Agility Robotics拟通过Churchill Capital Corp XI上市，成为美股稀缺的纯 humanoid robotics 标的

Humanoid robotics is one of the most talked-about themes in the market right now.

X has been going crazy with robotics over the last few days.

Tesla’s Optimus, Figure, Boston Dynamics, Unitree. The videos are everywhere, and the theme is front-of-mind for almost all investors.

Owning it directly is another matter.

For U.S. public-market investors, the exposure has been almost entirely indirect. You can buy a large EV and energy company and treat its robotics program as embedded optionality.

You can buy broad automation names that touch the theme from a distance.

But a U.S.-listed, pure-play humanoid company (one whose robots are already working in real facilities rather than running lab demos) simply hasn’t been available to buy.

That is about to change, and it raises a question worth taking seriously: what is a company like that actually worth, and how would you even buy it?

A few days ago, a U.S. humanoid robotics company entered into a definitive agreement to go public through a SPAC.

What makes it worth a closer look is the operating history behind it.

-

The robots have already moved more than 100,000 totes inside a live logistics facility.

-

They’re deployed alongside several logistics and automotive names you’d recognize immediately.

-

The company has disclosed a commitment worth more than $300 million, tied to 1,000 robots under a multi-year contract.

By their own account, no other U.S. pure-play is doing quite what they’re doing, and the product has made it out of the lab and onto the floor.

There’s also a timing element. The deal hasn’t closed and the ticker hasn’t changed over yet, which leaves a window to understand the business before the structure is finalized.

This report works through the entire company from the ground up. The robot, the software, the customers, the manufacturing, and the unit economics, and then through the deal mechanics that determine what public investors are actually buying.

The aim is to give you enough to form your own view of what it’s worth today and what it could be worth in five years.

Here’s what the full report covers:

-

Why the humanoid theme is hard to own through U.S. public markets

-

The operating company, the SPAC, and the post-close ticker and how the structure works

-

The customer problem the thesis starts from

-

The beachhead market. Where they’re starting, and why not the home

-

What the company sells (it’s more than a robot)

-

Why a humanoid form factor at all?

-

The hardware generations and why the newest version matters

-

Safety as the gatekeeper between a demo and a real business

-

The “physical AI” data loop and what it could mean for a head start

-

The software layer and the question of customer lock-in

-

Who is already running these robots in production

-

The $300M+ order. What it is, and what it isn’t

-

Can they build at scale?

-

Unit economics. Can one robot become a real business?

-

What the upcoming filing still needs to answer

-

The deal structure. Valuation, the PIPE, dilution, and lockups

-

Competition

-

Closing Thoughts & My Strategy

You do not want to miss this.

If you read my deep dive from 2 days ago, I would say this one is on par.

Just for a company you probably have never heard of.

*This report is for informational and educational purposes only. It reflects my personal research, interpretation, and opinions based on publicly available materials, company presentations, management commentary, SEC filings, customer announcements, and third-party sources believed to be reliable. It should never be treated as personalized investment advice, a recommendation to buy or sell any security, or a solicitation to enter any transaction.*

## Humanoids Are Coming to Public Markets

Humanoid robotics have been easy to see and hard to own directly.

The theme is all around and very front-of-mind for investors. We have Tesla with Optimus, Figure, Apprtronik, Boston Dynamics and Unitree.


But for U.S. public-market investing, that exposure is primarily indirect.

You CAN buy Tesla Optimus optionality, but obviously it’s inside a much larger EV/energy/Ai company.

You can buy automation names like Symbotic, Rockwell, Teradyne etc. But those vehicles are broader automation exposure and not direct versions of the humanoid thesis.

This is why the Agility/CCXI transaction is so interesting, and why I am introducing this today way before I get to my Humanoid layer in my investment research.


So let’s unpack all of this.

Using that quote above from their website, this explains the public market angle explicitly. No other public, pure play company in the U.S. is doing what they are doing. And they have made it out the lab.

The official merger release says Agility entered a definitive business combination with Churchill Capital Corp XI on June 24, 2026. After closing, the combined company is expected to operate as Agility and trade under ticker AGLT. The deal values Agility at a $2.5B pre-money equity value and is expected to provide more than $620M of gross proceeds, including roughly $420M of SPAC trust cash assuming zero redemptions and about $200M from a PIPE led by Foxconn.

Let’s seperate three things.

1 - Agility Robotics is the operating company. It builds Digit, Arc, safety systems, physical AI, and RoboFab. I will break all of these down throughout this report.

2 - CCXI is the current public SPAC vehicle

3 - AGLT is the expected post-close ticker after the merger.

We will unpack what all of this means throughout the report. If you are unfamiliar with how SPACs work, this could be confusing. But by the end, you will have a good grasp.

There is something here that is hard to put a number on. But it is the concept of **scarcity value.**


Like I mentioned, most humanoid companies are private, foreign-listed, or embedded inside larger companies.

We will need to keep this in mind when analyzing the financials and deciding what this company is worth today and what it could be worth in 5 years.

