It has been well over a year since AI ETF AGIX, the KraneShares Public Private AI & Technology ETF, added private-company exposure to its portfolio.

AGIX first added Anthropic in February of 2025, and later that year, added xAI. Since then, SpaceX merged with xAI, and the combined entity went public on June 12th, 2026.¹

Meanwhile, Anthropic grew its annualized run-rate revenue (ARR), which is an annualized estimate of revenue based on current revenue levels, from approximately $1 billion in early 2025 to $47 billion in May of 2026.² Additionally, on June 1st, 2026, Anthropic announced that it "confidentially submitted a draft registration statement on Form S-1 to the U.S. Securities and Exchange Commission for a proposed initial public offering of common stock."³

Within AGIX, we generally prefer capitalization table exposure to private companies through funding rounds or through buying shares from existing holders. Now, AGIX has a new class of direct investments in several exciting private AI companies.

Introducing AI ETF AGIX's New Private AI Company Class

AGIX's private exposure has centered on Anthropic (2.82% of AGIX as of 7/31/2026⁴) and xAI (now part of SpaceX) as core ways to access frontier AI development and the infrastructure that supports it.

AGIX's newest private holdings (Polymarket, Apptronik, Ayar Labs, and Nuro) share a common theme: they operate in specialized corners of the emerging AI economy that we believe are underrepresented in public markets.

Polymarket (0.44% of AGIX as of 7/31/2026⁴): Market-Based Intelligence for AI

Polymarket is a New York-based online prediction market where people trade on the outcomes of real-world events ranging from elections to sports to technology themes, with prices reflecting what the public believes will be the most likely outcome.

Those prices turn complex human opinions into clean probability numbers that update as new information comes in, creating a live data feed that AI models can use to reason about future scenarios and uncertainty, rather than relying on static surveys.

Polymarket capitalizes on its unique data by offering a dedicated application programming interface (API) suite, including its Gamma API for events and markets, CLOB API for real‑time prices and order‑book depth, and a Data API for historical and analytics views. These APIs are already being used by customers like Perplexity, which integrates Polymarket odds into its AI‑powered search experience via a commercial API agreement.⁵

Polymarket has attracted backing from investors including the Intercontinental Exchange (ICE), Founders Fund, and 1789 Capital⁶, and recently struck a major data-distribution and tokenization partnership with ICE, the parent company of the New York Stock Exchange.⁷

Apptronik (1.67% of AGIX as of 7/31/2026⁴): Physical AI Through Humanoid Robots

Apptronik, based in Austin, Texas, is building Apollo, a humanoid robot designed to work in warehouses, factories, and logistics centers performing heavy, repetitive tasks. Apollo relies on AI to understand and interpret its surroundings and coordinate its movements, which may enable it to work alongside people and other machines, turning AI from something that lives on a screen into a physical co‑worker.

Before the company was created, Apptronik's founders were selected to help build NASA's Valkyrie humanoid robot for the DARPA Robotics Challenge. Since Apptronik's creation in 2016, it has worked with NASA through multiple Small Business Innovation Research contracts to develop Apollo for both Earth and potential space applications.⁸ We believe that decade‑long collaboration has shaped Apollo's design and has helped differentiate Apptronik from peers that are newer to real‑world deployment.

Apptronik has drawn strategic investors, including Google, Mercedes-Benz, John Deere, AT&T Ventures, and Qatar Investment Authority, and has operational partnerships with Mercedes-Benz, GXO Logistics, and Jabil.^6,9

Ayar Labs (1.29% of AGIX as of 7/31/2026⁴): Photonics for AI Scale-Up

Ayar Labs is a California-based photonics company that replaces traditional copper wires inside servers with tiny optical links that move data using light instead of electricity. For AI, moving large amounts of data quickly between thousands of graphics processing units (GPUs) is just as important as the chips themselves. Traditional copper wiring is becoming a bottleneck, and Ayar's photonic connections are designed to unlock much higher bandwidth at lower power costs so data centers can run more powerful and faster AI models.

Ayar Labs has raised capital from Neuberger Berman and the Qatar Investment Authority and has strategic backers like AMD Ventures, NVIDIA, MediaTek, Intel, and Globalfoundries.⁶

Recently, we wrote a more in-depth piece covering Ayar Labs, how photonics work, and why they are an important part of the AI ecosystem.

Nuro (0.87% of AGIX as of 7/31/2026⁴): Turning Autonomy Into an AI Driver

Nuro is a California-based company that builds software to let cars drive themselves. Nuro's commercial strategy is centered on licensing its self-driving technology, called Nuro Driver™, to carmakers and mobility platforms; essentially an AI "brain" that is designed to see the road, predict what might happen next, and decide how the vehicle should respond in real time. By operating on a software licensing model, Nuro avoids the capital-intensive burden of manufacturing its own vehicles. We view this as a central competitive advantage for Nuro, which we covered in more detail when we announced AGIX's direct investment in the company.

Nuro's technology has attracted partners including Walmart, Domino's, and FedEx¹⁰, and investors such as Uber, NVIDIA, and SoftBank.⁶

Together, Nuro, Polymarket, Ayar Labs, and Apptronik represent four specialized but important pieces of the emerging AI ecosystem.

AI ETF AGIX Performance Update: Dynamically Adapting As The AI Story Changes

Now, let's dive into the numbers.

