According to a Bank of America Institute’s report titled ‘Physical AI, part 2: Humanoid robots’, shipments of humanoid robots per year can increase from 20,000 to 10 million by the year 2035. That is a big leap forward and reveals an upcoming trend in the global market.
However, finding a stock that gives investors exposure to robotics is not that straightforward. For example, in 2025, two humanoid robotics ETFs launched within a difference of just two weeks, and yet their portfolios are quite different.
One ETF includes companies such as UBTech Robotics (9880.HK, Hong Kong) and Tesla (NASDAQ: TSLA). While another ETF gets much of its exposure through component and auto-parts suppliers. If funds revolving around a singular theme cannot agree on what counts as a robotics company, the ‘category’ of robotics itself is not reliable for investors.
In this guide, we look beyond that label of ‘Robotics’ and look deeply at some of the best robotics stocks in 2026. To make things easy, we have categorized companies into pure-play robotics businesses, picks-and-shovels suppliers, theme riders, and established industrial automation firms, and then look at what the major robotics ETFs actually hold.
Note that this article is for informational purposes only and does not constitute financial advice. Investors should conduct their own research before making investment decisions.
Quick Comparison
| Company / fund | Ticker & exchange | Exposure type | Robotics share of business | Key figure / evidence | Main risk |
| UBTech Robotics | 9880 · Hong Kong Stock Exchange | Pure-play humanoid | Majority/direct | FY2025 full-size humanoid revenue was RMB820.6 million, or 41.1% of total revenue, with 1,079 units sold (UBTech FY2025 results) | Losses and commercialization risk |
| Unitree Robotics | 688836 · Shanghai STAR Market IPO process | Pure-play humanoid | Majority/direct | Humanoids accounted for 51.78% of 2025 revenue; secondary-market trading remained unconfirmed as of Aug. 10, 2026 (SSE filing) | Listing access and China exposure |
| Symbotic | SYM · Nasdaq | US pure-play robotics | Majority/direct | About $22.5 billion in contracted backlog as of June 27, 2026 (Q3 FY2026 presentation) | Customer concentration |
| Serve Robotics | SERV · Nasdaq | US pure-play robotics | Majority/direct | Q2 2026 revenue was $3.24 million versus a $64.13 million net loss (Q2 2026 results) | Cash burn and scaling risk |
| Richtech Robotics | RR · Nasdaq | US pure-play robotics | Majority/direct | Q2 FY2026 revenue was about $1.50 million; the company has reported 450+ robot deployments (SEC filing) | Micro-cap and execution risk |
| Nvidia | NVDA · Nasdaq | Picks-and-shovels / theme rider | Minor | Q1 FY2027 revenue was $81.6 billion; Edge Computing was $6.4 billion, but robotics revenue is not separately disclosed (Nvidia Q1 FY2027 results) | Robotics exposure diluted by much larger businesses |
| Teradyne | TER · Nasdaq | Picks-and-shovels | Segment | FY2025 Robotics revenue was $308.3 million, about 10% of $3.19 billion in total revenue (FY2025 annual report) | Robotics is only part of the company |
| ABB / Fanuc / Yaskawa / Rockwell Automation | ABBN · SIX Swiss Exchange / 6954 · Tokyo Stock Exchange / 6506 · Tokyo Stock Exchange / ROK · NYSE | Industrial incumbents | Segment to significant | Established industrial automation businesses with existing commercial revenue | Cyclical capital spending |
| KOID | KOID · Nasdaq | Robotics ETF | Fund | $310.9 million in net assets as of Aug. 10, 2026 (KraneShares) | Broad definition can dilute direct humanoid exposure |
| HUMN | HUMN · Cboe BZX | Robotics ETF | Fund | Top five as of June 30, 2026 included UBTech, Tesla, Harmonic Drive, Leader Harmonious Drive and Hyundai Motor (Roundhill) | Concentration and thematic volatility |
What Actually Counts as a “Robotics Stock”?
Saying that a company is a robotics stock can mean different things. It could either refer to a company that builds robots, supplies parts used inside the robots, sells industrial automation equipment, or simply has robotics as part of a larger business.
That is also why lists of top robotics stocks can look completely different from one another. Some focus on robot makers, while others include almost any company supplying the wider automation stack.
The pure-play test
For this list, the basic test is: how much of the company’s actual business comes from robots?
