Future wars will rely more on cybersecurity than military strength: Zscaler CEO Jay Chaudhry

Cybersecurity is increasingly becoming critical as technology advances, with future conflicts expected to depend more on cyber strength than traditional military force, said Jay Chaudhry, founder, chairman and chief executive of cloud security company Zscaler.

“While wars are becoming more and more sophisticated, I do think that cyber capabilities, especially those powered by AI (artificial intelligence), will play a much bigger role… I believe future wars will be won and lost many times without firing a single shot, because if your banking system is crippled or your power system is crippled, it creates a lot of chaos,” he said at the AI Impact Summit in Delhi on Wednesday.

According to Chaudhry, while AI offers endless possibilities, it also comes with substantial risks, making it essential to counter AI-driven threats using AI itself.


“When a cyberattack occurs, it begins with identifying the attack surface. What is your attack surface in the cyber world? It includes every public IP (internet protocol), every firewall, every VPN (virtual private network) and every application portal,” he said.

Chaudhry added that as applications are increasingly being built with the help of AI agents that generate code, new security flaws are also likely to emerge. In the short term, this agentic coding is expected to lead to a sharp increase in the number of security vulnerabilities.

“The most effective thing you can do in the cyber domain is to embrace zero trust, which means not trusting anyone,” he said.

India is now focusing on developing sovereign products and AI models tailored to local needs, with a growing emphasis on domain-specific systems, said Chaudhry.

“When we look at cybersecurity, we want our models to be highly focused and well trained on cyber telemetry and global cyber laws, because that is the domain we really need to address. By focusing on domain-specific capabilities, we reduce costs and significantly cut down the effort required to train these models and make them successful,” he said.