As shown in the chart below, AGIX (since it's inception on 7/17/2024) has outperformed both its public market benchmark and the Nasdaq 100 Index, which is a widely recognized and accepted technology benchmark.⁴

Over the past year, the fund is up +32.72%, with a +17.01% year‑to‑date (YTD) return⁴, even as AGIX has moved through several distinct phases of AI-related volatility.

We believe that, in addition to delivering strong returns, it is important to understand how AGIX is built to be dynamic.

AGIX is an actively managed strategy led by AI-native investors. By drawing on industry expertise in AI narrative shifts and translating it into a dynamically adjusting portfolio, it seeks to avoid locking into a static view of who the "AI winners" should be. Additionally, AGIX has a three‑bucket framework (AI Hardware, AI Infrastructure, and AI Applications) that is designed to sit on top of the evolving AI story and capture value across the entire AI ecosystem.

In AI Hardware, the memory chip boom has produced some of AGIX's biggest individual winners, but also sharp ups and downs as expectations for AI demand, pricing, and supply have changed. The PHLX Semiconductor Sector Index (SOX Index), a popular AI Hardware benchmark tracking the 30 largest U.S.-traded semiconductor design, manufacturing, and sales companies, is up +60.21% YTD but lost -20.58% in July.⁴ Meanwhile, in July, AGIX only fell -9.19% amid volatility in AI Hardware names.⁴

In AI Infrastructure, major cloud platforms have delivered mixed signals. The five largest hyperscalers (Amazon, Microsoft, Google, Meta, and Oracle) have committed over $660 billion in 2026 capital expenditures (CapEx), which has offset strong AI initiatives with concerns about near‑term spending patterns and margins.¹¹ We believe that investor caution on AI Infrastructure spending (which began as early as 2025) has shifted to "CapEx fatigue" and a broader AI downdraft. Additionally, we believe it may have contributed to the sharp July drop in AI Hardware names, as investors began pricing in heightened CapEx expectations and sentiment throughout the month, ahead of hyperscalers' earnings in late July.

Finally, in AI Applications, we believe the "SaaSpocalypse" has forced investors to rethink which application‑layer businesses are genuinely enhanced by large language models (LLMs) and which are more at risk of being rewired. In January of 2026, when volatility in the software sector occurred, the S&P North American Expanded Technology Index (SPNASEUT Index) closed the month down -14.63%.¹² During that same timeframe, AGIX only fell -2.89% amid volatility in AI Software names.¹²

The goal is not to eliminate volatility, but to use an active approach to keep the portfolio aligned with where AI adoption and economics are actually showing up. AI ETF AGIX is designed to move with the theme. Even through the volatility events of 2026, AGIX has returned +17.01% YTD.⁴

Conclusion

AGIX's evolution over the past year illustrates what makes the fund structurally distinct: its private company class is not static. What began as two high-conviction bets on frontier AI development, Anthropic and xAI, has grown into a differentiated portfolio of companies operating across autonomous driving software, prediction markets, photonics infrastructure, and physical AI. Each represents a corner of the emerging AI economy that we believe public markets have yet to fully price in.

---

Holdings are subject to change.

For AGIX standard performance, top 10 holdings, risks, and other fund information, please click here.

Citations:

1. xAI Company Website, "xAI joins SpaceX," as of 2/2/2026, and Bloomberg as of 6/12/2026.
2. Anthropic Company Website, "Anthropic raises $65B in Series H funding at $965B post-money valuation," as of 5/28/2026.
3. Anthropic Company Website, "Anthropic confidentially submits draft S-1 to the SEC," as of 6/1/2026.
4. Data from Bloomberg as of 7/31/2026.
5. Tech Crunch, "Prediction marketplace Polymarket partners with Perplexity to show news summaries," as of 8/12/2024.
6. Data from PM Insights as of 7/31/2026.
7. ICE Company Website, "ICE Announces Strategic Investment in Polymarket," as of 10/7/2025.
8. Nasa Spinoff, "Humanoid Robots Assist Assembly Lines," as of 1/6/2026.
9. Apptronik Company Website, "Apptronik Closes Over $935 Million Series A," as of 2/11/2026.
10. Business Insider, "Nuro's head of partnerships breaks down how the self-driving startup is helping transform the world of delivery," as of 12/21/2021.
11. Crosley, Blake. "The AI Memory Supercycle: How HBM Became AI's Most Critical Bottleneck," Introl, as of 1/3/2026.
12. Data from Bloomberg as of 1/31/2026.

Definitions:

Annualized Run-Rate Revenue (ARR): A company's current recurring revenue pace expressed as an estimated annual amount.

Capital Expenditures (CapEx): Spending by a company on long-lived assets such as data centers, factories, equipment, or infrastructure.

Capitalization Table (Cap Table): A record of a private company's ownership, showing its shareholders, securities, and ownership percentages.

Direct Investment: An investment made directly in a company's securities rather than through another fund or intermediary.

Net Asset Value (NAV): The per-share value of a fund's assets minus liabilities, generally calculated once each trading day.

Index Definitions:

Nasdaq-100 Index (NDX): A market-capitalization-weighted index of 100 of the largest technology companies listed on the Nasdaq exchange.

PHLX Semiconductor Sector Index (SOX Index): An index designed to track major U.S.-listed companies involved in semiconductor design, manufacturing, and sales.

S&P North American Expanded Technology Index (SPNASEUT Index): An index measuring a broad group of North American technology-related companies.

Solactive Etna Artificial Intelligence Index (SOLEAGIX Index): AGIX's public-equity benchmark, designed to represent companies selected for AI-related exposure.