‍AI is rapidly reshaping the cybersecurity landscape.

For the first time, non-human “agent” traffic accounted for more than 50% of Cloudflare’s (NET) network, while Datadog (DDOG) reported a 22x year-over-year increase in MCP (agent tool-call) volume, signaling a structural expansion for companies operating across the AI traffic stack, according to RBC’s review of roughly 24 software earnings calls. At the same time, frontier AI models are increasingly capable of identifying and exploiting vulnerabilities in minutes, while OpenAI launched its “Daybreak” cybersecurity models on Amazon Bedrock. RBC and Citizens favor CrowdStrike (CRWD) and Palo Alto Networks (PANW) as potential beneficiaries of the growing need to supervise and secure AI systems. Qualys also reported that autonomous remediation can reduce the vulnerability exposure window from 21 days to mere minutes.

(Source: RBC Capital Markets, 9 & 12 Aug 2026; Citizens, 12 Aug 2026)

AMD’s AI-infrastructure story is gaining momentum.

AMD (AMD) shares rose 6.5% to $514.39 on Aug. 14 as new developments strengthened the company’s position in AI infrastructure. Wedbush highlighted Anthropic’s reported 20-year, 191-megawatt data-center lease with RIOT Platforms as a positive read-through for AMD. RIOT has already delivered 25MW to AMD, with another 25MW under construction, while Anthropic is expected to deploy up to 2 gigawatts of AMD MI450 GPUs on AMD’s Helios rack platform beginning in 1H27, with Microsoft running Helios on Azure. AMD also filed a shelf registration for up to $5 billion in debt. The broader takeaway: AMD is securing both the supply and financing needed to compete more aggressively with Nvidia in the rapidly expanding AI-inference market.

(Source: Seeking Alpha, 11-14 Aug 2026)

Token Prices Reveal the Impact of New Model Launches.
While smaller models may be changing how users allocate their token budgets.

1. Open-weight models are the only tier to see a meaningful price decline
Matched-model prices fell 55%, while volume surged 365×. As cheaper models enter the market, usage is increasingly shifting toward them.

2. Frontier pricing held steady—until the latest Mythos launches
The introduction of Claude Mythos and Claude Mythos 5 pushed the median frontier-model price from $6 to $10 on Aug. 10. The frontier-to-open-weight price ratio has widened from 22× at the start of the series to 47× today, even as frontier-model volume grew 32×.

3. Small-model prices rose on paper, but fell in practice
Matched-model prices declined 1%, despite the median price rising roughly 160%, from $0.26 to $0.69. New Flash, Luna, and Haiku models are typically priced above the models they replace, while lower-cost 2025 tiers have rolled off. Even so, volume increased 47×.

(Source: BerriAI/litellmprice file, git history · blended $ per million tokens at 3 input : 1 output ·80 weeks from 02/25 to 08/10/26 · upper chart 5-week centredmean, markers = last 3 weeks raw · lower chart is a Jevons chain over modelspresent in both weeks, unsmoothed · list prices only.)

Token Budgeting in Full Display: Frontier, small, and open-weight models are increasingly serving different use cases.

1. Frontier models for complex tasks; smaller and open-weight models for everything else
Users are increasingly matching models to the task at hand—using frontier models when advanced capabilities are needed and smaller or open-weight models for more routine workloads.

2. Token volume is exploding across every category
Usage is surging across the board, with small models growing roughly 47× and open-weight models 365×, compared with 32× for frontier models. Users are moving down the model stack faster than vendors are cutting prices.

(Source: OpenRouterData API, rankings-daily · 1,018 trillion tokens over 579 days to 08/26 ·5-week centred mean over every week · markers = the last 3 weeks, raw · incompleteweeks not plotted · top 50 models a day · prices as pages 4 · 11–13.)

Estimates are illustrative and directional, based on OpenRouter routing data and assumptions used to extrapolate to the broader market. OpenRouter is not representative of all AI inference traffic, and certain traffic, including first-party enterprise usage and locally run open-weight models, is not observed. Figures should not be interpreted as audited market-size or market-share estimates. A price gap is a sudden jump or drop in an asset's price. A volume ratio is the proportional relationship between the volumes of two or more measurements.

Market Share Is Shifting - Measured by Token Volume: Chinese open-source labs now account for well over 50% of all routed tokens.

The mix is changing, but every model tier continues to grow. Open-weight models have surged from roughly one-fifth of routed tokens to approximately 70%. Frontier and small models have both lost share, but their absolute token volumes continue to increase as overall AI usage expands.

(Source: OpenRouterData API, rankings-daily · 1,018 trillion tokens over 579 days to 08/26 · 378models · 5-week centred mean over every week · markers = the last 3 weeks, raw · incompleteweeks not plotted · grouped by weight availability.)

----------------------------------------------------------------------------------------

As AI continues to reshape computing, software, and the broader economy, identifying the companies positioned to benefit from these long-term trends is becoming increasingly important. LRNZ seeks to provide concentrated exposure to companies at the forefront of AI and deep learning, focusing on businesses where advanced AI applications may create a distinct competitive advantage.

Learn more about the TrueShares Technology, AI & Deep Learning ETF (LRNZ)

Please click on the link to view the funds top ten holdings. Fund holdings and allocations are subject to change and should not be considered recommendations to buy or sell any security.

TRUE765