We led Standard Bots’ $200 million Series C because we believe it’s the company built to reinvent the robot arm for physical AI and to bring the industry back to America.
In this public thesis we will dive into:
- The birth of the robot
- A typical robot arm’s hardware and software architecture
- The technical limitations that hold back today’s arms
- How Standard Bots is reinventing the arm from first principles
Written by @JacklouisP
Introduction: The workhorse of automation
Six-axis arms are the dominant general-purpose robot. 90% of all automotive spot welds are done by a robot arm, they tend CNC machines around the clock, stack pallets by the billion, paint, polish, assemble, and inspect. They’ve done this for sixty years.
Why is the arm the dominant design for general robots? It's the optimal mix of simplicity and flexibility. Any object in space has six degrees of freedom, three of position and three of orientation, and a machine that does general work needs all six. There are many ways to get there, but if you optimize for simplicity and reach at a fixed location, the arm wins: six motorized joints stacked in a chain.
621,000 industrial robots were installed in 2025, up 15% on the year, with more than 4.6 million in operation.¹ Arm hardware generates roughly $20 billion a year in revenue.² We believe that number badly underplays what comes next.
The better measure of the opportunity is not the robot market but the labor within an arm's reach. Morgan Stanley's robotics research puts the US payroll addressable by robots at roughly $3 trillion a year by 2050, within a $5 trillion global market; Citi's estimate runs as high as $7 trillion.³ By our analysis of US occupational data, a quarter to a third of that payroll, roughly $600 to 700 billion a year, is stationary work done at a bench, a machine, or a line. That work doesn't need legs. It needs an arm that can be taught.
Arms often get overlooked, written off as commoditized, but the reality is the opposite: there is an enormous amount of space left for growth and progress. Robot density is rising in every industrialized country, yet even South Korea, the world leader, runs just one robot per ten manufacturing workers. We believe that ratio keeps climbing everywhere, with no obvious ceiling at one to one.⁴
There are two factors that explain why we believe Standard Bots is positioned to drive this progress:
The first might surprise you: America is not a real player in this space. The industry is dominated by Europe and Japan, with a rapid shift toward China. The country that invented robot arms has no horse in the race and this is a major strategic gap. Until now.
Second: incumbents have stopped innovating. Arms have been around so long we assume the design is fixed. The reality is the architecture has been frozen since the 1980s and it requires dramatic changes to adapt to AI.
To understand Standard Bots, you need to understand the robot arm: where it came from, and how it works. The next sections are that deep dive. If you’d rather cut straight to the thesis, skip ahead to “What Makes Standard Bots different?”
Here is a teaser: Standard Bots approaches robotics the way a frontier research lab would, while owning the customer relationship and delivering real industrial value today. That combination, we believe, could make it the leader in physical AI, a market forecast to surpass $430 billion in revenue by 2030.⁵
Problem 1: America invented the robot arm, then lost it
The industrial robot is an American invention. In 1961, the world’s first industrial robot arm, Unimation’s Unimate, went to work on a GM line in Ewing, New Jersey.⁶
For two decades, Unimation was the industry leader, but this wasn’t set to last.
First, the technology moved. The industry shifted from hydraulic robots, Unimation’s specialty, to electric ones. Japanese makers built for the future paradigm, Unimation built for the previous.⁷
As a result, its biggest customer defected. In 1982, GM, the largest robot buyer in America, formed a joint venture with FANUC, Unimation’s Japanese rival. Every robot GM bought from GMFanuc shrank the market for the domestic players.⁸ Westinghouse bought Unimation for $107 million in 1983, 5 years later it was sold off to Staubli of Switzerland. The company that invented the industry was gone within a decade of its peak.⁹
A 1983 US government report diagnosed this strategic blunder: American robots were complex, expensive, and needed frequent maintenance. Japanese and European machines were simpler, cheaper, and more performant.¹⁰ There is a lot we can learn from this story.
Now the industry is shifting again. Midea of China bought KUKA of Germany in 2016.¹¹ ABB of Switzerland is selling its robotics division to SoftBank of Japan for $5.4 billion.¹² This shift is accelerating as Chinese systems tend to be priced 20 to 40% below Western equivalents.¹³
The fact that the US lacks a local champion has always been a glaring gap and as European companies shift east, this gap has become even more critical to fill. Washington has noticed: the proposed American Security Robotics Act and GUARD Act would restrict Chinese robots from federal use, with bipartisan backing.¹⁴
Robot sovereignty isn’t protectionism; it’s about strategic relevance. We firmly believe that robotics is the most important tech trend of the coming decades and to successfully compete we can’t rely on strength in the model layer, we need a fully functioning local supply chain that comes from the creation of a local champion.