## The Overview

Agility is trying to become a humanoid labor platform for industrial work.
*

I thought this video was really cool. Give it a watch!

Agility is broader than just making a robot and narrower than general purpose humanoids everywhere.

They are starting with real physical labor inside manufacturing, distribution, and logistics environments.


I think that the strongest way to understand this company is through the customer problem.

Factories, warehouses, and logistics facilities need physical tasks performed all day. Moving bins, parts, trays. Loading and unloading conveyors. Feeding machines. Etc.

That can obviously be very repetitive, physical and have a high risk of injury.

Peggy Johnson (CEO) talks directly about that in the business combination call and said that Digit (one of their robots) picks up dirty automotive parts, paces them in a washer, unloads them, moves them into a dryer, and stacks them repeatedly.


So the thesis begins with customers having a labor problem and Agility trying to solve them with a humanoid worker.

**Material Handling is the beachhead**

This is the entry point for Agility. This means moving physical objects through a facility.

They say that their first markets are manufacturing, distribution. and logistics.


This is because they have large labor pools, significant labor spend, persistent hiring needs, physical repetitive work, structure workflows, and existing infrastructure designed around people.

This is one reason why this market has to come before the home robot dream.

Homes are very chaotic. All the objects are different, all the rooms are different for the most part.

Factories / warehouses are complex in their own way. But what they have is repeated workflows, defined processes, managers who track ROI, and budgets for automation. That gives Digit a narrower job to master first.

**Physical Labor Requires an Integrated System**

A human-centric form factor, embodied intelligence, and deployment systems in one unified platform.


That gives us their stack:

-

**Digit** is the physical interface.

-

**Embodied AI / physical AI** is the skill, movement, and control layer.

-

**Arc** is the deployment and fleet orchestration layer.

-

**Service and support** keep the system operating in customer facilities.


Awesome. Very easy to understand and very easy to visualize.

This is a great bridge into the next section.

## What Does Agility Sell?

The overview is a ‘deployable humanoid labor system’.

But this product is a combination of all the different layers.


A customer does not only need a robot that can move. They need a robot that can be deployed, assigned tasks, integrate into workflows, be monitored, serviced, updated, and expanded across a facility.

This is what they are trying to package.

**1 - Digit: The Physical Worker**

This is the physical body.


Fully autonomous humanoid with commercial deployments, designed to connect ‘islands of automation’ inside facilities and address hard-to-fill labor gaps in spaces where people already work.

Let’s take a second on that islands of automation statement.

A modern warehouse or factory often has multiple automated systems that each do one thing well.

The gap is between those systems.

Someone still has to pick up the object and move it from one process to the next.

That ‘in-between’ labor is where Agility is having Digit play.

So in a way it’s job is best understood as ‘mobile manipulation’.

It can move through the facility and interact with objects. Much different than a stationary robot arm which manipulates objects but stays fixed.

**2 - ARC: the control layer**

Arc is the software that makes Digit useable in a facility.


This is really where they start to look like infrastructure and more than just hardware.

If a customer is going to run multiple digits across a facility, it needs orchestration. Which robot get’s which task? Which robot needs charging? Which robot is down? You get it.

This is Arc’s job.

Jonathan Hurst (Co-founder & Chief Robotics Officer) in their call said that Arc connects Digit to customer operations, gathers data from the robot, coordinates hundreds or thousands of robots, interacts with AMRs (autonomous mobile robots) and robot arms, engages warehouse management systems, and sends robot health data back to Agility’s operations center.

So the customer buys the entire system because the robot has to plug cleanly into the operation.

**3 - Service and Support**

Industrial robotics lives or dies on deployment quality.


On-site service, online support, real-time monitoring, white-glove delivery, hands-on robot training, software updates, rapid response, workflow optimization, and help expanding to new workflows and sites.

The customer needs the robot to work in a production environment. If Digit goes down, gets confused, needs a part, needs a software update, or has to be adapted to a workflow change, the customer needs support. This is especially true for early humanoid deployments, where customers may lack internal robotics expertise.

This is also why RaaS (robotics as a service) fits the early adoption curve. Under RaaS, the vendor usually handles maintenance, hardware repairs, updates, and a larger share of technical responsibility. Agility’s FAQ frames RaaS as lowering the financial barrier, speeding deployment, providing end-to-end support, and allowing customers to scale more easily.

The deeper economics come later. For this section, the key idea is that Agility sells ongoing operational enablement, rather than a one-time machine drop-off.

**4 - Technology licensing**

This is early, but worth understanding.

Agility says the technologies developed for Digit may create value beyond its own humanoid deployments. Real-world deployment data as potential ecosystem-extension assets.

Peggy also talks about this on the call and says that Agility is expanding beyond humanoids through licensing actuator technology and fleet intelligence software into the broader robotics industry.

This creates a second possible path over time.

Path one is direct Digit deployments.

Path two is licensing pieces of the stack to others: actuators, control systems, safety systems, Arc, or fleet intelligence software.

