Tank Talks By Ripple Ventures

The episode has genuine pockets of non-obvious thinking - the LLM-as-platform-dependency risk illustrated by the Hootsuite/Twitter API analogy, the indium phosphide/zinc supply chain angle, and the pension capital gap - but the conversation is diluted by significant banter, vague assertions, and meandering topic-hopping that reduces the useful ideas-per-minute ratio.

“The fastest chipset or wafer in the world today is indium phosphide. Indium is a rare earth byproduct of zinc. Canada's got lots of zinc. So we are looking right now at. And by the way, that indium, um, compound, it's a semiconductor compound, 98% of it is largely produced in China.”

“we got $2 trillion of pension capital here that has almost zero, not quite, but almost zero in venture and growth in this country”

The Hootsuite/Twitter API analogy applied to LLM dependency is the standout original frame - it's concrete, earned from experience, and genuinely transfers to current decisions. Most other material (Canadian capital gap, purpose-driven founders, Canada vs. US ecosystem) recycles well-worn VC talking points without adding new angles.

“Twitter basically said uh, we're going, you want access to our APIs, you'll pay a 30% gross receipts tax on the revenues. The margin wasn't 30%.”

“there is full sovereignty and partial sovereignty. And I don't get into that nuance. But what Canada needs to do is find its spots in the chain.”

Ruffalo is a legitimate senior practitioner - decades at OMERS Ventures, early backer of Hootsuite and DuckDuckGo, current Kepler investor - with a genuine track record and institutional depth; he speaks from experience rather than theory. The host also contributes substantively as an active VC, making this a real operator conversation rather than an interview format.

“we had a near death experience with one of my large investments over a decade ago at Omer's called hootsuite”

“DuckDuckGo was really Google 2008. That was the reason why we had done it and we did this six months before Cambridge had come up”

The episode delivers a solid volume of named companies, specific numbers, and real anecdotes - SpaceX revenue, float size, Kepler GPU count, pension capital figures, indium phosphide sourcing percentages - though some figures are hedged or casually thrown out without sourcing, and several macro claims remain hand-wavy.

“The Starlink business is a great business. $20 billion revenue, 60% of revenues.”

“Kepler has 40 Nvidia GPUs spinning a lot. Not a lot. But it's more than zero. Yes, it's more than zero. And right now it's on other folks. And we already are computing the first global app right now on global wildfire protection.”

Matt Cohen occasionally pushes back with genuine conviction - particularly on the double-edged-sword technology debate and the 'this time is different' framing - but most exchanges devolve into mutual agreement and affirmation; follow-ups rarely drill into specifics and several interesting threads (LLM unit economics, Kepler's business model pivot) are dropped prematurely.

“I fundamentally disagree that in technology investments and investing or building technologies, it's always a double edged sword.”

“I'll tell you. I filter for it. I asked them when's the last time they went to Ottawa.”