The real lesson of the Unimation story: America can’t win by building a cheaper copy of a European or Japanese robot. This is China’s strength, the only way for the US to compete and win is: by changing what the product is.
The arm doesn’t just need to be reshored; it needs to be reinvented.
Problem 2: The incumbents stopped innovating
Fundamentally, robotics is a systems problem. Every component of the modern robot arm was designed for the last generation of automation: point-to-point, fixed-path programming. Record a path, replay it forever with no intelligent course correction.
Systems built that way can’t deal with variance. A different part, a changed process, a shift in lighting or fixturing, and the cell stops.
Solving variance expands the number of applications that are profitable to automate. It means:
- Robots that can handle a change in plan and adapt on the fly.
- Lower skill barrier, so their own people can deploy and retask the robot.
- Lower system cost, by bringing deployment in-house instead of paying integrators.
Current architecture can’t keep up, and that is the single biggest blocker to the next phase of automation growth.
AI is meant to solve this, but most of the AI push comes from model companies tackling one component of a complex system. That’s never how you get performance: the model is only as good as the motors, sensors, firmware, and interfaces underneath it, and none of them were designed for this new job.
To see why, you must open the arm up. What follows is a component-by-component look at how the machine works, what the choices are, and where the limitations bind.
How an arm works
First, know your arms. Cartesian robots run on three linear axes: rigid, cheap, boxed into their frame. SCARA arms are planar specialists, quick and precise but confined to one plane. Delta robots are the sprinters of the family - lightweight arms driven from a stationary base, capable of over a hundred picks a minute, but limited to small payloads in a small work volume.
Then there’s the articulated six-axis arm, inspired by the human arm, and the dominant design for a reason. It’s the optimal design for speed, flexibility, and cost. One arm, thousands of possible jobs.
The architecture across articulated arms is relatively consistent. A six-axis arm is a kinematic chain: six motorized joints in series, a controller cabinet, and software on top.
The controller plans a trajectory, chops it into position setpoints, and streams them to each joint’s servo drive. The drive switches current through the motor, the reducer converts speed into torque, the encoder reports position back. At the end of the arm, you have an end effector tailored to your job plus sensors to add real world feedback. This hasn’t changed for sixty years.
Before diving into each component, it's useful to illustrate why this system doesn't fit modern manufacturing. Take a deceptively simple factory task: inserting a cable into a connector.
The task requires aligning two parts that vary slightly in shape, then correcting live as contact unfolds, adjusting the angle of attack until the right force says the plug has seated.
The legacy robot fails. It replays a recording, and a recording can't capture variation. It can't sense the misalignment, can't course-correct mid-motion, and can't feel when the connection is made. Why? As we'll see: the program is fixed, set once at deployment by an integrator to run deterministically; the feedback loop is low frequency; and there is limited torque sensing. Each failure lives in a component, and the components are where we go next.
Decomposing a legacy arm
A robot is a full system and every component matters, but four rows do the heavy lifting: the motors and reducers together decide how well it moves and what it costs, the controller decides what the arm can respond to, and the programming decides who can use it at all.
Motors and drives
What they do. Motors turn current into motion and are powered and controlled by the drive electronics. Together they set the ceiling on payload, smoothness, speed, and how finely the joint can be controlled.
The choices.
Line up the arms from ABB, FANUC, KUKA, Yaskawa, and Universal Robots and the resemblance isn't branding. It's the same machine in different colors, because underneath the paint they draw from the same parts bin. Kollmorgen sells a "robot-ready" frameless motor line in 21 standard variants, sized for cobots; maxon, Nidec, Moog, and Novanta sell equivalents.¹⁵
Motors alone are roughly 20% of an arm's bill of materials, bought from the same catalogs by everyone. This is why the design feels fixed: because it has been. For decades the incumbents have assembled the same components around the same architecture, so every arm moves the same way, senses the same way, and hits the same ceilings. What looks like a mature market converging on the optimal design is closer to an industry that stopped asking the question.
