces three separate and disparate interpretations, each routing the money toward the reading association’s own members, that is lobby organizations doing the thing they were hired for. Fine. But what do we — as taxpayers, operators, builders, and citizens — actually need this capital to accomplish? The mandate has no theory of the gap it is closing. It has a pool of money and a queue of self-serving intermediaries lined up to administer it.

Let’s be clear. There is real distress in early-stage financing in Canada. NACO’s own 2026 data shows Canadian angel investment fell to a five-year low last year — $113.79 million across 490 deals, the third straight year of contraction, 57 percent below the 2021 peak. Weeks earlier, the CVCA reported overall venture investment had fallen to a nine-year low in the first quarter of 2026, with growth-stage deals nearly vanishing. Fewer deals, more capital concentrated in fewer of them. The distress is real. The question is not whether something is broken. The question is whether the thing on the table fixes it or feeds it.

The capital is not the problem

It is tempting to say Canadian capital is timid but this is a lazy argument — that it waits for proof while the Americans get in early. That is wrong, and the people who actually run money here deserve better than that slander.

There is good capital in Canada. There are good investors. Version One, Golden Ventures, Mantella VP, Wittington, Inovia — real funds, real discipline, real returns. True North buys secondaries in Canadian and Canadian-founded companies; it did not pretend it could out-cheque a US megafund on a primary round, so it found a different door into the same value. Plaza built a thesis on the exact place Canadian companies stall — the slog from Seed to Series A, where culture and operating muscle, not money, are what’s missing — and brought capital and operators to fix it, taking companies past $500 million in topline and into private-equity outcomes. (Disclosure: I’m an LP in a Plaza fund.)

That is not timidity. That is capable capital inventing differentiated ways to win. And here is what it tells you: these funds compete against far larger, far better-funded US firms for a smaller pool of local deals. The constraint isn’t courage, and it isn’t really dollars. It’s the size of the pond — and the best fish were taken out of it upstream, long before any Canadian investor saw a term sheet. To understand why, stop staring at the capital and look at where the founders are made.

Capital is the exhaust, not the engine

Here is what the founder-drain data actually says, and it is the opposite of what the associations are arguing. The American machine that pulls Canadian builders south does not win on money. The salary premium is real but it is backdrop. Money is the exhaust of this engine — it is how the participants get paid in the future, the off-gas of the combustion. It is not the fuel.

The fuel is ambition, operationalized. The conveyor starts in high school. It takes an eighteen-year-old who cannot yet picture what “making a dent in the universe” even looks like and surrounds them with a culture where going bigger is the water they swim in. It is a farm system, and it works because it offers a path: future potential, validation, a way to operationalize, and — eventually — a way to get paid. It feeds people in the smallest chunks. A high-paying job at OpenAI or Anthropic or Nvidia. A room with a founder who has hundreds of millions in exits. A small grant, a place to live, a problem worth chasing, the right person to meet. Each rung removes one obstacle and reveals the next.

By the time one of these founders raises a Seed or a Series A, the important thing has already happened. They have been activated. The round is downstream of the transformation, not the cause of it. And the activation does not happen in a government-funded accelerator. You cannot administer ambition into existence. You cannot expense immersion.

This is what Jesse Rodgers’ data is screaming. More than 500 US companies founded by someone with a Canadian connection, roughly $414 billion raised, the overwhelming majority with an education link to a Canadian university, more than half founded since 2023. Mila and Bain put the same truth in one devastating ratio: Canada has about 10 percent of the world’s top AI research talent and captures less than 2 percent of global AI venture investment. We build the companies. We make the talent. We just don’t keep the value. That is not a capital shortage. It is the activation, and the capture, happening somewhere else.

It’s not the olds versus the youngs. It’s gravity.

Let me be careful about what I am not saying. This is not the ambitious versus the unambitious, or the young versus the old. And it is emphatically not a shit-on-Canada piece. This is a fantastic country, and we should build here, stay here, and do great things here. We already do. That is how we have Tobi Lütke and Harley Finkelstein and the entire Shopify diaspora. That is Clio in Vancouver, PointClickCare in Mississauga, Ada in Toronto, and all the others. The education system that produces world-class founders is the asset, and some of them stay and build generational companies and seed the next wave.

