Large enterprises are shifting their cybersecurity budgets toward AI-driven tools to counter new threats. Spending is increasing for systems that monitor employee AI usage, AI-powered vulnerability discovery, and automated patching. To offset these costs, companies are cutting expenditures on traditional vulnerability management software, log collection tools, and manual penetration testing services.
Lumen Technologies expects to increase its cybersecurity budget by approximately 30% over the next year. The company is investing in AI scanning tools from providers like Anthropic (Claude Mythos), Palo Alto Networks, and startup Zafran Security. Lumen has also formed a dedicated team to use frontier LLMs for vulnerability scanning, while seeing an increase in external reports from AI-assisted researchers.
AppLovin is managing a near $1 million annual security software budget, which grew by about 10% this year. The company has contracted with early-stage startups Pluto Security and Fig Security to monitor employee AI behavior and identify vulnerabilities. AppLovin also uses Endor Labs for code scanning and is re-evaluating its spend on CrowdStrike and SentinelOne due to the superior performance of newer AI tools.
The market is seeing a rapid shift in procurement habits. Zafran reported securing three large bank contracts within five weeks of the release of Anthropic's Mythos model. Meanwhile, legacy vendors like Rapid7 have faced pressure, with the recipient reporting a slight revenue decline and implementing 12% layoffs to pivot toward AI products. In contrast, SentinelOne reported that its AI security segment's annual recurring revenue grew threefold in the second quarter.
Companies are also increasing their 'token budgets' to utilize models from Anthropic and OpenAI for system scanning. Veeam is offsetting the cost of the Mythos model via the Glasswing project by reducing third-party penetration testing. AppLovin uses Claude Opus 4.7 for weekly scans, spending several hundred USD per week on tokens. Palo Alto Networks reported spending over $1 million on Mythos tokens during a May testing phase, though it expects usage to stabilize.