Reducers
What they do. Motors spin fast with low torque; arms need slow, high-torque motion. So every joint runs through a precision gearbox at ratios of 50:1 to 160:1. The reducer is the single most expensive component in the arm.
The choices.
The limitations. Two problems, one economic, one physical.
The economic one: the reducer is the single most expensive part of the arm, roughly 35% of the bill of materials, and nearly all of them come from two Japanese companies. Nabtesco holds about 60% of the market for precision reducers in larger joints; Harmonic Drive Systems roughly 70–75% of the strain wave segment.¹⁶ The pairing is so standard in robot design that a disruption at either company would stall robot production worldwide.
The bitter irony: the strain wave gear is an American invention. C. Walton Musser developed it at United Shoe Machinery outside Boston in 1955, and it drove the wheels of the Apollo Lunar Rover. In an echo of Unimation, the US sold this integral innovation, and it was passed from owner to owner until it ended up in Japan.¹⁷
And Japan's grip is no safer than America's was. The reducer market is about $3.8 billion today¹⁸ and China is attacking exactly this layer: Leaderdrive is eroding Harmonic Drive's share with 500,000 strain wave units of annual capacity and a target of 1.59 million by 2027, while Shuanghuan Driveline, which supplies Tesla's Optimus and Unitree, brings on roughly 500,000 units of reducer capacity in 2026.¹⁹
The physical one: gearing is a trade. A high gear ration multiplies the motor's torque, which is what makes the arm strong. But it also multiplies friction, which cuts the joint off from the world: forces at the tool barely register back through the gears, and the joint becomes nearly impossible to back drive.
Tying this back to the connector example: a 100:1 reduction with high friction makes the joint nearly impossible to backdrive. When the plug meets resistance at the wrong angle, a human wrist gives and re-aligns; this joint can only push through. The stiffness that makes the arm strong makes it incapable of yielding, and insertion is a task of yielding.
Controller, firmware, and sensing
What they do. The controller plans motion and closes the loop; the firmware runs each joint; sensing is how the robot knows anything beyond its own position.
The limitations.
Legacy arms are position controlled. The controller receives a complete motion plan, sometimes thirty seconds of trajectory at once, and executes it blind. The robot knows where it is, but its knowledge of the environment is generally crude and infrequent. This is the critical flaw and essentially makes live course correction impossible.
One of the most important sensing gaps is torque. Torque is how an arm reads force and force is the robot's direct experience of the world, the signal both tasks and safety depend on.
Torque, if known at all, is often estimated from motor current, and through a high-ratio, high-friction gearbox the estimate is poor: research puts current-based torque error roughly five times worse than a real torque sensor.²⁰ Some arms use torque sensors at the joint but these are expensive and often limited to the wrist joint.
The entire install base was built for a world where the robot doesn't adapt on the fly. This is the hard ceiling on the industry's AI ambitions: a model can be brilliant and still be trapped behind a 1980s interface. It perceives in milliseconds but can only respond in waypoints. Whatever the software layer promises, the firmware caps fidelity.
Tying this back to the connector example: this is where the insertion dies. The arm can barely feel the contact, and the loop that could act on what it feels accepts corrections a hundred milliseconds late, for a contact that resolves in a few.
Programming
What it does. Programming is how humans get the task into the machine. It decides who can deploy a robot at all, which makes it the clearest window into the industry’s innovation drought.
The choices. Every robot ever sold is programmed through some combination of eight methods:
All eight remain in use today with very little adoption in the last two. It’s worth noting that practitioners split these into online programming, done on the robot itself, with the cell stopped while you teach and offline, written in simulation while production keeps running. Online incurs downtime and offline requires expensive software licenses and highly skilled engineers.
None of these methods are proprietary. Any company can implement any of them. The differentiation is execution, and whether the company has software DNA to make them usable.
It’s useful to think about three waves in robot programming
Wave 1 (1970s to 2000s): built for the integrator. Each incumbent invented a language: KRL, RAPID, KAREL. Written by controls engineers, for controls engineers. The architecture assumed the end user never touches the software. Robots lived in cages and integrators did everything, for a fee, every time anything changed.
Wave 2 (2008 to 2020): easier, but bolted on. Universal Robots invented the cobot and two genuinely better ideas: hand-guiding and flowchart programming. But they were layered onto legacy architecture by a hardware company, and the complexity bleeds through. Consider what a standard UR machine-tending job actually requires, per UR’s own tutorials:
The programming moved from the pendant to the flowchart but the skill level remains high.
