PHYSICAL AI & AUTOMATION
What the Q2 2026 rebalance tells us about where the robotics and AI story may be headed next
Every quarter, the ROBO Global Robotics & Automation Index and the ROBO Global Artificial Intelligence Index reconstitute and rebalance their constituents to capture a rapidly shifting landscape as secular industry trends continue to evolve. The additions and deletions go beyond portfolio housekeeping and reflect the index provider’s systematic methodology-driven view of where leadership (both technical and market) and innovation intensity currently sit within the physical and digital AI landscape. This quarter's changes tell a story: robots are gaining better senses and stronger limbs, working faster and processing more. Business software is being carved out and repackaged around automation and the AI infrastructure stack is consolidating around the chipmakers who move and process the data behind it all.
ROBO Global Robotics & Automation Index
Investment thesis: seeks to own the full value chain of the robotics economy which include the sensors that let machines perceive the world, the actuators that let them move through it and the software that automates the business processes around them.
At a Glance
| Action | Company | Subsector | One-Line Thesis |
|---|---|---|---|
| Added | Ouster (OUST) | Sensing | Chip-based lidar that gives machines depth perception at a price they can afford |
| Added | Schaeffler (SFFLY) | Actuation | A century of bearing know- how repackaged into the joints of humanoid robots |
| Added | Octave (Octave) | Business Process Automation | Hexagon's industrial software arm, now standalone, automating the sset lifecycle |
| Removed | Shibaura Machine (6104.T) | Factory Automation | A 150-year-old machine-tool maker that has been outpaced by higher-purity robotics plays |
The Additions
INDEX ADDITION SENSING
Ouster OUST
From the beginning, Ouster believed lidar* (Light Detection and Ranging) could be built differently. Instead of using the large, spinning mechanical systems common at the time, the company focused on a semiconductor-based design. Lidar acts as the "eyes" of autonomous machines, using laser pulses to create highly detailed 3D maps that allow robots, vehicles and industrial equipment to detect obstacles, navigate their surroundings and operate safely. A decade later and following its merger with Velodyne, Ouster has become the leading pure-play digital lidar company. Its single-chip CMOS architecture puts lidar on the same technology curve that has driven decades of innovation in semiconductors by delivering lower costs, higher performance and greater reliability with each generation. As lidar becomes more affordable, it moves from being a premium feature into a core enabling technology for delivery robots, autonomous trucks, warehouse automation, industrial robotics and smart infrastructure.
Why it's interesting now: Ouster is evolving beyond making sensors to helping machines understand what they're seeing. Its acquisition of Stereolabs added AI-powered cameras that interpret visual information, while its new REV8 OS lidar family combines highly detailed laser mapping with color imagery to give robots and autonomous vehicles a richer, more complete view of their surroundings. Instead of relying on a single sensor, Ouster is building systems that combine multiple types of data, making automation smarter, safer and more reliable.
By the numbers: The business is beginning to show greater operating leverage. Ouster generated $169 million in 2025 revenue, a 52% increase from the prior year, while substantially reducing its annual net loss as higher sales and improved margins supported financial performance.1
| TIES TO THE INVESTMENT THESIS Sensing is one the entry point to physical AI: a machine needs to perceive its environment before it can do anything productively in it. Ouster joins Ambarella (computer-vision silicon) and Jenoptik (photonics and optical sensing) in the index’s Sensing bucket giving the index a more complete picture of how robots are learning to see. |
INDEX ADDITION ACTUATION
Schaeffler SFFLY
Schaeffler built its reputation over more than 75 years on precision bearings before compounding that mechanical expertise across the auto, industrial and aerospace sectors. The company's own tagline is "We pioneer motion." Schaeffler is now applying that engineering heritage to humanoid robotics, combining bearings, motors, gearing, encoders and controllers into fully integrated actuator systems designed to power approximately 80% of a humanoid robot's joints. In simple terms, if the robot's software serves as the brain, Schaeffler builds the muscle and joints that enable it to move with strength, precision and control.
