Bank of America pointed out that the market's concerns about AI disrupting the cybersecurity industry seem to be somewhat overstated.
Last Friday, Anthropic launched a new security feature, Claude Code Security, for its Claude AI model. Investors are worried that AI may automate and replace a significant portion of the cybersecurity industry, leading to widespread sell-offs in the sector. On Monday, cybersecurity concept stocks continued to decline, with CrowdStrike and Zscaler down about 10%, Netskope plummeting over 12%, and the Global X Cybersecurity ETF dropping to its lowest level since November 2023.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/f0929ece-4f5c-4992-a6cb-15881135fc8b.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/f0929ece-4f5c-4992-a6cb-15881135fc8b.png"/>
According to the Wind Trading Desk, a recent report from Bank of America indicated that the market's panic over &#34;AI replacing cybersecurity&#34; is overestimated. The report stated that AI tools currently cannot replace complete cybersecurity platforms; their role is as enhancers rather than disruptors. The research pointed out that AI can indeed improve work efficiency in specific scenarios like code scanning, but in terms of visibility, control, and reliability—three key dimensions—current AI capabilities cannot replace a complete cybersecurity platform.
The report believes that comprehensive platform vendors represented by CrowdStrike and Palo Alto Networks, with their integrated data foundation and detection response capabilities, still have a solid moat. In contrast, independent vendors focusing on application security (AppSec) such as Checkmarx, Snyk, and toolchain companies like GitLab and JFrog may face greater competitive pressure and need to continuously prove their technological differentiation and value reconstruction capabilities in the AI era.
<h2>The Evolution of Code Scanning, Not a Security Revolution</h2>
The report pointed out that AI tools represented by Claude have achieved key breakthroughs in the field of development security. Unlike traditional scanners that rely on feature matching, Claude can perform in-depth reasoning analysis on component interactions, data flows, and business logic vulnerabilities, breaking through the boundaries of pattern recognition and significantly improving the accuracy of complex vulnerability detection.
However, the research report also clearly delineated the boundaries of Claude's capabilities: its impact is limited to the niche area of pre-production code analysis. This means that vendors using application security as a growth engine will face pressure for value reassessment and need to prove their irreplaceability in the AI era.
<h2>AI's Obvious Shortcomings: A Fundamental Gap Between Developer Tools and Runtime Security</h2>
The report pointed out that the reason AI finds it difficult to shake the core position of cybersecurity platforms lies in the fundamental differences between developer tools and runtime security environments.
The report summarized the characteristics of runtime security as: continuous monitoring, context awareness through multi-source signal fusion (integrating various signals from endpoints, identities, networks, and cloud environments), and an almost zero tolerance for errors, especially with a false negative rate that must be controlled within a precise range of 99.99%, as any omission could lead to serious consequences
In contrast, the vulnerabilities of reasoning models represented by Claude are evident: they are highly sensitive to prompt wording and easily lose context when exceeding established task boundaries. This instability may be tolerable in code reasoning scenarios, but it is fundamentally incompatible with the requirements for autonomous defense.
More importantly, current AI tools lack the real-time data collection capability of runtime sensors and are deficient in visibility into the execution process, let alone control plane permissions, which means they cannot perform critical response actions such as blocking processes or isolating terminals. Faced with the reliability thresholds required by security levels, these capability gaps constitute a moat that current technological paths find difficult to cross.
<h2>AI Reshapes the Cybersecurity Landscape, Platform Vendors Enjoy Moats</h2>
Bank of America Securities believes that AI's reshaping of the cybersecurity industry will follow the historical path of cloud computing. Just as the rise of hyperscale cloud service providers redistributed the value of the software stack and gave birth to new leaders like CrowdStrike and Palo Alto Networks, the AI wave will similarly spawn alternatives at the edge, but the ultimate beneficiaries will be those comprehensive platform vendors that can combine massive telemetry data, runtime sensor capabilities, and reproducible workflow systems.
In the application security field, independent vendors are facing direct impacts. Due to the high reliance of companies like Checkmarx and Snyk on code scanning functionalities, their core value is at risk of being partially replaced by Claude-like tools. Even though large vendors have deepened their layouts through acquisitions, such as CrowdStrike acquiring Bionic and Palo Alto Networks successively acquiring Bridgecrew, Cider, and Dig, research reports indicate that some of these model capabilities may also be diluted by AI. A similar logic applies to the cloud security posture management field, where some misconfiguration detection functions are expected to be enhanced by AI.
However, the report emphasizes that these capabilities exist merely as functional features within larger platforms, rather than being the core that defines platform value. The reason comprehensive vendors are difficult to disrupt is rooted in three structural moats: first, the accumulation of cross-dimensional telemetry data from endpoints, identities, to cloud workloads; second, the continuous monitoring capability of real-time execution behaviors; and third, the reproducible workflow systems that can integrate multidimensional signals in complex environments. These capabilities together form a competitive barrier that AI tools find hard to surpass

In response to the sector sell-off triggered by Anthropic's launch of a code security tool, Bank of America reported that the market's panic over "AI replacing cybersecurity" is overstated. The report points out that AI cannot replace a complete platform in terms of visibility, control, and reliability; its role is as an enhancer rather than a disruptor. The moat of comprehensive platforms remains solid, while independent vendors focusing on application security face pressure for value reassessment

- Bank of America indicates that market fears regarding AI disrupting the cybersecurity industry are overstated, noting that AI currently enhances rather than replaces cybersecurity platforms.  
- Specific AI tools like Anthropic's Claude improve code analysis accuracy but are limited to pre-production code, posing competitive challenges for application security companies.  
- The report asserts that comprehensive cybersecurity firms will maintain their advantages due to their integrated data capabilities and continuous monitoring, despite potential disruptions from AI advancements.

Will AI "kill" cybersecurity platforms? Bank of America refutes: it is an enhancer, not a replacer