The economics follow. Buyers still spend four to five times the robot’s price again on integration: specialist labour, tooling, fixtures.²¹ A $40,000 arm becomes a six-figure project, and any change to the part or process reopens the bill. Not to mention that every time you reprogram a line you need to pause production and incur downtime costs. This is why automation is predominantly found in very high-volume industries like automotive and electronics. Yet those industries are the exception: 74% of America's 239,000 manufacturers have fewer than twenty employees, these are high-mix shops where the part changes faster than the integration bill can be repaid.²²
Wave 3 (2019 onward): AI arrives, mostly as a retrofit. Rather than moving from position to position using simple logic the arm uses AI to decide the best next move in real time based on what it has learned from demonstrations. Where the legacy players are adding AI, it’s generally through partners: Universal Robots launched an imitation-learning system with Scale AI in March 2026.²³ That means two vendors, mismatched hardware and software, and nobody owning the outcome. And there's a harder limit than the org chart: as we saw at the controller, no legacy arm lets an external model drive it in real time. The retrofit is capped at the firmware. As we will see, Standard Bots is a pioneer here and the industry has noticed: at Automate 2026, Standard Bots’ Flux AI won the Automate Innovation Award for Vision, AI and Software.²⁴
Tying this back to the connector example: a deterministic program describes one insertion in advance, one path, one angle, one endpoint. The state of the art in adaptation is an integrator hand-scripting a search pattern, a wiggle written in code for one specific part. The whole difficulty of the task is that no two insertions are the same, and only the last two rows of the table drop the assumption that they are.
Summary
Most arm manufacturers buy off-the-shelf components from the same handful of suppliers and assemble them around a decades-old control architecture. Innovation has stalled, and the industry keeps building arms as systems for the last paradigm.
The prize for cracking Wave 3 is much bigger than the $20 billion arm market. It’s a share of what Standard Bots’ CEO Evan Beard has described as a $25 trillion global labor market.²⁵
To win the next generation, the arm needs to be reimagined. One company has done exactly that.
What makes Standard Bots different?
In short: Standard Bots redesigned the robot arm from scratch, every component optimized for AI, and built it on an American supply chain.
The product: an arm designed from scratch
Problem 2 is a systems problem, so the answer must be a systems answer. You can’t bolt intelligence onto a machine designed for replay; you must redesign the whole machine for its new job. That is precisely Standard Bots’ approach: rebuild the arm end to end, every component optimized for AI.
It started with one design goal: build an arm that AI can control in real time. That goal flipped nearly every component. From the outside it looks like another cobot, yet if you know what you’re looking for the inside is very different.
Nearly every row moves in-house, and every row now serves learning instead of replay. Looking deeper at a few components:
Torque sensing: Everyone else estimates torque from motor current or glues strain gauges to a flexure. Standard Bots uses a unique sensing method that the company reports is more accurate, cheaper, easier to manufacture, and repairable.²⁶ It gives every joint torque sensing without adding an expensive component per axis.
²⁷
Control: Real time feedback at 2kHz. Why does 2kHz matter? Because AI requires real-time course correction: the lower the frequency, the slower the feedback and the more errors that build up. Typical industrial arms aren’t built for this, and this hampers performance significantly.
Tying this back to the connector example, one last time: an operator demonstrates a handful of insertions and the model learns what varies between them, so nothing is fixed at deployment. Joints that measure their own deflection feel the few newtons between clicked and crushed. A new torque command every half millisecond adjusts the angle of attack as the contact unfolds, not a hundred milliseconds after. The plug seats, and this time the robot knows it.
And why can’t incumbents just follow? Because this is joint-level architecture. Matching it means redesigning the actuator, the electronics, the firmware, and the API, then re-certifying the product. That’s not an update. It’s a new robot.
Motors and reducers: The same joints are reused across a modular family. Spark from $29,500, Core from $37,000, Thor from $49,500, and Bolt, a bimanual droid, in beta.²⁸ One arm or two, stationary or mobile, sized to the job. The sizing is deliberate: Thor’s 30kg payload at two metres of reach is enough to palletize a full stack without a seventh-axis lift, which is exactly the heavy, high-value work the fastest-growing payload class is chasing. Customers buy the form factor the task needs, not the one the vendor makes.