Any pure-play robotics company would be the one that gets all or most of the business directly from selling robots or closely related services. This means that companies like UBTech, Symbotic, Serve Robotics, and Richtech Robotics are quite close to actual direct exposure to robotics companies rather than any other diversified tech company.
Why “humanoid” and “robotics” are not the same trade
There is another subgroup called humanoid robotics within pure-play robotics companies. Humanoid companies are the ones that are building both general or task-purpose robots that carry a form similar to humans.
Broader robotics includes warehouse automation, delivery robots, service robots, factory arms and cobots. They may benefit from some of the same advances in AI and hardware, but they are not the same market.
Picks-and-shovels: components, actuators, compute
Picks and shovels companies are those that are not likely to sell robots. These companies are instead focused on selling various parts used in robots. Actuators, precision gears, and computing chips are among a few examples.
Nvidia’s example of a ‘Theme Rider’
Nvidia (NASDAQ: NVDA) is probably the best example of what a picks-and-shovels company would be in this context. It supplies essential computing chips and platforms for the development in robotics. However, speaking of Nvidia, it does not report robotics as a material revenue area. For an investor, purchasing Nvidia’s stock is quite different from purchasing stocks from a company that directly sells robots.
Similarly, we can classify industrial automation companies within the same area. Companies like ABB, Fanuc, Yaskawa and Rockwell Automation already make money from real-world automation and robotics-related systems. While these companies may be much more established than humanoid robot startups, the performance of these stocks is unlikely to fully rely on humanoids.
Some companies may simply rely more on the software workflows rather than the hardware side of things. Companies such as UiPath focus more on robotic automation software rather than physical machines. Hence, that is an altogether separate category.
Figure, Agility Robotics and Boston Dynamics are private companies and cannot currently be bought directly through the public markets. Cryptopolitan has separately covered Agility Robotics’ potential route to becoming public.
Tesla also does not get its own entry here. Optimus might be relevant to humanoid robots, but Tesla is not currently reporting humanoid robots as a separate commercial revenue source.
So throughout this guide, “robotics exposure” ranges from companies that generate revenue directly from robots, all the way to companies that have robotics as part of a larger business.
Group 1: Pure-Play Humanoid Exposure
If we look at humanoid robot stocks, it is still quite a small and niche group in the larger market. Also, direct exposure does not necessarily correlate with easy access for investors. Let’s look at our list of pure-play humanoid companies.
UBTech Robotics (9880.HK): The clearest listed humanoid play
UBTech Robotics is perhaps the most obvious example of a publicly traded pure-play that is closest to humanoid robotics. This company, listed in Hong Kong, creates full-sized humanoid robots belonging to the Walker series, along with other education and service robots.
The humanoid robots are actually selling products too, not just a prototype. According to UBTech’s FY2025 annual results, full-size embodied humanoid robots generated RMB820.6 million in revenue, equal to 41.1% of total revenue, with 1,079 units sold during the year.
However, for investors considering this company, the lack of comfort lies in the financial picture. The same FY2025 results show a RMB789.8 million loss and RMB784.1 million in operating cash outflow. Despite sales that are increasing, the company has not yet economically scaled up.
For investors situated in the US, accessibility to this stock also happens to be a problem. It is worth noting that UBTech trades in Hong Kong rather than on a major US exchange. So, for certain US retail investments, buying this stock is not that straightforward.
Unitree Robotics: Direct exposure, but access is the problem
Unitree is another unusually direct robotics business. Its disclosed 2025 revenue breakdown showed humanoid robots accounting for 51.78% of revenue, while quadruped robots contributed another 41.62%.
That is about as concentrated a robotics revenue profile as investors are likely to find.
But Unitree is also a good example of why “public robotics stock” needs some qualification. Listing materials for its Shanghai STAR Market process identified stock code 688836 and an offering price of RMB 150.80 per share. As of August 10, 2026, secondary-market trading had not yet been confirmed.
Cryptopolitan recently reported on Unitree’s path toward a public listing. But even after it becomes tradable publicly, US investors may find it difficult to get exposure to this company due to the structure of China’s A-share market.
Group 2: US-Listed Pure-Play Robotics
The choices for US investors can hence become limited. Talking about US-listed pure-play robotics companies, Serve Robotics and Richtech Robotics all offer much more direct robotics exposure than a diversified technology company, but they are operating in very different parts of the market.
