Nikesh Arora explains that Palo Alto Networks monitors endpoint traffic for $20 to $25 annually using compressed local models. Feeding endpoint data directly into commercial LLMs like OpenAI would bankrupt customers on token fees. Companies should reserve expensive frontier models for rare high-stakes crises like active breaches.
Arvind Krishna notes that self-hosting optimized open-weight models cuts token costs from $20 or $50 down to $1 to $2 per million. Closed frontier models also pose intellectual property leakage risks via logged chain-of-thought traces. He projects open-weight alternatives will capture 75% to 90% of token usage.
Nikesh Arora observes that enterprise clients cannot reasonably stitch together 40 separate security products when they barely comprehend one tool. Just as 23 disparate sales systems merged into platforms like Salesforce, cybersecurity point tools will inevitably consolidate into unified platforms to eliminate manual glue code.
Arvind Krishna estimates that practical quantum computing will arrive around 2030, making current public-key cryptography obsolete. Because stolen encrypted data can be saved for years and cracked later, enterprises cannot wait. Organizations must adopt post-quantum algorithms now rather than delay upgrading until machines arrive.
Arvind Krishna highlights that typical enterprises maintain between 500 and 5,000 distinct software vendors, creating severe security vulnerabilities. Lightweight AI tools can easily replace around 490 of those vendors, driving budgets back to a handful of core platforms that protect distribution and reduce the corporate attack surface.
Nikesh Arora warns that companies are already late to quantum preparation, with roughly half of current IT budgets funding legacy security stacks. Organizations cannot rapidly swap trillions of dollars of infrastructure in a single year. Upgrading network traffic with post-quantum keys must happen immediately.
Nikesh Arora highlights how Nvidia mirrors the classic vertically integrated IBM mainframe model by owning hardware down to dirt and controlling developer applications. Operating the entire tech stack allows a company to manipulate pricing pools across layers, destroying a rival's margin while shifting profit into proprietary tiers.