The near-term thesis still centers on Digit deployments. Licensing is more of an upside extension. It shows Agility sees its internal technology stack as valuable outside its own robot fleet.

## Why a Humanoid Form Factor?

This is something that I wonder whenever I see companies like this.

Why make it look like a human? Is that really the best build to accomplish what needs to be accomplished?

Because many of the deployed systems are more like forklifts and robotic arms.

The answer that Agility provides is that they are building for **human infrastructure**. I thought that was interesting. So let’s unpack that a bit.


Factories and warehouses were designed around people. Aisles, doorways, shelf heights, conveyors, machines. You name it. They all assume a worker-sized body is moving through the space.

So Agility’s bet is that a humanoid can enter the environment with less facility redesign than traditional fixed automation.

Jonathan Hurst also said that Digit was NOT designed to imitate people. It ended up a humanoid because Agility started from the job and worked backwards. He says the robot needed to go where people go, perform workflows designed around people, move through narrow spaces, lift objects up high, operate in human workspaces, and interact physically with the world.

And their presentation backed this up.


Here is a cool video explaining some of the design behind Digit
*

So humanoids could look gimmicky but their argument is that Digit’s shape is an industrial design choice that is optimal for the deployed use cases.

I don’t want to spend too much time on this, but let me break down some of the different body parts and explain why they use them vs. other options.

**Legs**

Wheels are excellent on flat, controlled surfaces. AMRs are already useful because they move goods horizontally across warehouse floors. But wheels have limits. Jonathan’s explanation is that the robot needs to fit through human spaces while still lifting heavy things up high. A wide, statically stable wheeled base would become too large or too tippy. So Digit uses dynamic balance. It is actively balancing all the time, more like a person or animal.

But the tradeoff is complexity. Legs make the robot harder to build, control, certify, and service. So the investment question later becomes whether the adde flexibility is worth the added complexity.

**Upright Torso**

Keeps the robot narrow enough to move through human infrastructure. Places the compute, batteries, sensors, and arms in useful positions. Helps the robot balance because the torso can shift and lean as part of whole-body control. Also gives the robot vertical reach.

**Two Arms**

A single robotic arm can be great for smaller or highly repeatable objects. But when the object is large, heavy, flexible, or off-center, two arms help.

Jonathan says two arms allow Digit to lift larger objects, reposition objects between hands, simplify manipulation, help with whole-body balance, help catch itself during a fall, and help get back up.

## Digit v4 Versus Digit v5

V4 was the deployment proof and V5 is the scale attempt.


Digit v4 gives Agility credibility today. You can see the stats above.

There’s also 360 degree sensing, adaptive dexterity, learned whole-body control, and OSHA-recognized safety systems. It is currently handling parts in a real facility today.

But the deployment envelop is still limited. The main constraint is the workcell.

This means the robot can work, but the facility has to define where humans and robots interact. I guess it’s like a jailcell? That’s probably where the phrase comes from. Correct me if I am wrong!

But this reduces the range of tasks that Digit can cover and limits how freely it can move across the facility.

**Enter v5.**

Digit v5 is the version designed for commercial scale.

Again, you can check out all the stats above. This is a capability expansion.

The 35 to 50lb payload jump is significant. it opens up a much larger share of industrial materials.

The extra range can also reduce ‘exceptions’.

In robotics, exceptions kill productivity. If a robot can handle 70% of objects in a workflow, the human still has to babysit the remaining 30%. It the robot can handle 90%+, the economics look much different.

**The jump from 2:1 charge ratio to a 10:1 charge ratio is about utilization.**

A robot in these settings has to be available when the line or workflow needs it. Downtime for charging reduces output, complicates scheduling, and can require extra robots to cover the same workload.

And the max daily battery output rising from ~16 hours to ~22 hours in a 24-hour period means you can have multiple shifts.

Think about RaaS economics again. A robot that works more hours per day generates more valued for the customer and can support better payback for Agility.

**Vertical reach **is pretty self explanatory. I won’t spend time telling you why that’s better.

But **changeable end effectors **is not as common sense.

End effectors are the “hands” or tools at the end of the arms.

v4 has task-specific of fixed end effectors. This is logical for early deployments as you design the tool around a specific job, such as handling totes or baskets.

v5 is supposed to move towards changeable or dynamic end effectors. this expands the future skill set. A robot that can only use one type of gripper has a narrower job, but a robot that can change tools can support more workflows. So this opens up broader facility labor.

Jen Hunter (CFO) says v5 has already been built and they have produced units. Early production experience is validating the cost-reduction roadmap.

So obviously this is one main thing to track.

## Safety Is the Gatekeeper

This is the bridge between early deployments and scaled facility labor.

Can these machines operate around people, inside a real facility, with a risk profile that customers can accept?

Not only are products, equipment, and infrastructure at risk, but humans are at risk if a humanoid malfunctions.

These robots need to perceive people, predict motion, control force, stop safely, and behave in a way nearby workers can understand.

Peggy talked about this in the youtube video I linked earlier. When an LLM gives a wrong answer, the consequences are usually contained; when AI controls a powerful moving humanoid, people can get hurt.