Symbotic (NASDAQ: SYM): Robotics that already has a commercial market
Symbotic helps automate warehouses and their operations by using robotics, software, and AI. Using these, it moves and organizes items across large distribution networks. Hence, logically, the main customers of Symbotic include major retail and wholesale companies.
This is the main differentiating factor here. Symbotic is not really a humanoid play; rather, it revolves around making warehouse operations more automated, where robotics will handle repetitive and time-sensitive tasks.
Based on recent reports dated June 27, 2026, Symbotic has around $22.5 billion in backlog. The business is also very concentrated, as its largest customer was responsible for about 90.5% of the company’s revenue in Q3 FY2026. It is true that a large deployment can help a business scale up but the reverse also applies if that one large customer slows down spending.
Among US-listed names, Symbotic is still one of the clearer examples of a robotics company where the robots are already tied to a real commercial use case.
Serve Robotics (NASDAQ: SERV): Delivery robots move beyond the demo stage
Serve Robotics takes an approach that is noticeably different. It has small autonomous robots designed for sidewalk delivery. These robots can move food and other items over short distances without any human interference.
The major point is that Serve is already focusing on deploying robots in the commercial market. In Q2 2026, Serve reported $3.24 million in revenue against a $64.13 million net loss. Its network included more than 2,000 sidewalk-delivery robots and more than 100 Moxi robots across 44 cities.
But it is important to remember that deployment does not always correlate with profits. This company is still at an early stage. Scaling a delivery network can have its challenges, such as paying for robots, maintenance, operations, and a lot more.
Richtech Robotics (NASDAQ: RR): Direct exposure, but much earlier
Richtech Robotics is at quite an early stage. This company sells service robots that are used in restaurants, hospitality, and other commercial industries. It also has autonomous mobile robots in development, along with a humanoid program in its very early stages.
Richtech reported $1.50 million in revenue for Q2 FY2026 and said it had deployed more than 450 robots. However, the humanoid program is quite early, and no separate revenue from this program has been disclosed to date.
This means that investors would have direct exposure to robotics and it is not a side business for this company. The caveat is that Richtech is still small and its humanoid program is early.
Group 3: Picks-and-Shovels
Every company that makes robots requires compute power, chips, motors, gears, electronics, and many more components. This means that we can have another set of companies that supply these essential items.
These companies may not be producing robots directly, but they are very much beneficiaries of the robotics market. In this group, we will look at these pick-and-shovel companies.
Nvidia (NASDAQ: NVDA): Important to robotics, but still not a robotics pure play
If you search the term ‘robotics stock’ on Google, it is likely that Nvidia will appear on every list. This is because the company is expanding into robotics, as covered by Cryptopolitan earlier.
It supplies both simulation tools and computing chips necessary to develop robots. Yet, Nvidia does not report robotics as a revenue stream. Hence, it is not the main business model for this company. After all, computer chips are used in a wide range of other industries, not just robotics.
This makes Nvidia an indirect beneficiary of the robotics market. For Q1 FY2027, which ended April 26, 2026, Nvidia reported $81.62 billion in total revenue. Its $6.37 billion Edge Computing category includes robotics-related businesses, but Nvidia still does not disclose robotics revenue separately.
Teradyne (NASDAQ: TER): A real robotics business inside a larger company
Teradyne is a somewhat cleaner case because it actually owns robotics companies.
Universal Robots makes collaborative robot arms, or cobots, designed to work alongside people in industrial environments. Mobile Industrial Robots, usually called MiR, makes autonomous mobile robots for moving materials inside factories and other facilities.
Robotics is still only one part of Teradyne. Teradyne reported $308.3 million in Robotics revenue for FY2025, about 10% of its $3.19 billion in total company revenue. Its much larger business includes automated test equipment used by semiconductor and electronics manufacturers.
Harmonic Drive Systems (6324, Tokyo) and Leader Harmonious Drive Systems (688017, Shanghai): Selling the joints
Some of the more direct pick-and-shovel exposure sits much further down the hardware stack.
Harmonic Drive Systems (6324, Tokyo Stock Exchange Prime Market) and Leader Harmonious Drive Systems (688017, Shanghai Stock Exchange STAR Market) make precision strain-wave gearing. These components are used where robots need compact, accurate movement, including joints and actuators.
Both were among the top holdings of the Roundhill Humanoid Robotics ETF (HUMN) in Roundhill Investments’ June 30, 2026 holdings snapshot.