Why it's interesting now: Rather than betting on a single humanoid robot maker winning the race, Schaeffler maintains broad exposure across multiple manufacturers, establishing a classic "picks and shovels" position for the space. Schaeffler strengthened its position in humanoid robotics through a strategic partnership with Hexagon Robotics to supply high-precision rotary actuators for the AEON humanoid robot. The company also unveiled a back-drivable planetary actuator at CES 2026, highlighting its focus on production-ready motion systems designed specifically for next-generation humanoid robots.6
By the numbers: Schaeffler shares have significantly outperformed as it appears investors increasingly recognize the company's growing exposure to humanoid robotics, helping shift its perception from a traditional automotive supplier toward a key provider of robotics components.7
| TIES TO THE INVESTMENT THESIS Component suppliers are the lower-variance way to own the humanoid robotics wave, since they sell into the whole field instead of picking a single winner. Schaeffler deepens the index’s existing Actuation bench alongside Harmonic Drive Systems, THK, and Hiwin. The idea being that those who build the joints get paid regardless of which humanoid brand ultimately wins from a public adoption standpoint. |
INDEX ADDITION BUSINESS PROCESS AUTOMATION
Octave OCTV
Octave may be a fresh face in public markets, but its solutions have been helping some of the world's largest industrial operators run more efficiently for years. Created as a spin-off from several of Hexagon's divisions, including its industrial software businesses and Asset Lifecycle Intelligence division, Octave became an independent, publicly traded company in May 2026.
Think of Octave as the digital operating system for critical infrastructure. Its software connects the information generated during the design, construction, operation and maintenance of complex assets such as power plants, factories, refineries, utilities and transportation networks into one digital view. Instead of relying on disconnected drawings, spreadsheets, maintenance records and inspections, operators can monitor equipment, anticipate problems before they become costly failures, improve safety and make better decisions throughout an asset's life. As AI, sensors and automation generate ever-larger volumes of operational data, platforms like Octave help transform that information into actionable insights.
Why it's interesting now: Octave joined the Index as a result of Hexagon's spin-off, with Hexagon shareholders receiving one Octave share for every ten Hexagon shares they owned. While the company's inclusion reflects that corporate action, its software aligns closely with the investment thesis by providing the digital intelligence layer that helps make industrial operations more connected, automated, and efficient.
| TIES TO THE INVESTMENT THESIS This is the index’s Business Process Automation subsector at work: robotics isn't only about physical machines; it's also about automating the workflows and decisions that run the industrial world around them. Octave keeps that side of the thesis represented as Hexagon's businesses continue to separate into more focused, pure-play pieces. |
The Removal
INDEX REMOVAL FACTORY AUTOMATION
Shibaura Machine 6104.T
Shibaura Machine traces its roots back to 1875 and spent decades operating as Toshiba Machine before adopting the Shibaura name in 2020. Today it's a leading global maker of precision injection-molding machines, machine tools, and die-casting equipment, with its TM Robotics line of 6-axis, SCARA and Cartesian robots living inside its control-systems segment.
Why it's interesting now: This is a relative-ranking exit, not a fall from grace. The index scores every constituent on technical leadership, market leadership and innovation intensity and Shibaura has been ceding ground on all three: its robotics categories are mature and crowded, its growth tracks cyclical machine-tool and molding demand. investors continue to move more towards cutting-edge AI technologies as opposed to the traditional automation players.
| TIES TO THE INVESTMENT THESIS Shibaura held its index seat for years and simply got out-scored by higher-purity pure-plays like Ouster and Schaeffler. The broader takeaway is that the index is designed to evolve alongside the robotics and automation industry. As new technologies emerge and leadership changes, the index seeks to reflect the companies that are helping shape the next generation of innovation. |
ROBO Global Artificial Intelligence Index
Investment thesis: seeks to own the highest conviction segments and purest expressions of the AI value chain including; the chips, infrastructure and platforms most directly levered to AI adoption. With the AI landscape shifting from large scale model training to running (inferring from) them more efficiently and cheaply at scale, the index rotates towards the primary beneficiaries positioned to capture this next wave of the transition.