Standard Bots publishes a comparison of Core against the FANUC CRX and UR10e.²⁹ The results are clear:
Lower list price, higher payload, and tighter repeatability, per the company’s published specifications, with assembly in America.³⁰
Taught, not programmed
Hand guiding has existed for decades: move the arm, record waypoints, replay them. It’s just hard coding with your hands instead of a pendant. Imitation learning is different in kind. The robot doesn’t store your motion. It learns the task and generalizes to variation you never showed it.
Standard Bots’ goal is to take process designed by ML engineers and make it suitable for anyone regardless of education level.³¹ Here’s how it works:
Step 1: Capture. Demonstrate the task in the method that suits the operator.
Onboard vision records the human’s actions in tandem with the other sensors, collecting structured training data.
Step 2: Review. The operator replays the captured data, labels moments as good or bad, and annotates.
Step 3: Train. The user can train their own models or fine tune an open-source model in the cloud, stored and versioned like software releases.
Step 4: Run. The robot performs the task autonomously on a learning model. It builds its own strategies and self-corrects around changes and obstacles it was never shown.
Step 5: Recover. When it hits something it can’t handle, a human teleoperates in and fixes it. The correction is captured at the exact point of failure and feeds the next training run. Retrain, redeploy, expand.
Below the AI sits a UX designed for ease of use. No-code routines on a touchscreen, pre-built playbooks for machine tending, palletizing, and welding, native integration with PLCs, sensors, and fieldbus equipment, a built-in 3D simulator, fleet dashboards, and a full Python SDK for the expert end.³²
This is critical, AI is not a blunt instrument to be used for every task. The operator can use traditional programming for repeatable operations and then use learned approaches only where they are needed.
At Automate Evan announced the future: an AI sits above all of this and streamlines how you program the arm: describe the task in text, and the robot generates its own step-by-step program.³³
What it adds up to: the company’s claim is that a factory worker becomes a robot operator in a day.³⁴ Some examples: Shaw Barrels, a multi-generation family manufacturer in Pennsylvania, reports a 4x productivity increase through automating a deep-hole drilling machine.
HomeGrown Lifting in Kentucky reports 2x throughput increase on palletizing months after pivoting into a new product line. Neither used an integrator.³⁵
Teach-in-a-day is what could collapse the four-to-five-times integration multiplier. The robot was never the expensive part. The expertise was. Standard Bots seeks to remove that expertise.
Standard’s AI philosophy
Beard laid out their philosophy in a recent essay with Not Boring, worth reading in full.³⁶ The short version: physical AI won’t be built in the lab. It gets built in the real world, by earning your place with value today.
Robotics is bottlenecked on data, and the data that matters is on-robot, task-specific, and from real deployments. Video helps pretraining. Simulation helps multiply edge cases. But for the final policy there is no substitute for data collected on your own hardware doing real jobs; data misaligned with your hardware can cost 100 to 1,000 times more volume.³⁷
Standard Bots spent eight years doing the hard, unglamorous work, vertically integrating the machine, earning its place on factory floors and used it to pioneer the AI on top. It's the rare position from which delivering customer value and advancing physical AI are the same act. Evan's stated goal of accounting for 10% of all new US industrial robot deployments within a year would scale both at once.
It’s important to state that this is a platform, not a bet against the labs. If frontier models crack general manipulation, Standard Bots owns the hardware, the customers, and the deployment layer, and can onboard better external models.
A US-first supply chain
The rebuilt product answers Problem 2. The rebuilt supply chain answers Problem 1.
Supply chain: motors, drives, and PCBs are built in Glen Cove, New York. In Beard’s words, the company designs everything except bearings, gearing and chips, with a goal of manufacturing “from metal in robots out” in America by 2027.³⁸ Given what we saw above about the component chokepoints inside every robot joint, owning that layer isn’t patriotism, it’s insurance against supply disruption, and it strips tariff and freight exposure out of the cost structure that import-dependent arms carry.
Another way to put it is that the factory itself is the product. Standard Bots has spent years building the production process and written its own software across supply chain, quality, and assembly. Designing a great robot is one problem; ramping its production is another, and the formula for the second is as proprietary as the first.
The founders
We’ve spent this piece on product and technology, but a company lives and dies by its founders. We believe Evan Beard, David Golden and James Cordle are uniquely built to grow and scale this one.