The current state is the workcell we just discussed. But this is too limiting if a customer wants broad facility coverage.

The unlock is **cooperative safety.**

This is a big claim for v5. Cooperative safety means Digit can operate near people in shared industrial spaces without needing a traditional safety barrier around every task.

They wrote a blog on it here if you would like to read it:

[https://www.agilityrobotics.com/content/built-for-the-real-world](https://www.agilityrobotics.com/content/built-for-the-real-world)

But this is the unlock for scaled humanoid adoption.

It requires three main things:

1 - Detection (understanding people and surroundings)

2 - Reaction (responding safely in real time)

3 - Anticipation (moving in predictable ways)

Now we need to understand the difference between **cooperative** and **collaborative.**


There are stages here.

Cooperative = designed to operate outside a workcell

Collaborative = share the same workspace and jointly handle tasks with humans

They are focused on cooperative safety first.

But as they progress along the safety spectrum, you get expanding use cases.


General public, elder care, home service, hospitality etc.

Hopefully you can see the vision.

Their near term market is industrial rather than consumer. Which I think is really smart. But if they can succeed in this progression, we are looking at a massive massive TAM.

Safety is connected to:

**Perception:** detect humans accurately.
**Classification:** distinguish people from non-people.
**Motion planning:** choose safe movement paths.
**Payload handling:** place the object down safely.
**Whole-body control:** stop, lower, balance, sit, and power down.
**Functional safety:** behave correctly even when systems fail.
**Certification:** prove the safety case to third parties.

And they are leaning on Nvidia for part of this architecture.


Not only are they partnered with NVIDIA for their AI Infra, but they also invested in them. They have a pretty solid track record of investments.

There is also a validation and tooling angle here. NVIDIA gives Agility a credible physical AI / edge compute / safety ecosystem partner. Agility still has to prove the full robot-level safety system.

The risks are pretty clear. They call out product safety and liability in their own presentation.


Design flaws, errors, defects, or unintended AI-driven behaviors (yikes) could lead to recalls, injuries, product liability claims, litigation expense, insurance issues, and brand damage.

## Physical AI

This is where Agility becomes more than a warehouse story.

A humanoid robot needs AI that can act through a body.

Digit has to turn all of it’s understanding that it takes in from it’s environment via it’s sensors and turn it into movement.

This is what Agility means by physical AI or embodied AI.

They also draw a difference between Semantic and Physical AI.


Semantic AI is the ‘understanding’ layer. It helps the robot interpret the environment. It can recognize a tote, identify a conveyor, understand a work instruction, reason about a sequence, or know that an object needs to move from one place to another.

Physical AI is the ‘execution’ layer.


How does this exact robot, with this body, these motors, this payload, actually perform the movement?

That is the harder part for humanoids.

**Foundation models are useful but incomplete**

Large language models and vision-language-action models can help with generality. They can help robots understand scenes, commands, and task context.

But they do very little by themselves if the robot cannot execute.

Agility says it starts with physical capability first.


Agility’s sequence is physics → physical AI → deployment → scale. The success metric is useful labor in customer environments.

**Why practice and simulation are central**

Hurst says Agility can provide examples through teleoperation, motion capture, or animation. But then the robot has to practice. That practice happens through reinforcement learning and simulation, where the robot refines the skill until it works for Digit’s own body.

This is similar to how a human might learn. Watching someone lift a box helps. Actually lifting boxes over and over teaches your body how to adjust grip, posture, balance, and timing.

Digit needs that same kind of embodied practice.

Simulation is useful because it lets the robot practice many variations cheaply and quickly. Real deployment is useful because the physical world always has details that simulation misses.

So the learning loop is:

Demonstration gives the rough task.
Simulation gives repeated practice.
Real-world deployment exposes edge cases.
Deployment data improves the next model.

**Why deployment data is valuable**

Each deployment generates movement and workflow data, which improves embodied AI, expands autonomy and dexterity, opens new applications, and supports more deployments.


Peggy says the same thing in the call: real operational data is currency for building foundation models for robots. More deployment data improves the AI model, which accelerates skills, which opens new verticals.

This is the bull-case logic behind Agility’s head start. If Agility has more real-world humanoid deployment data than competitors, and if that data actually improves performance, then every deployment can make the system more useful.

## Arc - The Software Layer

Arc is the system that makes a fleet of Digits usable inside a real operation.


A single robot doing one task can be impressive. But a customer running robots across a factory or warehouse needs something else entirely. They need task assignment, workflow integration, fleet monitoring, uptime visibility, charging coordination, exception handling, and data collection.

That is Arc!

A warehouse or factory is already a software-driven environment.

There are systems that manage orders, inventory, machines, conveyors, AMRs, labor schedules, workstations, production lines, and fulfillment flows. A robot has to plug into that environment. It cannot live as a disconnected machine doing random tasks.

This is why Arc is important. It turns Digit from a standalone robot into a deployable workforce layer.

