WCBR
WisdomTree Cybersecurity UCITS ETF - USD Acc

Published 14 September 2026
Associate Director, Quantitative Research
Director, Digital Assets Research
Artificial intelligence is rapidly changing the economics of digital resilience. Frontier models are becoming better defensive tools, but they are also lowering the cost, expertise and time required to mount sophisticated attacks, while creating an entirely new attack surface that enterprises need to protect (Figure 1). The result could be a powerful new cycle of spending on both traditional cybersecurity and the security of AI systems themselves. Crypto faces the same technological escalation from a different starting point: in systems where assets are natively digital and transactions can be irreversible, the cost of security failure can be immediate.
Figure 1. AI is reshaping the cybersecurity landscape and is redefining the economics of cybercrime.

Source: CrowdStrike, “Global Threat Report” published February 2026. IBM, “X-Force Threat Intelligence Index 2026”. IBM, “Cost of a Data Breach Report 2026”. Check Point, “AI Security Report 2025”.
Investors initially interpreted rapid advances in frontier AI as a potential threat to cybersecurity companies. As Figure 2a shows, cybersecurity stocks came under significant pressure around the turn of the year and in the first quarter of 2026, alongside the broader software sector. In that period, investors reassessed the potential for AI to disrupt established software and security businesses. But that narrative rapidly reversed, with Anthropic’s Claude Mythos marking a clear inflection point for cybersecurity (Figure 2b). Mythos became the first frontier AI model to solve AISI’s cyber ranges – simulated network environments designed to test the cybersecurity capabilities of frontier AI models1.
Released initially only to trusted partners and critical software infrastructure companies, Mythos underscored how quickly frontier AI capabilities are advancing from assisting cyber professionals to performing increasingly sophisticated offensive tasks. For cybersecurity investors, this has shifted the narrative from AI as a potential disruptor of cybersecurity spending to AI as a powerful new catalyst for it.
Figure 2. Performance of leading cybersecurity strategies in Europe.
a. Period from 31st August 2025 to 10th April 2026.

b. Period from 10th April 2026 to 11 September 2026.

Source: WisdomTree, Bloomberg Finance L.P. Based on net total returns in USD. Strategies are represented by the indices tracked by the top 5 biggest cybersecurity ETFs in Europe. WisdomTree Cybersecurity is represented by the WisdomTree Team8 Cybersecurity UCITS Index. You cannot invest directly in an index. Historical performance is not an indication of future performance and any investments may go down in value.
The speed at which frontier models are acquiring cyber capabilities has been striking. In July, models being evaluated by OpenAI circumvented controls intended to isolate them from the internet, chained vulnerabilities together and ultimately accessed parts of Hugging Face's production infrastructure2. OpenAI described the episode as an unprecedented cyber incident involving state-of-the-art capabilities. In turn, Anthropic has separately disclosed incidents in which Claude models gained unauthorised access to real third-party systems during cyber evaluations.
These experiments do not mean commercial AI models are autonomously attacking companies. They do, however, demonstrate what increasingly capable models can do when those capabilities are placed in the hands of a malicious actor. Anthropic's latest threat intelligence3 shows AI already being used across the cyber kill chain, allowing adversaries to operate faster, over a broader surface and with fewer resources. In economic terms, AI can lower the marginal cost of an attack while increasing its potential scale.
An August breach at Manchester Airports Group exposed data relating to an estimated 8.7 million customers across Manchester, Stansted and East Midlands airports4. Personal data included email addresses and, for some customers, phone numbers, postcodes and vehicle registrations. Operations were unaffected, but stolen personal information can become fuel for subsequent phishing, impersonation and social-engineering attacks. AI can make it much cheaper to personalise and scale these activities.
For cybersecurity providers, this escalation is increasingly translating into demand. CrowdStrike reported its best quarter on record in August, with net new annual recurring revenue rising 51% year on year to $333 million5. CEO George Kurtz explicitly described a “Mythos moment” that highlighted that cybersecurity has become the foundation for secure AI adoption. Palo Alto Networks similarly added almost $1 billion of net new next-generation security ARR in its latest quarter, with CEO Nikesh Arora arguing that advances in AI are making CIOs to prioritise cybersecurity6.
Importantly, this spending opportunity extends beyond conventional cybersecurity. AI security is emerging as a new category in its own right. Enterprises adopting AI need to protect models, agents, training data and AI applications against risks ranging from data leakage and prompt manipulation to model attacks and unauthorised agent behaviour. This is incremental to the broader need for endpoint, identity, network, cloud and data security rather than simply replacing it.
The opportunity is becoming increasingly visible even to the architects of the AI boom. Nvidia CEO Jensen Huang recently described cybersecurity as likely to become AI’s “next major use case”7, arguing that AI-assisted coding is accelerating both the speed at which software can be exploited and the speed at which vulnerabilities must be fixed. Yet the same dynamic is also prompting calls for caution from within the AI industry. Anthropic CEO Dario Amodei has called for a slowdown in frontier AI development8, so that safety measures and our ability to understand and control increasingly capable systems can keep pace.
Crypto’s security problem is broadening, not easing. TRM Labs recorded 207 hacks and exploits in H1 2026, the highest six-month incident count on record and more than double the 83 reported a year earlier9. While total losses of US $972 million were materially below H1 202510, that is no victory lap. It mainly reflects the absence of another mega-heist, not weaker adversaries.
AI is pushing this risk sharply higher. The commercial impact is already visible in crypto. Chainalysis estimates that scams with identifiable on-chain links to AI-tool vendors extracted an average of US $3.2 million per operation in 2025 versus US $719,000 for other scams, or 4.5 times more, while generating nine times as many transfers per day11.
Figure 3: AI-enabled crypto scams extract more value

