Open-SourceAI Is Closing the Gap

Open-source AI is catching up to frontier models faster with each generation. SemiAnalysis found that the time required for an open model to match the first closed model of each LLM era has fallen from more than a year in the early scaling era to roughly 5 months today in the agentic era. DeepSeek’s V4 Flash Vision Exp recently reinforced the trend, outperforming Anthropic’s Opus 4.8 on visual and agentic benchmarks at a fraction of the cost.

Exhibit: Open Models Are Catching the Frontier Twice as Fast Each Era

(Source: Semi Analysis, 21 Aug 2026)

The result: model intelligence is becoming more accessible, while users increasingly choose models based on the task and cost.

AI Usage Is Moving Down the Model Stack

Frontier models remain important for complex tasks, but smaller and open-weight models are handling an increasing share of routine workloads. Token volumes have grown roughly 34× for frontier models, 84× for small models and 386× for open-weight models. Pricing tells a similar story. Open-weight median prices have fallen approximately 66%,from $0.62 to $0.21, while frontier prices declined about 26%. The gap between the two has widened from roughly 22× to 47×.

Exhibit: Token Budgeting in Full Display

(Source: Source: OpenRouter Data API, rankings-daily · 1,193 trillion tokens over 595 days to08/26 · 5-week centred mean over every week · markers = the last 3 weeks, raw ·incomplete weeks not plotted · top 50 models a day · prices as pages 4 · 11–13.)

The takeaway: users are increasingly matching model capability to the task, rather than relying on a single model for everything.

Compute Remains the Bottleneck

AI models may be getting cheaper, but the infrastructure required to run them remains scarce. All five Graphics Processing Units tracked remain above their 2026 lows, with the A100-80GB up approximately 35% from its low.

Exhibit: Compute Shortage Continues – GPU Pricing

(Source: Source: SkyPilot catalog git history · $ per single GPU-hour · 5-week centred mean over every week · markers= the last 3 weeks, raw · indexed to 100 at 08/25, B200 at its 10/25 floor ·average excludes B200.)

That scarcity could help preserve an advantage for frontier labs. SemiAnalysis estimates that frontier AI tokens can generate approximately $100 million per megawatt annually, giving leading labs significant incentive to compete for scarce compute.

The financing required to build that infrastructure is also drawing attention.Nvidia (NVDA) agreed to provide more than $100 billion in credit support for a new OpenAI data center in Ohio, alongside a $1.5 billion equity investment and exclusive compute-supply agreement. Nvidia closed at $214.72on Aug. 21, down 0.98%.

Meanwhile,AMD (AMD) continues to expand its role in AI infrastructure, with reported involvement in Google’s next-generation Tensor Processing Unit development and multi-year Helios rack agreements with Meta, OpenAI and Anthropic. AMD closed at $473.25 onAug. 21, up 0.81%.

Agentic AI Is Expanding the Cybersecurity Opportunity

As AI agents begin acting autonomously, traditional identity systems may not be enough to control what non-human users can access and do. Morgan Stanley estimates the emerging agentic identity opportunity could add $33 billion to today’s $24 billion identity-security market, creating a ~$60 billion market by 2030 in its base case.

Exhibit: Agents Could Roughly Double the Identity-Security Market by 2030

(Source: Morgan Stanley, 20 Aug 2026)

With non-human identities potentially reaching 100× the number of human identities, the need for real-time controls around autonomous agents could create a structural opportunity for cybersecurity providers such as Crowd Strike(CRWD) and Palo Alto Networks (PANW)

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

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.

‍

Before investing, carefully consider theTrueShares ETFs investment objectives, risks, charges and expenses. Specific information about TrueShares is contained in the prospectus and a summary prospectus, copies of which may be obtained by visiting www.www.true-shares.com.Read the prospectus carefully before you invest.

The fund is distributed by Paralel DistributorsLLC, Member FINRA. Paralel is not affiliated with TrueMark Investments, LLC and BlackHill Capital Management.

TRUE774

‍

‍