Hurst explains this in the business combination call when he says Arc connects Digit to customer operations, collects robot data, coordinates fleets, interacts with AMRs and robot arms, engages warehouse management systems, and sends robot health data back to Agility’s operations center.

**This changes the business model**

A robot sale is a hardware transaction. But software and orchestration can become recurring, high margin sales.

Thi sis part of their scaling model. They list Arc adoption, service attachment, and installed-base utilization as KPIs (key performance indicators) for software and post-deployment services. It’s also tied to potential ecosystem-extension revenue.

So if Digit deployments grow, Arc could grow with them.

More robots means more need for coordination.

More facilities means more need for repeatable deployment templates.

And so on.

That is why Arc can eventually support a higher-quality revenue mix. The robot is the expensive physical asset. Arc is the layer that could generate recurring software value and deepen customer lock-in.

**Arc as a potential standalone product**

Hurst says Arc could become useful even beyond Digit because it engages with warehouse management systems and other forms of automation. That is an interesting upside angle.

If Arc can coordinate humanoids, AMRs, robot arms, and other automation systems, it could become more than internal Digit software. It could become an orchestration layer for broader physical automation.

This is still early for sure. But it is nice to see the vision.

**Arc can create customer lock-in**

If Arc becomes deeply embedded in a customer facility, switching costs rise.

A customer may start with one workflow. Then Arc learns that workflow. The customer adds more robots. More tasks get mapped. More facilities copy the same deployment pattern. More uptime and productivity data flows into the system.

Over time, the customer has more than robots on the floor. They have operating history, workflows, integrations, training, service processes, performance metrics, and deployment templates.

That can make Arc a sticky layer.

## Customer Proof

Think of this in four stages.

**Stage 1: Demo**
The robot performs a task in a controlled environment.

**Stage 2: Pilot**
The robot enters a customer facility and tests a narrow use case.

**Stage 3: Production workflow**
The robot performs repeated work in a live operation.

**Stage 4: Scaled fleet deployment**
The customer expands robots across more workflows, more sites, and more facilities.

Agility is trying to show that Digit has crossed into Stage 3 with customers like GXO and Schaeffler, while Digit v5 is supposed to push the company toward Stage 4.


The investor presentation says Agility has 9 committed customer facility deployments, 65,000 hours of operations, and customers including Schaeffler, Toyota Motor Manufacturing Canada, Amazon, GXO, and Mercado Libre. It also shows recent wins across logistics, manufacturing, automotive, and consumer brands.


Agility says Digit moved more than **100,000 totes** at GXO’s Flowery Branch facility. That is useful because tote movement is repetitive, measurable, and throughput-driven. A robot can look good once but moving totes thousands of times forces the system to deal with repeated grasping, placement, walking, object variation, and workflow timing.

**GXO** validates the “in-between labor” thesis. Digit is moving material through the gaps between other automation systems. Agility’s own writeup says Digit evolved from picking items on and off an AMR to a conveyor into stacking totes onto a different floor location. That tells you the robot is being tested across more than one motion pattern inside the same broad logistics environment.

**Schaeffler **gives manufacturing validation. This was Peggy’s example in the transaction call. Digit picks up dirty automotive parts, places them into a washer, unloads them, moves them into a dryer, and stacks them repeatedly. That is a manufacturing workflow, with process steps, dirty parts, machines, and repetition. The combination of GXO and Schaeffler shows how Digit can play is differnet verticals.

**Toyota Motor Manufacturing** Canada is important because the agreement followed a pilot and moved into a Robots-as-a-Service agreement. Agility’s release says TMMC plans to deploy Digit to support employees with manufacturing, supply chain, and logistics operations. It also says the companies will evaluate additional use cases where robots and AI could augment automotive production. Automotive manufacturing is a serious operating environment. Automakers care about worker safety, ergonomics, repeatability, uptime, and process control.

There are a few more examples but I will move on.

**The Customer Acceleration Program**

This is Agility’s customer conversion funnel.


This tells us how Agility is trying to industrialize selling.

They are trying to create a repeatable process:

-

Validate skill fit.

-

Replicate onsite workflow.

-

Measure KPIs.

-

Align safety and infrastructure.

-

Run a RaaS pilot.

-

Convert into ongoing deployment.

Their conversion rate here will be one of the most important operating metrics after they are public.

## The $300M Order

This is a serious demand signal.


This represents potential multi-year value expected to be realized over time, is subject to contractual milestones, relates to 1,000 Digit v5 robots under a three-year RaaS contract, includes warrants issued to the purchaser that vest proportionately as robots deploy, and is not a measure of current period revenue.

This tells us this is not clean backlog in the way we might think about a mature industrial company. This is really encouraging, but we do need to stay grounded on what it is and what it isn’t. It isn’t realized revenue today.

But a $300M+ commitment tells you a customer sees enough potential to engage at fleet scale, at least on paper. And this is tied to 1,000 robots! That indicates the customer is thinking about deployment density, multi-year usage, and larger-scale labor replacement or augmentation.

## The RaaS structure matters

The order is tied to a three-year RaaS contract.