At a Glance
| Action | Company | Subsector | One-Line Thesis |
|---|---|---|---|
| Added | Cerebras Systems (CBRS) | Semiconductors | Wafer-scale AI chips built for the inference era, the most credible US NVIDIA challenger |
| Added | Marvell Technology (MRVL) | Semiconductors | Custom hyperscaler silicon and the interconnects that stitch AI data centers together |
| Removed | HubSpot (HUBS) | Applied Software / CRM | Seat-based CRM pricing under pressure as AI compresses the applied- software moat |
| Removed | CoStar Group (CSGP) | Applied Software / Data | A data-moat marketplace model facing a higher bar as generative search reshapes the category |
The Additions
INDEX ADDITION SEMICONDUCTORS
Cerebras Systems CBRS
Cerebras started in 2015 with a bet most of the chip industry considered absurd: instead of carving a silicon wafer into hundreds of individual chips, build the entire wafer as one processor. A decade later, that bet is the WSE-3 which is the largest chip ever made, roughly the size of a dinner plate, packing 4 trillion transistors and 900,000 cores. It’s like replacing an entire room full of computers working together with one extraordinarily powerful supercomputer. Fewer moving pieces mean less time spent coordinating and more time computing. The company that set out to build "impossible technologies" now positions itself as the fastest AI inference and training platform available and the most credible US challenger to NVIDIA at the silicon level.
Why it's interesting now: Cerebras is squarely positioned for the industry's biggest structural shift: as AI workloads tilt from training new models toward serving existing ones at scale, architectures that sidestep the memory-bandwidth bottleneck of clustered GPUs may become more valuable. Cerebras claims up to roughly 15x GPU speed on supported models. It IPO'd on Nasdaq on May 14, 2026, popped about 68% on day one, and raised $5.55 billion, briefly the largest US tech IPO since Uber in 2019, before SpaceX's offering topped it.2
By the numbers: IPO raised $5.55B; shares rose ~68% on debut.2
| TIES TO THE INVESTMENT THESIS Cerebras reinforces the index’s semiconductor backbone alongside compute leaders like AMD (Advanced Micro Devices) whose multi-year AI infrastructure partnership with OpenAI underscores the industry's enormous demand for computing power. The addition represents a direct investment in the idea that the next phase of AI infrastructure spending may be increasingly driven by inference (the process of putting trained AI models to work) rather than training alone. |
INDEX ADDITION SEMICONDUCTORS
Marvell Technology MRVL
Marvell spent its early decades as a storage and networking chip company, building the silicon that powers hard drives and network switches. Over the past few years, it has reinvented itself around AI data centers and is now one of just two leading independent designers of custom AI chips for the world's largest cloud providers.3 Rather than selling the same chip to every customer, Marvell designs application-specific integrated circuits (ASICs) - purpose-built processors, often called XPUs, that are tailored to the unique AI workloads of companies like Amazon, Microsoft and Google. These custom chips can deliver higher performance and greater efficiency while helping reduce reliance on NVIDIA's general-purpose GPUs. Marvell also provides the high-speed networking technology (including optical connections, Ethernet switches, digital signal processors (DSPs), and storage controllers) that allows thousands of servers and AI chips to communicate with one another, effectively acting as the digital plumbing that keeps modern AI data centers running.
Why it's interesting now: The Marvell thesis is converting from its paper-based designed wins to recognized scaling revenue. Data center now makes up about 75% of the business and two hyperscale custom-compute programs hit high-volume production in early 2026, with a 2nm ASIC platform already on the roadmap.3,4
By the numbers: Q3 FY2026 data-center revenue reached about $1.52 billion, up 38% year over year.5
| TIES TO THE INVESTMENT THESIS Marvell extends the AI-connectivity and custom-silicon theme the index has been building through names like Astera Labs, Credo, and Lumentum on the optical side. It sits near the center of that map, levered directly to hyperscaler capital spending regardless of which AI model or application ultimately wins. |
The Removals
INDEX REMOVAL APPLIED SOFTWARE / CRM
HubSpot HUBS
HubSpot built its name on inbound marketing and has broadened over the years into a full customer platform, now reframed around an AI-powered "Smart CRM" and agentic workflows, still carrying the "grow better" banner it launched with.