What pulls the others away is not a character flaw. It is gravity — the simple mass of the US market, its density, its capital, its proximity to people who have done it. You feel that gravity most when you have the least holding you down. At eighteen, with no mortgage, no kids, no house, the things that make Canada saner and safer rank below your own future, because the future is the only asset you have and it is still unwritten. You have opportunity, a paycheque, and the knowledge that you can always come back. Or not. Density wins that argument more often than not, and it is nobody’s fault that it does.

You cannot out-subsidize gravity. The mistake we keep making is believing the government can build a cheap facsimile of the American machine — replicate the place, fund the imitation, and bend the physics. It can’t. What it can do is recognize that we already make the founders, and ask the only question that matters: why don’t we capture more of them?

Go to market

Here is the simplest test for every proposal on the table. Why are we giving lobby organizations public money to do what the market is supposed to do? If the thesis is good, the market funds it. If the market won’t fund it, that is information — not a market failure for the government to paper over. The market test is the entire point: it is the filter that separates “I can allocate capital and return it” from “I can write a white paper saying I could.” All three associations are asking the government to spare them that filter. Run each through it.

CSCA wants to be anointed manager of an emerging-manager fund-of-funds. But a fund-of-funds is something you raise. There is no reason a brand-new coalition of breakaway micro-institutional investors and community organizers should be handed that mandate. If the job is real, it goes to a HarbourVest or a Northleaf — outfits with actual track records — under a new or expanded mandate. Anointing an advocacy coalition as the allocator is not the market; it is the opposite of the market. Go raise it.

NACO wants public money to match private angels two-to-one. Angels deploy their own money — that is the market, already functioning. A 2:1 match means the public takes the majority of the position and risk while the private investor keeps the carry. That is not filling a gap; it is a transfer. And the $250 million “infrastructure” half is regional economic development — programs and events — wearing a federal-innovation costume. If you want to host SaaSNorth in Ottawa or Startupfest in Montreal, those are regional economic-development choices that should stand on their own economics, not a national innovation-capital strategy. It is funding the rinks and calling it player development.

CVCA wants public capital steered to growth-stage institutional managers. This is Canadians Canadianing — the safe, legible, committee-approved thing that makes no difference. Late-stage is the stage with the most available capital; foreign investors already account for 80 to 90 percent of the dollars in $50-million-plus financings. Adding a domestic top-up to the best-funded stage changes nothing about the conveyor or about emerging-manager formation. It is motion without consequence.

Three proposals, one answer. The fight between them was never about closing a gap. It was about which lobby organization gets to skip the market and hold the cheque.

What actually works doesn’t look like this

The things that work here already exist. They just don’t look like more government money to lobby organizations — and notably, every one of them went to market.

TandemLaunch in Montreal is a venture studio, not an accelerator. It builds deep-tech companies from the ground up — sources an idea from frontier academic research, assembles a founding team, injects real capital, and accelerates the company until it can spin out and stand alone. It reverses the conveyor instead of feeding it: rather than nurturing talent from one local university only to watch it leave for San Francisco, it pulls research and talent in from close to a hundred universities around the world and builds the companies in Montreal. It charges no management fee and no carry; the principal makes money only as a limited partner alongside his investors, with the overwhelming majority of capital going into companies rather than overhead. He eats his own cooking.

Inovia and Mila’s Venture Scientist Fund is the other shape of right. It is one of the incumbents — a multi-billion-dollar firm — choosing to build company-creation off the research stack rather than lobby for a matching program. It engages at the earliest moment of company formation, often before a startup formally exists, integrated with the national AI institutes and the universities that actually produce the talent. Valérie Pisano described it as a bridge between the lab and the market where researchers and founders don’t even have to cross the hall. That is the farm system, built where the diamonds are, by people betting their own fund on it.

Notice what they share. They are bets on the research stack and the educational system that creates these founders. They build companies; they don’t administer programs. They engage before the deal exists, not after the risk is gone. They make money only if the companies do. And they raised their capital in the market. None of them is a cheque to an association to help organize a room of angels.

The government keeps picking the wrong seat

There are three seats government can take in venture, and it keeps choosing the two that don’t work.

It can be a permanent LP — passively allocating, no market signal, picking by committee and political mandate. That is the NACO match and the CSCA fund-of-funds: public money in forever, disciplined by nothing.

