The buzzword "P(doom)" — a mathematical function reflecting the probability of AI-induced mass destruction — is rising, and so are shares of cybersecurity companies. Palo Alto Networks, Crowd Strike, and Fortinet have more than doubled over the past six months. AI enables bad actors to create deepfakes, craft personalised phishing emails, and trick AI agents into leaking data, driving organisations to spend heavily on protection. Cyber security spend is projected to grow by 10 per cent annually between 2025 and 2030, per a TD Cowen report citing Gartmore data.

Analysts estimate annual revenue growth of about a fifth for the next two years for Palo Alto and Crowd Strike, according to S&P Capital IQ. However, questions remain about whether traditional vendors can sustain this momentum as LLMs increasingly offer software scanning services. Since Anthropic released its Mythos model, LLMs have been able to spot thousands of previously unnoticed vulnerabilities.

Yet traditional vendors retain advantages through deep customer embeddedness and network effects, with over 80 per cent of attacks centred on human issues like stolen identities, according to Bernstein.