The Rundown 3/13/26: Canada’s Defence Tech Push, Constellation’s AI Test, and the Private Credit Mess

In this episode of Tank Talks, Matt Cohen and John Ruffolo unpack a volatile moment across software, capital markets, AI, and Canadian industrial policy. The conversation opens with Constellation Software’s AI-era challenge, as new president Mark Miller faces investor skepticism around whether legacy vertical market software can maintain its moat in a world increasingly shaped by AI-driven productivity, automation, and code generation.

From there, Matt and John examine Salesforce’s decision to raise billions in debt to fund share buybacks, questioning whether this is smart balance-sheet engineering or a red flag that large software companies are running out of offensive growth options. The episode then turns to the private credit market, where redemption gates, liquidity pressure, and fears around AI infrastructure lending raise deeper concerns about leverage, accounting, and systemic fragility.

Back in Canada, the discussion shifts to the country’s defence industrial strategy and why the real opportunity is not just traditional military spending, but dual-use investment across AI, quantum, satellites, aerospace, and strategic infrastructure. The episode closes with a look at Andrej Karpathy’s open-source Auto Research project and what it signals about the speed of AI progress, the democratization of research capabilities, and the growing pressure on knowledge workers and software engineers to keep up.

If software moats are weakening, private credit is wobbling, and defence dollars are becoming innovation dollars, where will the next real edge come from?

Constellation Software, AI Pressure, and the Future of Vertical SaaS (00:43)

Matt and John break down Constellation Software’s latest numbers, the market’s growing skepticism toward legacy software businesses, and the bigger question of whether mission-critical vertical SaaS can stay resilient as AI chips away at traditional moats. They explore why trusted workflows and proprietary data still matter, but also why even durable software businesses may face long-term pressure.

Salesforce’s $25 Billion Debt Bet and What It Really Signals (06:28)

Matt and John unpack Salesforce’s plan to raise massive debt for share buybacks, debating whether this is efficient capital structure management or a defensive move from a software giant with fewer compelling growth opportunities. The bigger issue is what this says about confidence, capital allocation, and the mood inside mature SaaS companies right now.

Private Credit Redemption Gates and the Fear Beneath the Surface (10:49)

A wave of redemption limits across major private credit funds becomes the next flashpoint. Matt and John explain why retail money flooded into the asset class, how managers were pushed into riskier lending, and why the underlying concern is no longer just liquidity management, but whether private credit has been pricing equity risk like it was safe debt.

Canada’s Defense Strategy Is Really a Dual-Use Tech Strategy (16:29)

Matt and John shift to Canada’s defense industrial strategy and the National Research Council’s planned investment, arguing that the real opportunity is in dual-use innovation. Rather than thinking only in terms of tanks and submarines, John reframes defense spending as investment in AI, quantum, satellites, aerospace, and strategic infrastructure that can serve both government and enterprise customers.

The AI Catch-Up Panic Is Real (21:26)

Matt and John zoom out from markets and policy to the personal reality of AI acceleration. John admits he feels both energized and behind, capturing the exact tension many operators and investors feel as new tools emerge faster than most people can realistically absorb them.

Andrej Karpathy Auto Research and the One-GPU Research Lab Moment (22:58)

The episode closes with Andrej Karpathy’s open-source Auto Research project and why it matters. Matt explains how autonomous research loops, overnight experimentation, and low-cost GPU access could dramatically speed up model tuning, product testing, and AI development, making advanced experimentation far more accessible than before.

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/

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit tanktalks.substack.com