It can be the GP — investing directly, picking winners itself. That is BDC, and the evidence is damning. In BDC’s own federal review, investors said the agency had become “our primary competitor” and “the Goliath of growth capital in Canada,” and asked it to return to its mandate of complementing the private market by backing existing GPs through funds-of-funds. The feedback was never actioned; BDC is accelerating direct investing instead. Canadian research, meanwhile, finds that firms backed by government-sponsored venture capital underperform their privately-backed peers on IPOs, M&A, and patents — and that government-backed funds displace more effective private investment. And as of this month, the new federal AI strategy doubles down: a Canadian Tech Growth Fund explicitly designed to let the government take direct equity stakes in “the most promising Canadian AI firms,” plus a sovereign wealth fund positioned to back “national champions.” They were told, in February, that the direct-investor seat crowds out the market. In June they announced more of it.

This isn’t only a venture problem; it’s the same instinct running through the whole strategy. As Joseph Fung has argued, Ottawa’s new AI strategy isn’t a failed builder strategy so much as a successful distribution strategy — a “digital welfare state” optimized to make Canadians comfortable users of AI rather than competitive builders of it: literacy programs, certification, watermarking, subsidized adoption, and the state as venture capitalist. The $750-million fight is that instinct in miniature. It is a debate about how to distribute public money among intermediaries, not about how to build companies that can win.

Canada already has the bones of this. The Venture and Growth Capital Catalyst Initiative co-invests in privately-led funds-of-funds and subordinates the public dollar — private investors get their capital back plus a preferred return first, the public takes its share after. That is real discipline, and it is the right instinct. But it stops exactly where Yozma’s genius begins: there is no buyout and no sunset. The public goes in and never comes out — and when the money runs down, we simply run the program again, as we have since 2013. We have built the catalyst without the exit. A subsidy that sweetens private returns in perpetuity is not the same as a one-time bet that builds a private industry and then leaves.

The third seat is the one that works: catalytic, priced, minority, time-limited — and then gone. Israel’s Yozma is the reference, and the lesson is the principle, not the photocopy. Yozma took minority government stakes in privately-led funds with a call option: the private partners could buy the government out at cost plus a modest return. Most did. The state seeded a generation of fund managers and exited, and Israeli venture went from tens of millions to billions in a decade. The discipline is the buyout. The point is the exit. The only seat that catalyzes rather than crowds is the one government refuses to take — because it requires giving up the picking, the ownership, and the announceable stake in a national champion. The politics rewards the photo. The economics rewards the structure that builds managers and disappears. They point in opposite directions, and we keep choosing the photo.

So what would actually help

I won’t pretend I have the precise instrument, because I don’t think anyone honestly does — and false confidence is how we got the proposals on the table. But the test any answer has to pass is clear: does it address the conveyor — the activation and the capture — or does it just move public money through the people who already had it? And the better instruments share a shape: they change private incentives and let private capital allocate on its own judgment, rather than putting the government in a seat.

The cleanest lever is the tax system. Make Canadian capital want to fund emerging managers and the companies they back. A capital-gains treatment for primary investment in early-stage active Canadian businesses — eliminate or sharply reduce the tax on the gain, broadly defined, the way the US treats qualified small-business stock (§ 1202) and the way Build Canada has proposed for Canadian companies valued under $100 million at entry — pulls private money to market without the government picking a single winner. It rewards the outcome, not the allocation. You only benefit if the investment actually works.

One hard-won warning, because Canada has run this play badly before. Do it through the gain, not the purchase. In the 1990s and 2000s, the Labour-Sponsored Investment Fund regime offered a tax credit for buying into labour-sponsored funds — a subsidy on the act of purchase, regardless of whether the fund invested well. The result was bloated, high-fee, underperforming capital that, by at least one well-known analysis, crowded out hundreds of better private investments a year. A credit for buying the fund is the gong show. A lower tax on the gain you earned is the opposite: it rewards real returns, draws private capital on merit, and keeps the government out of the allocation entirely. The same logic applies to the idea of forcing the Maple 8 pension funds to allocate to Canadian venture: a mandate that overrides fiduciary judgment is just a hidden tax on the plan’s growth — paid by the sponsors and future contributors of a defined-benefit plan, or straight out of members’ accounts in a defined-contribution one — and the very need to legislate it is the admission that the managers would not choose the allocation on its merits.

