20MinuteLeaders

The episode contains a handful of genuinely useful concepts - store-now/decrypt-later, crypto agility, and the distinction between large quantitative models vs. GenAI - but is padded with background scene-setting, biographical detours, and high-level optimism that dilute the useful insight-per-minute ratio.

“one in five breaches come because of a credential that was lost or not updated at the right time”

“Words are really good, but maybe not enough. We need to bring chemistry. We need to bring physics. We need to bring biology”

The AI-quantum convergence narrative is increasingly mainstream, and 'store now, decrypt later' is well-rehearsed in security circles; the more interesting framing around large quantitative models trained on equations rather than words is fresher but underdeveloped and not pushed to a contrarian conclusion.

“Gen AI is great, but it works with words. And words can only get you so far”

“large quantitative models, which are models of AI trained on equations that we know have been working for 100 years, not just on words, but also on numbers, quantitative”

Fernando is a genuine deep-tech operator with Google X tenure and current enterprise responsibility at a well-funded, real-client company working with DoD and major banks; he's a practitioner, not a thought-leader, though he sits below founder/C-suite level and the conversation doesn't fully exploit his operational depth.

“I spent six years at Google X working in many different projects from robotics, machine learning or AI when it was called machine learning”

“that's a lot of what we're doing with the department of defense and with many of the banks and many different players across the the globe”

There are a handful of concrete anchors - Peter Shor's 1994 algorithm, the 1-million-qubit threshold, '1 in 5 breaches,' and the 2017 Attention paper - but client outcomes are unnamed, no revenue or adoption metrics are cited, and most claims about industry impact stay at the level of generality.

“in 1994, Peter Shor published an algorithm that said if one day we have a quantum computer, this algorithm will break encryption as we know today”

“one in five breaches come because of a credential that was lost or not updated at the right time”

The host asks reasonable stage-setting questions and makes some effort to synthesise and drill down (e.g. the crypto-agility follow-up), but there is no meaningful pushback on vague claims, the opener is laden with flattery, and the conversation never forces the guest into genuinely uncomfortable or specific territory.

“if the two of you are working together, then that means that you are a phenomenal being”

“Well, they love it. They love the tool because they say this is not only helping me with a future threat”