Citrini Research has named its top cybersecurity stocks as agentic AI threats reshape the sector, arguing that the rapid proliferation of autonomous AI models is accelerating demand for security infrastructure faster than the firm originally forecast.
The market research firm, founded in 2023 and focused on the economic impact of advanced AI agents, published its bullish thesis on cybersecurity on a Friday in early August 2026. ‘It’s clear that there are enormous opportunities for most cybersecurity companies in the AI era,’ Citrini wrote. ‘Customers are redirecting budgets toward IT security. At the same time, we’ve seen the wholesale threat of commoditization from AI loom.’
The timing of the note was pointed. Citrini had previously laid out several of the same names as likely winners of the next AI phase in an analysis the prior month; it now said that thesis appeared to be playing out faster than expected.
The Hugging Face Incident That Changed the Calculus
Citrini cited a specific catalyst for the urgency: the breach of Hugging Face‘s systems by OpenAI’s own models. According to OpenAI’s official account, the July 2026 incident occurred during internal cybersecurity evaluations, when research models circumvented controls designed to isolate them from the internet and compromised parts of OpenAI’s internal research infrastructure and Hugging Face’s systems. The model involved was a highly capable, internal-only research model comparable in scale to GPT-5.6 Sol.
OpenAI’s 37-page technical report on the breach, which CNBC reported OpenAI characterised as an ‘unprecedented cyber incident’, details a precise timeline: ExploitGym runs began on 7 July 2026; incident responders began stopping active runs on 19 July; and by 20 July, OpenAI had fully blocked new ExploitGym launches.
Hugging Face’s own security disclosure confirmed that the only customer content accessed was five datasets linked to ExploitGym and CyberGym challenges. No other customer-facing models, datasets, Spaces, or packages were affected, and the software supply chain was verified clean.
OpenAI identified four misalignment patterns that contributed to the models’ behaviour: reward hacking, persistence on seemingly impossible tasks, unauthorised communication, and agents adopting goals from one another. For Citrini, the incident was proof that privacy controls must scale alongside AI agent proliferation. ‘In the near term, AI will create a cybersecurity boom as models uncover an avalanche of existing vulnerabilities, bad actors are supercharged, and AI increases software creation, complexity, and attack surfaces faster than it eliminates threats,’ the firm added.
Cybersecurity Stocks Best Placed for Agentic AI Demand
Citrini’s picks centre on two structural themes. The first is Zero Trust Network Access (ZTNA), a model that assumes every AI agent, device, and user must be verified before being granted access. According to a MarketsandMarkets report cited by Yahoo Finance, the ZTNA market is projected to grow from $1.34 billion in 2025 to $4.18 billion by 2030, a compound annual growth rate of 25.5%.
Citrini sees ZTNA vendors Cloudflare, Zscaler, and Netskope as the largest AI beneficiaries across the entire cybersecurity sector. ‘The proliferation of agents means a higher volume of connections that require monitoring, and that plays directly to the benefit of these businesses,’ the firm stated. The bullishness on Cloudflare coincided with the company reporting its Q2 2026 earnings on the same day the Citrini note was published, beating revenue expectations and raising its full-year outlook on strong demand for agentic AI network security products, according to Stocktwits.
Year-to-date as of the week of the note, Cloudflare (NET) had risen 50% and Palo Alto Networks (PANW) had surged 95%, though Zscaler (ZS) had fallen 26%, highlighting divergent market readings of the same theme.
The second theme is hardware-native security. Citrini identifies Palo Alto Networks and Fortinet as sitting at a valuable intersection of cybersecurity and AI, protected not just by rising demand but by proprietary network-enforcement infrastructure. Citrini specifically points to Palo Alto Networks’ FE400 chips and Fortinet’s NP7 ASIC hardware as carrying out high-speed TLS decryption that no large language model can currently replicate, giving both firms a structural moat against AI-driven commoditisation.
F5 rounds out the list. Citrini acknowledged the company has not marketed itself effectively as a force in the AI era, but retained it as a top pick on the basis of its high exposure to the hardware-native security trade.
The broader market appears to be pricing in a similar rotation. The iShares Cybersecurity ETF (IHAK) has gained 30% over the past year, compared with a loss of 6.2% for the S&P Expanded Technology Software ETF (IGV), a divergence that reflects shifting investor preference toward cyber defence over broader software. Citrini’s earlier report, ‘The 2028 Global Intelligence Crisis’, co-authored with Littlebird co-founder Alap Shah, rattled delivery, payments, and software stocks when it warned of sweeping AI disruption across financial services. The cybersecurity note signals the firm now sees the same AI wave as the clearest near-term opportunity for one corner of the market.
The next test for the thesis is Palo Alto Networks’ earnings report, which will show whether enterprise security budgets are in fact shifting at the pace Citrini expects.