So the customer pays recurring fees instead of buying the robot upfront. Agility keeps ownership of the robot and is responsible for more of the deployment, maintenance, and support burden.

This is attractive for customers because it lowers upfront capital requirements. They can adopt the technology without treating it like a major capital purchase.

It is attractive for Agility because it can create recurring revenue and a larger lifetime value per robot.

But it also shifts risk onto Agility.


Under RaaS, Agility has to fund or finance the robot, deploy it, maintain it, repair it, update it, support it, and keep it productive. If the robot breaks, needs service, or underperforms, Agility carries more of that burden.

So the RaaS order can be high quality if the robot works well, but it can become expensive if service costs run higher than expected.

**Purchaser warrants matter**

The footnote in the image a couple sections back says the contract includes warrants issued to the purchaser that vest proportionately as robots are deployed.

That means the customer may receive equity upside as deployments occur.

This can be interpreted two ways.

The positive interpretation is that it aligns a major customer with Agility’s success. The customer has a reason to support scaled deployment because they participate in some upside if the company works.

The cautious interpretation is that it is a commercial incentive attached to the order. We need to understand the dilution, strike price, vesting terms, and whether the order would look the same without those economics.

This is exactly the kind of detail that the S-4 needs to clarify.

## RoboFab

This section is about whether Agility can actually build the robots at scale.

RoboFab is Agility’s humanoid robot manufacturing facility in Salem, Oregon.

Agility calls it the world’s first full-scale humanoid factory. They say it has peak capacity of 10,000 robots annually and uses a modular workcell design that can support expansion.


The $300M+ Digit v5 order is tied to 1,000 robots. That means Agility needs the factory, supply chain, quality systems, field operations, and service organization to support real output. A 1,000-robot order is a major manufacturing test.

## The engineering / manufacturing feedback loop

For a product like Digit, manufacturing teaches engineering.

When you build more units, you learn which parts are hard to assemble, which components fail, which tolerances are too tight, which suppliers are inconsistent, which repairs take too long, and which designs create field issues.

That feedback can lead to fewer parts, lower BOM (bill of material), easier assembly, faster testing, and improved liability.

**Domestic supply chain**

They repeatedly emphasize their U.S. manufacturing and domestic sourcing. **75%** of Digit parts are sourced from the U.S.

This have several implications.

First, it supports the “U.S. industrial capacity” narrative. Agility is positioning itself as a domestic humanoid robotics company at a time when physical AI, robotics, reshoring, and China competition are all politically relevant.

Second, it may help with enterprise and government-related customer perception. Some customers may prefer domestic supply chains for resilience, security, compliance, or procurement reasons.

Third, it could reduce some geopolitical risk, but not eliminate it. The risk section in the deck still says Agility relies on global supply chains for semiconductors, batteries, sensors, actuators, and specialty alloys.

## **Unit Economics**

This is all about whether the robot can become a real business.

If Agility deploys one Digit, how much revenue can that robot generate, how fast does Agility recover its cost, and what margin could the company earn over time?

The unit economics in Agility’s presentation are **illustrative**.


That means they are based on management assumptions, internal estimates, and expected future business models. They are useful for understanding the company’s financial logic, but they are not audited results and not proof of what margins will actually be.

**RaaS and ownership**

Agility has two main ways to sell Digit.

**RaaS** means Agility keeps ownership of the robot and the customer pays recurring fees.

**Ownership** means the customer buys the robot upfront, then pays for software and maintenance.

These two models have very different financial shapes.

Like we said earlier, RaaS gives Agility recurring revenue, but Agility carries more asset and service burden. Ownership gives Agility more upfront revenue, but less recurring lifetime revenue per robot.

The deck says RaaS can generate about $500k of cumulative revenue over a five-year useful life, while ownership can generate about $400k over five years.

RaaS includes annual subscription fees plus a roughly $25k one-time deployment fee. Ownership includes an upfront Digit purchase, a roughly $20k deployment fee, and annual Arc subscription plus maintenance services.

So basically

RaaS = smoother recurring revenue, higher lifetime revenue, more operating responsibility.

Ownership = more upfront revenue, faster cash recovery, lower lifetime revenue.

**The customer ROI math**

Agility’s customer ROI pitch is built around labor replacement or labor augmentation.


This graphic does a good job of laying it out so I won’t just repeat it.

Agility wants Digit priced below the customer’s fully burdened labor cost while still generating strong revenue per robot.

**BOM**

This is the direct hardware component cost of building the robot.

Agility says current Digit v4 BOM is about $125k and that is has a path toward roughly $30k BOM over time, driven by engineering improvements, supplier engagement, manufacturing experience, scale, and component cost-down. Jen Hunter emphasized on the call that Agility has already reduced v4 costs and is applying those learnings to v5.


This is a huge swing factor.

If Digit costs $125k to build, RaaS payback takes longer and margins are lower.

If Digit eventually costs $30k to build, the same annual revenue per robot becomes much more powerful.

That is why the BOM curve is central to the thesis.