Why it's interesting now: The change reflects how the industry is evolving, rather than any deterioration in the company's execution. Top- and bottom-line growth face headwinds as AI compresses seat-based software pricing and erodes the moat that used to protect applied-software business models. On the index’s refreshed AI-purity ranking, an applied-CRM name now sits below the infrastructure leaders coming in.
| TIES TO THE INVESTMENT THESIS This is the flip side of the Cerebras and Marvell additions: as the index tilts harder toward the AI backbone (chips and infrastructure) application-layer software names that merely use AI, rather than enable it, become comparatively less "pure" expressions of the theme. |
INDEX REMOVAL APPLIED SOFTWARE / DATA
CoStar Group CSGP
CoStar started in 1987 as a commercial real estate data provider, built that into an industry standard and layered on marketplaces (LoopNet, Apartments.com, Homes.com) to become, in its own words, commercial real estate's leading provider of information, analytics and online marketplaces. It has been an index constituent for years.
Why it's interesting now: The removal reflects a changing competitive landscape rather than weakening business performance. As generative AI reshapes how users discover information, companies whose competitive advantage has traditionally come from proprietary data may face greater competitive pressure and on the index’s refreshed relative ranking it now sits below the semiconductor and infrastructure names taking its place.
| TIES TO THE INVESTMENT THESIS The fundamentals here remain strong; the open question the index is implicitly asking is whether a vertical data-and-marketplace business stays a top-purity AI expression as the value chain reshapes around the infrastructure layer. For now, it is answering that question by rotating toward the picks-and-shovels providers instead. |
What This May Mean for Investors
- Robotics is maturing from mechanical to sensory. The ROBO Global Robotics & Automation Index’s rotation toward Ouster (perception) and Schaeffler (actuation) reflects a broader industry shift: the hard part of robotics is no longer just building a machine that can move, it's building one that can sense, interpret and react to its environment.
- Component suppliers offer diversified exposure to the humanoid race. Rather than picking which humanoid robot company wins, the ROBO Global Robotics & Automation Index is leaning into suppliers like Schaeffler that profit regardless of which brand ultimately dominates.
- AI's center of gravity is shifting from training to inference. The ROBO Global Artificial Intelligence Index’s addition of Cerebras is a direct bet that serving AI models cheaply at scale may be where the next wave of infrastructure investment lands.
- Custom silicon is going mainstream at the hyperscalers. Marvell's addition tracks the growing hyperscaler push to reduce NVIDIA dependence through custom ASICs. An emerging trend with a potential long runway as more cloud providers pursue their own chip programs.
- The AI-purity bar keeps rising for application-layer software. HubSpot and CoStar's removals signal that being an AI "user" isn't enough to stay in a high-purity AI index. The companies enabling AI at the infrastructure level are increasingly favored over those simply applying it to an existing business model.
The ROBO Global Robotics & Automation Index ETF (ROBO) seeks to track the performance of the ROBO Global Robotics & Automation Index. Click here to learn more about ROBO.
The ROBO Global Artificial Intelligence ETF (THNQ) seeks to track the performance of the ROBO Global Artificial Intelligence Index. Click here to learn more about THNQ.
Sources:
* LiDar: (Light Detection and Ranging) is a remote sensing technology that uses pulsed lasers to measure variable distances and build precise, three-dimensional models of objects and environments.
