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

Brief Summary

Arvind Krishna and Nikesh Arora discuss quantum-safe security, open-weight AI, cyber defense, AI regulation, and the coming consolidation of enterprise software.

Key points

• Quantum progress, measured in milestones. The panelist lays out the steps to watch: error correction on a running quantum computer, then scaling up its computations. A tenfold increase each year could bring startling capabilities by 2029.
• The quantum deadline is closer than it looks. A quantum computer could break today’s cryptography around 2030, but replacing IT infrastructure takes time. The panelist says organizations should start moving to quantum-safe systems now.
• Use the right model for the job. For routine enterprise tasks, smaller, specialized models can be cheaper and accurate enough. Frontier models make more sense when the stakes—like stopping a breach—justify the cost.
• Why open-weight models could dominate. Lower costs, control over intellectual property, and geopolitical considerations all point toward open-weight models taking most enterprise token usage. The panelist predicts they could account for 75–90%.
• Attackers need one vulnerability. AI can help identify security flaws, but its findings still need verification. Attackers only need one vulnerability to succeed, while defenders must find and patch every one.
• Share vulnerabilities, strengthen defenses. Rather than having every company investigate and patch vulnerabilities alone, the panelist proposes pooling findings through trusted parties and distributing fixes.