**Product margin targets**

Potential product margins above **70%** over time, depending on adoption model and scale.


The deck says this excludes corporate SG&A and R&D. That means these are not company-level operating margins.

This just means the per-robot economics could become very attractive before corporate expenses.

## What We Need From the S-4

In a SPAC merger, the S-4 functions like a combined proxy statement and prospectus.

It gives shareholders the information needed to vote on the transaction and evaluate the new public company. Agility’s presentation specifically says Churchill intends to file a Form S-4, and that investors should read the proxy/prospectus when available because it will contain important information about Churchill, Agility, and the business combination.

The S-4 will contain historical revenue, gross margins, cash burn and expense base, customer concentration, order definitions, RaaS accounting, warranty / service / field-support burden, cap table and dilution, cash-at-close and runway, risk factors with more detail.

So until then, we can study the company and understand everything at our disposal.

But the financial case remains incomplete.

## CCXI Deal Structure

What are public investors actually buying through CCXI?

A company can be interesting while the transaction structure still creates dilution, redemption risk, warrant overhang, and trading volatility.

CCXI is Churchill Capital Corp XI, a SPAC.

A SPAC is a public shell company that raises cash first, then later merges with an operating business. In this case, the operating business is Agility Robotics.

The expected path is:

CCXI today → merger closes → combined company becomes Agility Robotics → ticker changes to AGLT

The presentation says Agility expects to merge with Churchill Capital XI, close in Q4 2026, and trade under ticker AGLT. It also gives the headline transaction terms. $2.5B pre-money equity value, $620M expected proceeds, a $200M common-stock PIPE, and 100% Agility insider rollover and lock-up


This deal is designed to put growth capital into Agility, rather than cash out existing shareholders.

Existing holders roll. PIPE investors come in at $10. Agility gets cash to fund v5 production, customer deployments, manufacturing, software, safety systems, and operating scale.


**The PIPE**

PIPE stands for private investment in public equity. It means investors commit capital alongside the SPAC transaction, usually at a set price. Here, the PIPE is common equity priced at $10 per share.

The presentation says the PIPE is $200M from new and existing investors. The merger materials say Foxconn is the lead PIPE investor, and Peggy also said insiders committed over $60M to the PIPE with Foxconn leading.

The PIPE does two things.

First, it adds capital beyond the SPAC trust.

Second, it signals that strategic and existing investors are willing to fund the company at the transaction valuation.

**Lockups**

Lockups control when insiders can sell.

Existing Agility shareholders and Churchill XI sponsor shares are subject to a 180-day lock-up after closing. There can be early release if the post-closing company’s volume-weighted average trading price equals or exceeds $12.00 over any 15 trading days within any 180-trading-day period.

This creates a timing layer. Lockups reduce immediate insider selling pressure after close. But once shares unlock, supply can increase. If the stock trades strongly enough to trigger early release, that supply event can arrive sooner.

So the lockup is both a confidence signal and a future trading calendar item. This unlock period can provide extreme volatility.

**The $10 reference price**

The whole transaction is built around a $10 reference.

PIPE investors buy at $10.
The SPAC trust value is around $10 plus interest.
The ownership table is shown using a $10 trust value per share.

At $10, the illustrative basic share count of 336.2M implies roughly **$3.36B** of basic equity value. The transaction sources table shows $2.5B rollover equity plus $620M expected gross proceeds. The difference comes from SPAC mechanics such as sponsor/founder shares and the way the illustrative cap table is presented.

That is why we have to look past the headline pre-money value.

The better question is…what is the fully diluted market cap at the actual trading price after close?

If AGLT trades above $10, valuation expands quickly. If it trades below $10, the market is discounting the transaction value.

## Competition

Agility has one of the cleanest commercial deployment stories, but it is competing against better-funded humanoid companies, China’s faster unit-scale ecosystem, and older automation tools that may solve many of the same customer problems at lower complexity.

The way I see it, Agility faces three kinds of competition:

**1. Direct humanoid competitors**
Tesla Optimus, Figure, Apptronik, Boston Dynamics Atlas, 1X, Sanctuary, Unitree, UBTech, AgiBot, Fourier.

**2. Existing automation alternatives**
AMRs, AGVs, robot arms, conveyors, sortation systems, automated storage systems, fixed automation cells, warehouse automation platforms.

**3. Enabling-layer companies**
NVIDIA, Arm, robotics software platforms, actuator suppliers, sensors, chips, simulation tools. These can help Agility, but they can also arm competitors.

Agility’s own risk-factor slide names Tesla, Figure AI, Apptronik, 1X, Boston Dynamics, and Sanctuary AI as well-funded humanoid/bipedal competitors. It also says AMRs, fixed warehouse automation, and AI-driven software can compete with Digit at lower cost or lower operating complexity.

The main competitive question a customer is asking is:

What is the cheapest, safest, most reliable way to get this physical task done?

Sometimes the answer may be Digit.

Sometimes it may be an AMR plus a human worker.

Sometimes it may be a conveyor.

Sometimes it may be a robot arm.

