Arvind Krishna & Nikesh Arora: Quantum Threat, Open vs Frontier AI, AI Cyber Warfare & SaaS Collapse

The episode opens with a topic the hosts had been circling for days: cryptography. Arvind Krishna's entry point is a DARPA competition called the Quantum Benchmarking Initiative, which began with a couple of hundred entrants and is down to roughly thirteen — his shorthand for how much of the field is real. His working assumption is not that quantum computers work today but that a machine capable of breaking current cryptography arrives somewhere around 2030, with credible estimates spread between 2028 and 2033.

He is careful about what that claim rests on. Post-quantum cryptography is not speculative: the mathematics is understood and the code exists, and his complaint is that organisations have been too lazy to adopt it. Nor is the guarantee absolute — he compares its rigor to the claim that today's cryptography resists today's classical computers, which rests on complexity-theoretic assumptions rather than a proof.

The practical sting is retroactive. An encrypted disk stolen today can be held for five years and opened once the machine exists, which converts a future capability into a present-day data-loss problem.

Asked to define the milestones he would accept as evidence, Krishna gives two. The first is a demonstration of error correction or mitigation on a running machine — not a simulation, not a side argument — and he expects claims next year that third parties should audit, with one or two real advances by 2027.