Evan Beard’s first three companies were software. Etacts, a contact manager he co-founded out of Duke in 2009 with Howie Liu, who later founded Airtable, went through Y Combinator and sold to Salesforce within a year of launch.³⁹ ArmorHub, a web security scanner, sold to Spirent in 2014.⁴⁰ A Plus, a media company co-founded in Ashton Kutcher’s living room, reached 50 million monthly visitors within two years.⁴¹
That was just a warmup, in 2017 Beard started building robots from his New York City apartment, funding it with personal savings. Before committing, he has described interviewing around a hundred companies that used robots from the incumbents. He heard the same complaints on repeat: they automate only a fraction of what we want, the interfaces need experts, the deployment risk is ours, and none of them are made in America. Read that list again. It’s the two problems of this piece. The company is the answer to those hundreds of calls.
The early years were lean. Seed funding closed just before the COVID lockdown and the company nearly ran out of money; Beard moved his own savings into the account to keep the doors open while the team debugged electronics.⁴² It moved into a Glen Cove factory, raised a $63 million Series B from General Catalyst, Amazon’s Industrial Innovation Fund, and Samsung Next in 2024,⁴³ and spent eight years doing what software founders are told not to do: building the whole machine.
Co-founder David Golden came from the other direction: he co-founded Bowery Farming, one of the most ambitious vertical-integration plays in agriculture technology, and built and sold Leappay before that.⁴⁴ Beard brings the software DNA and the go-to-market. Golden brings scaling physical operations.
Today the company has expanded far beyond the founders. One leader to mention is President Zach Tomkinson. Before joining Standard Bots he was a sales leader at Universal Robots, the company that defined Wave 2. There he learnt firsthand that outsourcing sales and integration meant outsourcing your relationship with the end user. As a result, at Standard Bots he has built a team of sellers and applications engineers more than 50 deep. Distribution in robotics is as hard to build as the robot, and it compounds with every deployment. We believe this go-to-market approach is a moat in itself.
That pairing explains the company. Standard Bots sells robots the way a software company sells products: direct, inbound, self-serve, no integrator channel. The DNA is the strategy.
Our investment
In June 2026 we led Standard Bots’ $200 million Series C at a $1 billion valuation.⁴⁵
America needed an arm champion. The arm needed rebuilding. One company was built for both.
Arms, AI, and supply chains don’t work as layers. They work as a system. By vertically integrating and designing from first principles, Standard Bots can build the best product for the customer. The data flywheel, the sovereign supply chain, and the platform position all fall out of that one decision.
And the arm is the wedge, not the endpoint: the same modular joints extend into mobile bases and new form factors, one hardware platform configurable to the job. The prize is physical AI and we believe Standard Bots is building the position to lead it.
Standard Bots has stated a goal of supplying 10% of new US industrial robot deployments within a year.⁴⁶ We believe even that undersells it. Manufacturing is entering a phase change, from programmed machines to physical AI, and the arm will be one of its foundational components.
The industrial robot was invented in Ewing, New Jersey, in 1961. Sixty miles up the coast in Glen Cove, New York, we believe it's being built back.
Risks
We’re clear-eyed. Scaling hardware manufacturing is unforgiving, and the plan demands execution across manufacturing, deployment, and software at once.
The incumbents are responding: UR’s imitation-learning launch with Scale AI shows the giants see the same future.⁴⁷ China is the bigger competitive question: price competition is intensifying, Chinese reducer and robot capacity is scaling, and Chinese AI-native entrants could pair low cost with capable models. The proposed US restrictions are proposed, not law.
Reshoring itself is a risk. If the wave stalls or policy support fades, a tailwind we’re underwriting becomes a headwind.
If foundation models commoditize manipulation faster than expected, hardware differentiation could erode, though the platform position hedges this.
Important Disclosures
Standard Bots is one of a number of portfolio investments held by RoboStrategy, Inc. This discussion of a single portfolio company should not be viewed as representative of the Fund’s portfolio as a whole. Information regarding the Fund’s full portfolio is available in the Fund’s filings with the SEC.
RoboStrategy, Inc. holds an investment in Standard Bots, and FP Strategies LLC, the Fund’s investment adviser, therefore has a financial interest in, and an incentive to present favorably, the company discussed herein. The fair value of the Fund’s investment in Standard Bots is based in part on the financing round described in this article. Nothing herein constitutes investment advice or a recommendation to purchase or sell any security.