And if the associations want a role — there is a good one, and it isn’t holding the cheque. CVCA has convening power, credibility, and a network. Use it to help emerging managers find market funders: broker the LP relationships, de-risk first-time GPs for private capital, be the connective tissue rather than the conduit for public money. That is a job an industry body is actually built to do. It just doesn’t come with $750 million to allocate.

I am not an angel-group member and I am not lobbying for a slice of this envelope. I have spent twenty-five years in this ecosystem as an operator, an advisor, and a limited partner in early-stage funds — so I have skin in this world, and you should weigh that. But I am also a taxpayer. And as a taxpayer who knows exactly how this machinery works, I find it ludicrous that the public would write a cheque to a lobby organization to do what the market exists to do.

We make world-class founders on a schedule. We keep some — and the ones we keep build extraordinary things. We lose others to a gravity we cannot subsidize away, and we capture far too little of the value our own universities create: ten percent of the world’s AI minds, two percent of the money. We will not close that gap by handing $750 million to an association to run another program, host another event, or anoint itself an allocator. The things that work look nothing like it. If the thesis is good, go to market. If it isn’t, no cheque from Ottawa will make it so.

David Crow is a venture advisor and founder-finance educator with 25 years in early-stage technology as an operator, investor, and advisor. He holds limited-partner positions in early-stage funds, including a Plaza fund referenced above, and is not a member of any of the associations referenced in this piece.

---

References

1. The conveyor / founder drain — Jesse Rodgers, “The Conveyor,” Barn Ventures: https://barnvc.com/the-conveyor

2. Dominion List (US companies with Canadian founders dataset):

https://dominionlist.com/

1. 10% talent / 2% capital; Inovia + Mila Venture Scientist Fund — BetaKit: https://betakit.com/mila-and-inovia-launch-venture-fund-to-turn-ai-research-diamonds-into-startups/

2. Three-association fight / CSCA $75–150M ask — The Globe and Mail: https://www.theglobeandmail.com/business/article-jesse-wiebe-canadian-startup-capital-association-alternative-proposal/

3. CSCA launch — BetaKit: https://betakit.com/former-startup-tnt-leader-launches-coalition-for-canadas-early-stage-investors/

4. NACO “Seeding Growth” proposal: https://www.nacocanada.com/seeding-growth

5. Angel investment five-year low — NACO via GlobeNewswire (May 21, 2026): https://www.globenewswire.com/news-release/2026/05/21/3299653/0/en/Canadian-Angel-Investment-Falls-to-Five-Year-Low-at-113-79-Million-in-2025-while-Women-s-Participation-Hits-a-Record-40-NACO-Reports.html

6. CVCA nine-year low — The Logic: https://thelogic.co/news/naco-canada-angel-investing-2025/

7. BDC venture landscape (foreign capital share at late stage): https://www.bdc.ca/en/about/analysis-research/canadian-venture-capital-landscape

8. BDC direct-investment criticism / “primary competitor” quotes — BetaKit: https://betakit.com/the-feds-asked-investors-for-candid-feedback-on-bdc-it-was-never-actioned/

9. Government VC underperformance / crowding out — Montreal Economic Institute: https://www.iedm.org/government-venture-capital-programs-fail-to-deliver/

10. New federal AI strategy / Canadian Tech Growth Fund — The Globe and Mail (June 4, 2026): https://www.theglobeandmail.com/business/article-canada-ai-strategy-evan-solomon/

11. TandemLaunch — BetaKit (Fund IV close): https://betakit.com/tandemlaunch-closes-37-million-fund-iv-to-create-and-back-deep-tech-startups/

12. Yozma — overview: https://en.wikipedia.org/wiki/Yozma

13. Qualified small-business stock treatment (US comparison) — IRS: https://www.irs.gov/publications/p550

14. Labour-Sponsored Investment Fund failure — Fraser Institute: https://www.fraserinstitute.org/studies/crowding-out-private-equity-canadian-evidence

15. Capital-gains exemption for sub-$100M Canadian companies (QSBS-style) — Build Canada, “Let’s Build the G7’s Fastest Growing Economy”: https://www.buildcanada.com/memos/g7s-best-economy

16. Entrepreneurship decline (57% / formation gap) — Speer & Jackson, The Hub (June 3, 2026): https://thehub.ca/2026/06/03/a-57-decline-the-hunter-prize-for-public-policy-takes-on-the-curious-case-of-canadas-missing-entrepreneurs/