Source: Chainalysis. 13 January 2026. Historical performance is not an indication of future performance, and any investment may go down in value.
In other words, AI is expanding the pool of capable attackers and allowing sophisticated groups to run more campaigns in parallel. The result is not merely a higher volume of attempted compromises, but a faster and more adaptive threat environment in which vulnerabilities can be discovered and weaponised before smaller crypto ecosystems have the people, tooling or liquidity to respond.
Harmony’s September proposal to sunset its own Layer 1 and reissue ONE on Ethereum is therefore more than another struggling-chain migration. A blockchain launched as an Ethereum rival is effectively arguing that the fixed cost of independently securing a network against state-backed actors and increasingly capable AI-enabled threats now outweighs the value of sovereignty.
As attack capability scales, network security is becoming a scale business. Ecosystems without sufficiently deep validator, developer and operational-security resources may find that shared security, or settlement on an established chain, is the more durable option.
Taken together, developments across cybersecurity and crypto point to the same underlying reality for investors: as AI increases the speed, scale and sophistication of digital attacks, the economic value of resilience is rising. In cybersecurity, this is translating into greater demand for existing security solutions as well as a new generation of AI-specific security tools. In crypto, repeated hacks and exploits demonstrate the consequences when digital infrastructure cannot keep pace with increasingly capable adversaries. More capable attackers, a larger digital attack surface and new AI-specific security requirements could support structural growth in cyber spending for years to come. Despite this increasingly compelling growth outlook, current valuations do not appear to fully reflect the potential scale of the opportunity that frontier AI could create for cybersecurity (Figure 4).
Figure 4. Current valuations don’t seem to have priced a new era of cybersecurity risk.

Source: WisdomTree, FactSet. Period from 30 October 2020 to 11 September 2026. Based on Price-to-Sales multiple of the WisdomTree Team8 Cybersecurity UCITS Index. You cannot invest directly in an index. Historical performance is not an indication of future performance and any investments may go down in value.
The WisdomTree Cybersecurity strategy, which tracks the WisdomTree Team8 Cybersecurity UCITS Index, is designed to provide exposure to companies positioned across these evolving security themes. In March 2026, the Index expanded to include AI Security as a distinct theme, reflecting the view of WisdomTree and Team8, its cybersecurity expert partner, that protecting AI systems is becoming a structural category of cybersecurity expenditure in its own right. For investors seeking to participate in the build-out of digital resilience and the secure adoption of AI at scale, the strategy offers diversified exposure to companies positioned to benefit as cybersecurity evolves from an IT requirement into a strategic necessity and an essential foundation of the AI economy.
1AI Security Institute blog post How fast is autonomous AI cyber capability advancing? | AISI Work.
2OpenAI, The Hugging Face incident and the road ahead | OpenAI
3Anthropic, “Detecting and countering misuse of AI: September 2026”, published 10th September 2026.
4Manchester Airport plc, Manchester Airport Data Security Incident - Update & FAQs
5CrowdStrike, Q2 2027 Earnings Call.
6Palo Alto Networks, Q4 2026 Earnings Call.
7Nvidia CEO Huang Says AI Will Transform Cybersecurity Industry - Bloomberg
8Dario Amodei — We Must Pace the Frontier
9Source: H1 2026 Crypto Hacks Reach Record High as Losses Fall Below USD 1 Billion | TRM Labs
10Source: H1 2026 Crypto Hacks Reach Record High as Losses Fall Below USD 1 Billion | TRM Labs
11Source: 2026 Crypto Crime Report: Scams

Associate Director, Quantitative Research
Elvira has been a member of the WisdomTree Research team since September 2018. With over nine years of experience in the ETF and asset management industry and an academic background in quantitative finance, Elvira combines technical expertise with thematic strategy insights to contribute to thought leadership research and the development of new innovative strategies at WisdomTree. As a lead specialist in thematic strategies, she supports the periodic review and rebalancing of thematic portfolios, delivers quantitative insights, bespoke analysis for clients, strategic thought pieces as well as commentary on market trends and thematic strategies. She also develops comprehensive product collateral designed to support client needs.

Director, Digital Assets Research
Dovile Silenskyte is a director of digital assets research at WisdomTree. Before joining WisdomTree in May 2024, Dovile worked as an index equity product strategist at BlackRock. Currently, she is responsible for conducting analyses for in-house digital assets publications and assisting the sales team with client queries about products and markets. Dovile holds an MSc in Finance from Texas A&M University – Commerce, and she is also a chartered financial analyst (CFA).