This article contains forward-looking statements within the meaning of Section 27A of the Securities Act of 1933 and Section 21E of the Securities Exchange Act of 1934. Forward-looking statements are not historical facts but are based on current expectations, estimates, projections, beliefs, and assumptions about the Fund, our current and prospective portfolio investments, our industry, and our beliefs and assumptions. Words such as “anticipates,” “expects,” “intends,” “plans,” “will,” “may,” “continue,” “believes,” “seeks,” “estimates,” “would,” “could,” “should,” “targets,” “projects,” and variations of these words and similar expressions are intended to identify forward-looking statements. These statements are not guarantees of future performance and are subject to risks, uncertainties, and other factors, some of which are beyond our control and are difficult to predict, that could cause actual results to differ materially from those expressed or forecasted in the forward-looking statements. The Fund undertakes no obligation to update or revise any forward-looking statements, whether as a result of new information, future events, or otherwise.
Investing in shares of RoboStrategy, Inc. involves a high degree of risk and is highly speculative. You should read the discussion of the material risks of investing in our common stock in the “Risk Factors” section of the Prospectus. You may lose part or all of your investment.
This article is not an offer to sell or a solicitation of an offer to buy any securities. Any offering of securities may be made only by means of a prospectus meeting the requirements of Section 10 of the Securities Act of 1933, as amended.
Shares of closed-end investment companies frequently trade at a discount to their net asset value (“NAV”). If shares of our common stock trade at a discount to our NAV, investors will face increased risk of loss. There is no assurance that an active trading market will develop or be sustained, or that shares will trade at or above NAV.
The Fund’s investments in private companies are valued at fair value as determined pursuant to procedures approved by the Fund’s Board of Directors. Fair values are necessarily subjective, and there is no assurance that such values will be realized.
Views expressed herein are those of FP Strategies LLC, the Fund’s investment adviser, and do not necessarily represent the views of RoboStrategy, Inc. or its Board of Directors.
Certain information contained herein has been obtained from published sources prepared by third parties, including statements made by Standard Bots and its officers, and has not been independently verified. While such information is believed to be reliable, we assume no responsibility for its accuracy or completeness.
Sources
¹ FR preliminary 2025 figures, presented at Automate 2026: Manufacturing Dive, June 2026, https://www.manufacturingdive.com/news/us-robotics-rebounded-2025-on-track-more-growth-ifr-automate-2026/823874/; operational stock per IFR World Robotics 2025, https://ifr.org/ifr-press-releases/news/global-robot-demand-in-factories-doubles-over-10-years. Third-party market data may be revised and definitions vary by source.
² Straits Research, Industrial Robots Market Report, https://straitsresearch.com/report/industrial-robots-market, 2024 estimate of $20.8 billion in annual revenue.
³ Morgan Stanley Research, "Humanoid Robot Market Expected to Reach $5 Trillion by 2050," April 2025: approximately one billion humanoid units in use by 2050, with 63 million in the US addressing roughly $3 trillion in payroll, modeled from occupation-level labor data across 831 US job classifications; see also Morgan Stanley, "The Humanoid Robot Market Outlook," 2024. Citi GPS, "The Rise of AI Robots: Physical AI is Coming for You," December 2024 projects a market of up to $7 trillion by 2050. These analyses model humanoid robots; we reference them as estimates of automatable physical labor. Our stationary-work estimate is based on classifying major occupation groups in the BLS Occupational Employment and Wage Statistics survey by workstation mobility; production occupations, which are predominantly fixed-station work, account for the majority of the total, with smaller contributions from material moving and food production roles. Long-range projections of this kind are necessarily speculative.
⁴ IFR robot density statistics, 2024 data.
⁵ Research and Markets, “Physical AI Market Set to Surpass $430 Billion by 2030,” July 2026. Third-party forecasts for the physical AI market vary widely by scope and methodology.
⁶ Association for Advancing Automation, “Joseph Engelberger and Unimate”. George Devol filed the first industrial robot patent in 1954.
⁷ Unimation, Wikipedia. Westinghouse acquired Unimation for $107 million in 1983 and sold it to Staubli in 1988 as the industry shifted from hydraulic to electric robots.
⁸ Chris Miller, “Robotics Manufacturing: The Rise of Japan”, covering the GM-FANUC joint venture and the 1983 US government assessment of domestic robot competitiveness.
