Is Your AI Finished When It Works, or Only When It Can't Be Broken? with Qing Zhang

About this episode

Most teams call an AI project done the moment it works. Qing Zhang, Managing Director at LDV Partners, sees the deals before the rest of us do — and she thinks the finish line is somewhere else entirely. Jonathan Green talks with her about why AI is quietly redefining security, identity and the perimeter all at once, what happens when one person is running thousands of agents that talk to each other without anybody watching, why the honest answer to "is our AI secure?" is uncomfortable at board level, and which human skills survive when the tools keep changing names. Key Takeaways: • AI redefines security, identity and the perimeter simultaneously. Agents are digital companions that can do anything a person can, and they talk to each other without you knowing — so one person running thousands of agents is effectively running a small army you have to track. • Chips are the headline bottleneck but only one link in a long value chain that starts at power and ends at deployment. The measure that matters is intelligence per watt. • Price per token is commoditizing, so stop pricing on tokens and start benchmarking to value. Enterprises are paying for consumption; they need it tied to th