1. Ouster, Inc., "Ouster Announces Results for Fourth Quarter and Full Year 2025," news release, February 26, 2026.
2. CNBC. 2026. "Cerebras Pops 68% in Nasdaq Debut, Pushing the AI Chipmaker's Market Cap to $95 Billion." May 14, 2026.
3. Reuters. "Marvell Sees Custom Chip Revenue Topping $10 Billion by 2029 as AI Demand Grows." May 27, 2026.
4. Marvell Technology, Inc. Custom AI Investor Event. Accessed July 3, 2026.
5. Marvell Technology, Inc., Marvell Technology, Inc. Reports Third Quarter of Fiscal Year 2026 Financial Results, December 2, 2025.
6. Schaeffler AG, "Humanoid Robotics: Schaeffler Enters into Strategic Partnership with Hexagon Robotics," news release, April 22, 2026.
7. Reuters, "Schaeffler Sees Humanoid Robotics Orders in Three-Digit Million Euros by 2030," May 5, 2026.
Carefully consider the Funds’ investment objectives, risk factors, charges and expenses before investing. This and additional information can be found on the Funds' full or summary prospectuses, which may be obtained at www.roboglobaletfs.com. Read the prospectus carefully before investing.
Investing involves risk, including the possible loss of principal. International investments may also involve risk from unfavorable fluctuations in currency values, differences in generally accepted accounting principles, and from economic or political instability. Emerging markets involve heightened risks related to the same factors as well as increased volatility and lower trading volume. Narrowly focused investments and investments in smaller companies typically exhibit higher volatility. There is no guarantee the funds will achieve their stated objective. ROBO and HTEC are diversified. THNQ is non-diversified.
The liquidity of the A-shares market and trading prices of A-shares could be more severely affected than the liquidity and trading prices of other markets because the Chinese government restricts the flow of capital into and out of the A-shares market. The funds may experience losses due to illiquidity of the Chinese securities markets or delay or disruption in execution or settlement of trades.
The risks associated with investments in Robotics and Automation Companies include, but are not limited to, small or limited markets for such securities, changes in business cycles, world economic growth, technological progress, rapid obsolescence, and government regulation. Robotics and Automation Companies, especially smaller, start-up companies, tend to be more volatile than securities of companies that do not rely heavily on technology. Rapid change to technologies that affect a company's products could have a material adverse effect on such company's operating results. Robotics and Automation Companies may rely on a combination of patents, copyrights, trademarks and trade secret laws to establish and protect their proprietary rights in their products and technologies. There can be no assurance that the steps taken by these companies to protect their proprietary rights will be adequate to prevent the misappropriation of their technology or that competitors will not independently develop technologies that are substantially equivalent or superior to such companies' technology.
The risks associated with Artificial Intelligence (AI) Companies include, but are not limited to, small or limited markets, changes in business cycles, world economic growth, technological progress, rapid obsolescence, and government regulation. Rapid change to technologies that affect a company’s products could have a material adverse effect on such company’s operating results. AI Companies also rely heavily on a combination of patents, copyrights, trademarks and trade secret laws to establish and protect their proprietary rights in their products and technologies. There can be no assurance that the steps taken by these companies to protect their proprietary rights will be adequate to prevent the misappropriation of their technology or that competitors will not independently develop technologies that are substantially equivalent or superior to such companies’ technology. AI Companies typically engage in significant amounts of spending on research and development, and there is no guarantee that the products or services produced by these companies will be successful.
The risks associated with Medical Technology Companies include, but are not limited to, small or limited markets for such securities, changes in business cycles, world economic growth, technological progress, rapid obsolescence, and government regulation.
Diversification may not protect against market risk.
Beginning September 2, 2020, market price returns are based on the official closing price of an ETF share or, if the official closing price isn't available, the midpoint between the national best bid and national best offer (“NBBO”) as of the time the ETF calculates current NAV per share. Prior to September 2, 2020, market price returns were based on the midpoint between the Bid and Ask price. NAVs are calculated using prices as of 4:00 PM Eastern Time. The returns shown do not represent the returns you would receive if you traded shares at other times.
The Funds are distributed by SEI Investments Distribution Co. (SIDCO) 1 Freedom Valley Drive, Oaks, PA, 19456