A sweeping majority of corporate security leaders expect to raise cybersecurity spending next year, driven by artificial intelligence's growing ability to discover and exploit software vulnerabilities faster than human defenders can patch them. According to a new survey of 100 chief information security officers at companies with more than 1,000 employees, published today by venture capital firm Andreessen Horowitz, 87% anticipate increased budgets for 2027. The report documents a fundamental shift in how enterprises protect themselves, as AI simultaneously creates new threats and offers tools to counter them.

The data reveals how quickly priorities have changed. When security chiefs recalled their 2026 priorities, exposure and vulnerability management and SIEM topped the list. For 2027, all seven of the most-selected initiatives involve AI. SOC automation received the most first-place votes, while other teams prioritized offensive security, remediation, or securing their own AI deployments first. Internal AI agent security, SOC automation, and offensive security made the largest gains from the prior year, with code security, remediation, and AI governance following closely. Despite AI handling more of the work, nearly half of respondents expect security headcount to increase in 2027, another half expect it to stay flat, and only 4% plan cuts. The ambition is to cover more ground: investigate alerts that might otherwise wait, test more applications, and close vulnerabilities sooner.

Purchase plans reflect the urgency. Within the next twelve months, 45% plan a new purchase or upgrade for internal AI agent security, followed by offensive security at 42% and AI usage and governance at 38%. Yet production adoption remains low: only 18% report production use of AI SOC tools, and 13% report production use of either AI offensive security or AI exposure and vulnerability remediation. Many are already piloting these capabilities or expect to reach production within a year. The report notes that 92% of companies have at least limited production use of employee generative AI tools, 83% use coding assistants or agents, and 55% have MCP servers or other agent-to-application integrations in production. Security concerns are already slowing deployment in many cases: 63% of responses describe security slowing at least some AI adoption through reviews, approval requirements, and restrictions on particular uses.

The report traces the pressure back to concrete incidents that demonstrate AI's offensive capabilities. This summer, OpenAI's agents broke out of a cybersecurity evaluation and breached Hugging Face's production systems, exploiting vulnerabilities and using stolen credentials to work their way into another company's infrastructure. More recently, researchers working through OpenAI's Patch the Planet initiative helped uncover critical flaws in PHP, which powers roughly 70% of websites, that had been sitting undiscovered for more than a decade. The authors explain that models are getting better at finding weaknesses and figuring out how to exploit them, with progress visible in the AI Security Institute's evaluations where newer models complete increasingly complex cyber tasks. Bain estimates that AI can compress a typical attack sequence from roughly four weeks to about eighteen hours, leaving much less time to find an exposure and fix it. Meanwhile, the environment security teams must protect keeps expanding as employees ship applications they couldn't have built a year ago and agents gain access to company documents, source code, and production systems.

Looking ahead, security leaders expect testing to become far more frequent and embedded in development processes. Today, 18% say most of their offensive testing runs continuously against production assets; 52% expect it to within a year. Looking five years out, 44% expect testing to be embedded in CI/CD pipelines, triggered by code or infrastructure changes. The report argues that this brings security testing into the process of building and maintaining software, so a release can trigger a new test and a fix can be checked as soon as it's deployed. When asked where they would look first for AI security capabilities, 44% default to features from an existing security platform, compared to 21% who start with an AI-native security vendor. But the first call isn't always the last one: 47% said they would choose a startup over their incumbent if it solved a capability gap their existing vendor cannot address. The authors conclude that this next wave will produce important security companies, built around capabilities that enterprises are only beginning to need and will soon find difficult to operate without. The industry may struggle to distinguish signal from noise, however—in some categories the report identifies as many as 43 competing vendors, forcing security chiefs to narrow evaluation lists from seven or eight down to three for hands-on testing. Platforms retain an advantage through existing relationships and procurement approvals, but they leave room for specialists who can solve problems the incumbents haven't yet addressed.