In this episode of Tank Talks, recorded live at the Global Startups Conference, Matt Cohen and John Ruffolo take the stage for a wide-ranging conversation on the future of Canada’s innovation economy, the AI infrastructure race, and why this moment feels different, even if John still refuses to fully believe “this time is different.” The conversation opens with SpaceX’s historic IPO, massive valuation hype, and the question of whether public market demand can support a new wave of AI and frontier tech giants like OpenAI and Anthropic.
Matt and John then dig into one of the biggest strategic questions facing founders today: should startups build on top of frontier AI models, or will those platforms eventually come for their margins? John draws a sharp comparison to Hootsuite’s dependence on social media APIs, warning founders not to build businesses where a monopolistic platform can eventually “come calling.” From LLM unit economics and inference costs to local models, edge compute, AI sovereignty, and Canada’s weak position in the full AI stack, this episode breaks down why real moats may come from deep tech, defense, energy, chips, space infrastructure, and hard-to-build businesses.
The discussion also tackles Canada’s AI strategy, the tension between innovation and regulation, the rise of dual-use defense startups, the shortage of domestic growth capital, and whether Canada is becoming a farm team for U.S. acquirers. John and Matt close with a candid look at family offices, immigrant founders, Canadian ambition, and what actually separates fundable founders from the noise: purpose, focus, and the ability to build something hard when everyone else is chasing the latest shiny object.
SpaceX’s IPO and the return of the hype machine (02:48)
Matt and John open with the massive SpaceX IPO, its soaring valuation, and whether the market is being driven by fundamentals or pure scarcity-fueled hype. John argues that discounted cash flows still matter, even when investors are caught up in the next great frontier tech story.
Satya Nadella’s warning to AI founders (05:56)
Matt brings up Satya Nadella’s warning about relying too heavily on frontier models. The discussion explores why businesses built on top of OpenAI, Anthropic, or other LLM platforms may eventually face direct competition from the very infrastructure they depend on.
Canada’s AI strategy: long overdue, but too unfocused? (16:14)
Matt and John assess the government’s AI strategy and the promise that Canadian AI adoption could add massive GDP growth. John says the strategy contains useful objectives, but risks becoming a laundry list without a clear answer to the question: which pedal are we actually pressing?
Building trust in AI without creating regulatory capture (21:46)
The audience asks how Canada can build trust in AI adoption. John argues for clear guardrails, but warns that large AI players may eventually welcome heavy regulation because it protects incumbents and locks out smaller competitors.
Defense tech is hot again, but not every startup is real (25:19)
Matt and John discuss the surge of interest in dual-use defense technology. John warns that when government money appears, everyone suddenly claims to be a defense company, making it harder to separate serious builders from PowerPoint tourists.
Is building in Canada patriotic or financially irrational? (33:19)
Matt asks the blunt question: in 2026, is staying in Canada a patriotic endeavor or a financial mistake? John argues Canada has the talent, ecosystem, and raw materials, but lacks confidence and ambition at the capital layer.
Why Canada needs real growth capital, not just early-stage funding (37:34)
John explains why he created Mavericks to address the gap in Canadian growth equity. The issue is not founder ambition, but the lack of domestic capital willing to write meaningful checks once companies need to scale past the early stage.
Family offices, education gaps, and Canada’s missing innovation capital (43:56)
Matt explains why many Canadian family offices are still learning how venture and startup investing work. Unlike real estate or private equity, venture requires patience, a tolerance for the J curve, and a different understanding of risk and return.
Canada’s AI edge may be hiding in resources, minerals, and chip substrates (49:43)
The episode closes with a discussion of Canada’s possible edge in AI infrastructure through natural gas, rare earth materials, zinc byproducts, indium phosphide, and semiconductor supply chains. Matt and John argue that Canada’s issue is not a lack of resources, but a lack of permission, capital, and long-term conviction to build around them.
Connect with John Ruffolo on LinkedIn: https://ca.linkedin.com/in/joruffolo
Connect with Matt Cohen on LinkedIn: https://ca.linkedin.com/in/matt-cohen1
Visit the Ripple Ventures website: https://www.rippleventures.com/