The appeal of this part of the supply chain is fairly simple. If humanoid production eventually scales across several manufacturers, component suppliers do not necessarily have to predict which robot brand wins. More robots can still mean more demand for the parts inside them.
The risk is just as obvious: that thesis depends on production scaling well beyond demonstrations and limited deployments. If humanoid volumes disappoint, anticipated component demand can disappoint with them.
Sensors, power, and everything around the robot
The definition gets broader again with companies such as Sensata Technologies (NYSE: ST), Monolithic Power Systems (NASDAQ: MPWR), Hexagon AB (HEXA B, Nasdaq Stockholm), Horizon Robotics (9660, Hong Kong Stock Exchange) and Magna International (NYSE: MGA; TSX: MG).
They supply very different things. Sensata has sensing technologies, Monolithic Power makes power-management semiconductors, Hexagon operates in measurement and industrial technology, Horizon Robotics develops computing technology for intelligent vehicles, while Magna is a major automotive supplier.
None should be mistaken for a pure robotics company. The tradeoff is diluted exposure. Their shares can be driven much more by automotive, semiconductor, metrology or broader industrial demand than by robot production itself.
According to KraneShares’ July 31, 2026 holdings data, all five were the largest positions in the KraneShares Global Humanoid Robotics and Physical AI Index ETF (KOID). Hexagon accounted for 2.59% of the fund, followed by Horizon Robotics at 2.44%, Magna at 2.34%, Monolithic Power at 2.33% and Sensata at 2.27%.
That tells us something important about the sector.
Once “robotics” expands into “physical AI,” the investment universe can quickly move away from companies actually selling robots and toward the much wider supply chain around them.
For some investors, that diversification may be the point. But it is a very different exposure from owning UBTech, Symbotic or Serve Robotics.
Group 4: Industrial Incumbents
Not every robotics company is a start-up. If we look at ABB, Fanuc, Yaskawa Electric and Rockwell Automation, these companies are selling automation equipment to different global factories. Talking about industrial motion control systems and robot arms, Yaskawa and Fanuc particularly have high exposure. At the same time, Rockwell is more toward the industrial control and automation side.
For investors interested in humanoid stocks, this group might appear less interesting. However the tradeoff is that this group already has a commercial base and revenue stream. They would hence be more stable than a humanoid startup.
But then again, industrial automation revolves around factory investments and capital-expenditure cycles. So if the demand for manufacturing in global industries is slow, the longer-term automation trends would also slow down.
Robotics ETFs: What’s Actually Inside Them
Finding the best robotics ETF is not as simple as choosing the fund with “robotics” or “humanoid” in its name. A basket can spread company-specific risk, but what investors actually own still depends heavily on the holdings.
And there is another problem: the funds themselves do not agree on what robotics exposure should look like.
The clearest example is KOID and HUMN. Both launched in June 2025 around the humanoid robotics theme. Look inside them, though, and they are making very different bets.
The comparison below uses fund data published by KraneShares for KOID and Roundhill Investments for HUMN. KOID’s holdings are dated July 31, 2026, while HUMN’s holdings snapshot is dated June 30, 2026.
| KOID | HUMN | |
| Full name | KraneShares Global Humanoid Robotics and Physical AI Index ETF | Roundhill Humanoid Robotics ETF |
| Launched | June 4, 2025 | June 26, 2025 |
| Style | Index-based, tracking the MerQube Global Humanoid Robotics and Physical AI Index | Actively managed |
| Expense ratio | 0.79% gross / 0.69% net; fee waiver stated through Aug. 1, 2028 | 0.75% |
| Top holdings | Hexagon AB-B 2.59%, Horizon Robotics 2.44%, Magna 2.34%, Monolithic Power 2.33%, Sensata 2.27% as of July 31, 2026 | UBTech, Tesla, Harmonic Drive, Leader Harmonious Drive and Hyundai Motor as of June 30, 2026 |
| What you are mostly buying | Components, sensors, metrology and the wider physical-AI supply chain | More concentrated humanoid and humanoid-adjacent exposure |
KOID’s holdings show how broadly a humanoid robotics fund can define the opportunity. Rather than concentrating on companies that build finished robots, much of the portfolio sits further down the physical-AI supply chain, including sensing, computing, measurement and automotive components.
The logic is that these suppliers may still benefit if humanoid production scales, even if the eventual winning robot brands are difficult to predict.