⁹ Unimation, Wikipedia. Westinghouse acquired Unimation for $107 million in 1983 and sold it to Staubli in 1988 as the industry shifted from hydraulic to electric robots.
¹⁰ Chris Miller, “Robotics Manufacturing: The Rise of Japan”, covering the GM-FANUC joint venture and the 1983 US government assessment of domestic robot competitiveness.
¹¹ KUKA AG; Midea Group acquired a controlling stake in 2016-17.
¹² ABB Group press release, October 2025.
¹³ Digitimes, August 2024.
¹⁴ House Select Committee on the CCP, GUARD Act press release; FDD analysis, March 2026. Both bills are proposed legislation.
¹⁵ Kollmorgen, TBM2G robot-ready frameless servo motors; joint motor selection factors per Kollmorgen and industry guides on torque density and cogging in robotic joints.
¹⁶ Nabtesco share per Nabtesco Corporation; Harmonic Drive strain wave segment share per Global Info Research, 2025, which puts it at 71.37% of global harmonic reducer revenue.
¹⁷ Musser's invention per Harmonic Drive Systems; Lunar Rover per Harmonic Drive SE; sale to Japan's Teijin Seiki in 1991 per company profile.
¹⁸ Dataintelo, Industrial Robot Reducer Market Report.
¹⁹ Leaderdrive capacity per company profile; Shuanghuan per Next Financial, "The Joint Problem," 2026.
²⁰ Joint torque sensing vs motor current estimation accuracy per Measurement (ScienceDirect), 2022 and related literature.
²¹ Daniella Tola, “Enabling Digitalization in Modular Robotic Systems Integration,” PhD dissertation, Aarhus University, 2024 (arXiv:2401.02227), which notes integration can run four to five times the cost of the hardware; see also Standard Bots, “How much do robots cost,” 2026.
²³ Universal Robots and Scale AI press release, March 2026.
²⁴ Association for Advancing Automation (A3), Automate 2026 press release, Business Wire, July 2026. The Automate Innovation Awards recognize new products introduced to the market in 2025.
²⁵ Evan Beard and Packy McCormick, “Many Small Steps for Robots, One Giant Leap for Mankind,” Not Boring, January 2026.
²⁶ Evan Beard and Packy McCormick, “Many Small Steps for Robots, One Giant Leap for Mankind,” Not Boring, January 2026.
²⁷ Torque control and joint performance specifications per Standard Bots, accessed July 2026, including ±0.1 Nm torque repeatability. Company-reported figures; not independently verified.
²⁸ Standard Bots published pricing, accessed July 2026. Starting prices, subject to change.
²⁹ Standard Bots Core product page and published comparison, accessed July 2026; FANUC CRX specifications; UR10e specifications. Competitor prices vary by configuration.
³⁰ Standard Bots Core product page and published comparison, accessed July 2026; FANUC CRX-10iA/L specifications; UR10e specifications, Universal Robots e-Series brochure.
³¹ Standard Bots AI platform, accessed July 2026.
³² Standard Bots software, accessed July 2026.
³³ Standard Bots AI platform, accessed July 2026.
³⁴ Standard Bots software, accessed July 2026.
³⁵ Standard Bots published case studies, accessed July 2026. These results have not been independently verified, and individual customer results are not representative of all deployments.
³⁶ Evan Beard and Packy McCormick, “Many Small Steps for Robots, One Giant Leap for Mankind,” Not Boring, January 2026.
³⁷ Evan Beard and Packy McCormick, “Many Small Steps for Robots, One Giant Leap for Mankind,” Not Boring, January 2026.
³⁸ Standard Bots company website, accessed July 2026. The 2027 goal is a company target.
³⁹ TechCrunch, December 2010; Y Combinator company directory.
⁴⁰ Spirent press release, 2014.
⁴¹ Inc./Business Insider, 2015; Forbes 30 Under 30, 2016.
⁴² New York Post, “Standard Bots is training robots to think for themselves,” October 2024.
⁴³ Standard Bots Series B announcement, July 2024.
⁴⁴ Company materials and public profiles.
⁴⁵ Standard Bots press release, PR Newswire, June 2026.
⁴⁶ Standard Bots press release, PR Newswire, June 2026.
⁴⁷ Universal Robots and Scale AI press release, March 2026.