Sometimes it may be a specialized automation cell.

Sometimes it may be another humanoid.

A humanoid wins only when flexibility is valuable enough to justify the complexity.

**Tesla Optimus**

The Tesla threat is scale, capital, manufacturing experience, autonomy data culture, and brand.


Tesla has enormous manufacturing infrastructure, AI engineering resources, custom silicon ambitions, and a public equity base that already gives Optimus massive optionality value. If Tesla proves Optimus can work in its own factories first, then sells externally later, it could become a brutal competitor.

Compared with Tesla, Agility’s advantage is focus and commercial warehouse/manufacturing deployment today. Tesla’s advantage is scale and capital.

**Figure**

Figure 03 is being pushed as a general-purpose humanoid for home help, powered by Helix AI for unpredictable home environments.


Figure’s bull case is that broader embodied AI and dexterous manipulation can unlock larger markets faster. Agility’s bull case is that narrower industrial deployment produces real customer value sooner. The strategic question is whether the winner starts narrow and expands outward, or starts general-purpose and compresses down into many tasks.

**Apptronik**

Apollo is designed to bring AI into the physical world through mobility, dexterous manipulation, and learning new skills, targeting physically demanding work in environments built for people. It specifically calls out manufacturing, warehousing, logistics, retail, goods-to-person, person-to-goods, packing, kitting, inspection/sorting, and machine/tool tending.


That overlaps heavily with Agility.

The difference is form-factor strategy. Apptronik says Apollo has both bipedal and wheeled mobility options, allowing it to adapt to different environments and workflows. A wheeled humanoid-style robot may have lower locomotion complexity in structured facilities. If the customer’s floor is flat and the workflow is compatible, wheels could be cheaper, safer, and easier to maintain than legs.

**Boston Dynamics Atlas**

Atlas is aimed at enterprise applications and industrial automation, with material handling, minimal supervision, barcode scanning, workflow integrations, and autonomous charging/battery swapping.


Boston Dynamics has deep robotics credibility, Hyundai ownership, and a history of world-class motion control. Atlas also carries the prestige of being one of the most advanced humanoid platforms in the world.

**Unitree and China**

Unitree represents cost, production velocity, and China robotics ecosystem scale.


If China players can produce capable humanoids at dramatically lower costs, they can pressure the entire market’s hardware pricing. Even if Agility owns the premium enterprise U.S. deployment lane, global benchmarks may compress what investors are willing to pay for robot hardware.

That said, Unitree and Agility may initially serve different markets. Unitree has been known for lower-cost robot platforms and volume. Agility is focused on U.S. enterprise industrial deployments with safety, workflow integration, and service/support.

There are many more, as I initially laid out. But I will stop here for the sake of length.

Just know that while this is the first opportunity at a U.S. listed pure-play, it isn’t the only humanoid maker.

## Closing

That’s all I’ve got! If you have made it this far, congrats! That was a long one. But it is worth understanding and getting familiar with.

I really like the position that this company is in and the stage they are in.

But in order for me to invest, I need to know their financial situation.

I do believe that the stock over the coming days/weeks will be very volatile and could appreciate significantly. But I still do not love the mechanics of SPAC’s and will personally sit this out until I get more clarity on the finances.

If I do end up taking a position in my Physical AI port, I will let you all know!

## *Financial Disclosure*

*At the time of publication, I have no position in Churchill Capital Corp XI ($CCXI), Agility Robotics, or any expected post-merger Agility Robotics security, including the anticipated ticker $AGLT. I may initiate a position in the future, but any such decision would depend on valuation, Form S-4 disclosure, cash-at-close, redemption levels, Digit v5 execution, customer conversion, and broader market conditions.*

*This report is for informational and educational purposes only. It reflects my personal research, interpretation, and opinions based on publicly available materials, company presentations, management commentary, SEC filings, customer announcements, and third-party sources believed to be reliable. It should never be treated as personalized investment advice, a recommendation to buy or sell any security, or a solicitation to enter any transaction.*

*Humanoid robotics, SPAC mergers, early-stage hardware companies, and pre-scale robotics businesses carry substantial risk. These risks include technology delays, manufacturing challenges, customer concentration, order conversion risk, safety and liability exposure, dilution, warrant overhang, redemptions, financing needs, competitive pressure, regulatory uncertainty, and the potential loss of invested capital.*

*Company-provided projections, unit economics, order values, TAM estimates, production targets, margin assumptions, and future deployment timelines are inherently uncertain. Many figures discussed in this report are forward-looking, preliminary, illustrative, unaudited, or dependent on management assumptions. Actual results may differ materially from those presented or discussed.*

*I receive no compensation from Agility Robotics, Churchill Capital Corp XI, or any related party for this report. I have no business relationship with the company, its management team, Churchill Capital Corp XI, or the transaction sponsors.*

*Readers should conduct their own due diligence, review all relevant SEC filings, read the Form S-4 once available, consider their own risk tolerance and financial situation, and consult a qualified financial advisor before making any investment decision.*
The market might think so...