HUMN takes a different route. Its June 30, 2026 holdings snapshot put UBTech, Tesla, Harmonic Drive, Leader Harmonious Drive and Hyundai Motor among its largest positions.
So despite launching only three weeks apart, the two funds barely look alike at the top of their portfolios. KOID is closer to a diversified physical-AI supply-chain basket. HUMN puts more weight behind companies more directly tied to humanoid development and hardware.
That also changes the risk. That makes HUMN more exposed to changes in sentiment around humanoid-focused names. KOID spreads that exposure across a much wider group of component and industrial companies, but in doing so it becomes less of a direct humanoid bet.
The Themes Humanoid Robotics ETF (BOTT) is another fund built specifically around the humanoid theme. But its presence also reinforces the same problem seen with KOID and HUMN: a fund name alone does not tell investors how directly the portfolio is tied to companies actually building humanoid robots.
That is why holdings, weighting methodology and fees matter more than the label. A humanoid ETF can lean toward robot makers, component suppliers, semiconductors or a mixture of all three.
The older generation looks different again. The Global X Robotics & Artificial Intelligence ETF (BOTZ), ROBO Global Robotics and Automation Index ETF (ROBO), and ARK Autonomous Technology & Robotics ETF (ARKQ) all predate the current humanoid boom and take a broader view of robotics and automation.
Their portfolios are therefore more likely to capture industrial automation, autonomous systems and the wider robotics ecosystem rather than a concentrated bet on humanoid commercialization. For someone searching for the best robotics ETF, that distinction matters more than simply comparing fund names.
The Risks Nobody Puts in the Listicle
Perhaps the biggest risk associated with robotics is that not all robotics companies will make it big, even if the larger market continues to grow. We would say that is true for companies in all markets, though, not just robotics.
The biggest gap for companies right now is to show that the robots can actually be useful in the long term. Of course, humanoid robots will be able to walk and lift objects in controlled environments, with much more ease than an actual human can. But would they actually perform similarly well in a commercial or everyday setting? Hence, the return on investment on humanoid robots remains a question.
Due to this very reason, the shipment forecasts do not necessarily translate to the fact that the manufacturers already have a large number of orders. If the deployment of these robots in the market is too slow and their value is not established, companies may hit a wall.
Next comes the risk with policies, especially revolving around China. Chinese companies are trying to advance robotics at a quick speed. They are already trying to automate industries with the help of robots. For example, BYD was reportedly developing humanoid robots for its own factories. Yet restrictions on imports and tariffs could mean slow growth for companies.
Cryptopolitan has previously reported on US action against Chinese state-subsidized robotics imports.. Investors also have to deal with different listing structures and market-access rules. UBTech trades in Hong Kong, while Unitree’s route to the public market is through Shanghai’s STAR Market.
Concentration can show up in less obvious places too. Symbotic depends heavily on a relatively small group of large customers. Thematic ETFs can have a different version of the same problem. HUMN is more concentrated around humanoid and humanoid-adjacent names, while KOID spreads exposure across a broader physical-AI supply chain.
Finally, there is sentiment risk. Robotics stocks can reprice quickly when commercial results fail to keep pace with demonstrations, deployment targets and long-range forecasts. That can hurt early-stage names even when the underlying technology is still progressing.
None of these risks means robotics adoption has to stop. They mean the technology can keep improving while individual stocks, business models and even entire parts of the theme struggle along the way.
Final Verdict: Decide Which Exposure You’re Actually Buying
There is no single formula for choosing between robotics stocks because investors are not really choosing from one market.
UBTech offers some of the most direct listed humanoid exposure, while Unitree is similarly concentrated in robotics but remains subject to listing and access constraints. Symbotic, Serve Robotics and Richtech give US investors more direct public-market exposure, although the businesses range from established warehouse automation to much earlier-stage service robots.
Picks-and-shovels companies such as Nvidia, Teradyne and precision-component suppliers offer another route. They may benefit even if the eventual winners in humanoid robotics are still unknown, but the tradeoff is less direct exposure.
Industrial incumbents sit at the more established end of the spectrum, while robotics ETFs spread the risk further. Even there, the holdings matter more than the label. KOID and HUMN showed just how differently two funds can interpret the same humanoid theme.
The better question, then, is not which robotics stock is best. It is what kind of robotics exposure an investor actually wants.