Market — Cyberstrata

01 — The problem, with evidence

The robot arm is a quarter to two fifths of a working robot cell. Know-how and integration are most of the rest.

Finding	Value	Unit	Evidence	Source
The robot is the cheap part
Robot arm's share of an installed robot cell's cost	25–40	%	Reported	LA NPDT
Integration's share of the same cell	25–35	%	Reported	LA NPDT
Average cobot price	27 000	€	Reported	Qviro
Full cobot system against the cobot alone	2–3	×	Reported	Qviro
Renting does not deliver work
Humanoid rental in Hangzhou, per day, with an operator; used for displays, film shoots and education	3000	yuan	Reported	Interesting Engineering · 2025-03
AgiBot rental orders in its first 3 months	5500	orders	Reported	company claim via press · 2026-08
Robot work sells, but only from owned fleets
Formic: Series A, plus over $100 million in debt to buy equipment (2022)	26.5	$ million	Reported	SupplyChain247
Robonnement: robots bought and operated at customer sites, in 5 countries	200	robots	Reported	Robonnement
There are many robots, and more each year
Industrial robots in operation worldwide	5	million	Reported	IFR World Robotics
Share of those robots standing idle in a given night or month	—		Open	first interviews

Some robots already prove their own work: autonomous floor scrubbers report a coverage map and the exact area cleaned after every run (Avidbots).

02 — Who pays, who supplies

First customers: buildings that need floors cleaned overnight. First supply: fleets that come with their operators.

No.	Who	Pain	Buys	Confidence
Demand
D1	Facility managers and building owners (offices, malls, airports); cleaning contractors that need overflow capacity	Labour shortage, night work, one-off deep cleans	Cleaning jobs done by robots and their operators, priced per m² or per night	Medium
D2	Small and mid-size manufacturers with one repetitive task	Upfront cost; doubt that it will work for them	Palletising or machine tending, per hour or per shift	Medium
D3	Events and promotion	Attention	Humanoid appearances	Low value
Supply
S1	Fleet operators: cleaning contractors with robot fleets, robot-service providers, rental fleets	Robots stand still between contracts	Bring robots plus the operators who know them	Low to medium
S2	Integrators	Demo robots idle between projects	Bring test robots and deployment know-how	Low to medium
S3	Owners with idle robots and no operators	Capital tied up	Bring robots only; join through S1 or S2	Low

Biggest open question: whether fleets have free capacity in the windows customers need. Cleaning fleets often already work at night, so supply may come from fleets between contracts, daytime-idle robots or seasonal capacity.

03 — Landscape

Rental platforms list robots and service operators own them. Neither plans a job across many owners' fleets.

The renter still carries the hard part: does this robot fit the task, who integrates it, who runs it, and is it in good condition. Listing sites leave that open; service operators solve it by buying everything.

Player	Where	Model	What it leaves open
Rental marketplaces
Sharebot	US	Peer-to-peer: owners list robots idle between projects; rentals by day, week or month; payment held until the rental starts	Fit, integration and insurance are left to renter and owner
RoboShare	US	Matches businesses with owners of idle robots; rentals and service bookings through local partners; announced 2026-07	No traction published
Humanoid rental
AgiBot rental	China	Rental from the maker, about 3500 yuan per day including shipping and an operator	Robots need an operator; mostly events and exhibitions
Qingtian Zu	China	Shanghai leasing platform bundling hardware, content, operator and logistics	Entertainment and promotion
Robots as a service
Formic	US	Owns and runs complete robot cells; the customer pays per hour of uptime	Buys every robot; debt-financed fleet
Robonnement	Switzerland, Germany	Buys and operates robots at customer sites for a fixed monthly fee	Buys every robot
Skyline Robotics	US	Window-cleaning robot Ozmo on a 45-storey New York tower, sold as a service	One system type, owned by the operator
Bayer cobot rental	Germany	Palletising cobots for rent; part of the rent counts toward buying	One task, one owner
RobCo	Munich	Modular arms and software for small and mid-size manufacturers	Sells its own hardware
Skill app stores
Unitree UniStore	China	Motion apps for Unitree robots	One maker; mostly dances and moves
OpenMind	US	App store for humanoids and quadrupeds from several makers	Consumer and service apps; quality varies
UR+	Denmark	Over 500 certified add-ons and application kits for Universal Robots arms	One maker's ecosystem
Cautionary case
READY Robotics	US	One operating system for many robot brands; raised $41.5 million; shut down in 2024	Added cost on top of each maker's controller

"Not found" means not found in a few searches. In Europe we found rental and service operators but no robot marketplace; that is a gap to confirm, not a proven one. Prices and volumes are company claims or press reports.

04 — How we earn

At an assumed 15 % commission, one 20 000 m² office building is worth €1800–4500 a month to us.

The value is in recurring jobs in large buildings. First-time setup, such as mapping and elevator access, has to be paid back over repeat nights.

Step	Value	Evidence
Office maintenance cleaning in Germany, labour included	€2–5 per m² per month	Reported
Cleaning value of a 20 000 m² office building	€40 000–100 000 per month	Estimate
Share robots can deliver: floors, not toilets or desks	30 %	Assumption
Robot-deliverable value	€12 000–30 000 per month	Estimate
Commission	15 %	Assumption
Cyberstrata revenue per building	€1800–4500 per month	Estimate
100 buildings	€2.2–5.4 million per year	Estimate

Price range: clean-invoice, 2026share and commission: our assumptions

Market size, method first

Revenue = sites × cleanable area × price per m² × robot share × commission

We fill each term from interviews and pilots, not from market reports. AgiBot puts China's humanoid leasing market at 1 billion yuan in 2025, rising to 10 billion in 2026: a company projection, mostly display work, but a sign of how fast robot-time markets can grow.

Revenue line	When
Commission on each job's value, assumed 10–20 %	Day one
Planning and quotes for complex one-off jobs	Possibly day one
Recurring-job contracts on saved plans	Early
Verification, financing partners, skill royalties	Later

05 — Why now

Service robots already work alone and report their own work. Deploying them for a specific job is now the bottleneck.

01

Autonomy is proven in service work

Floor scrubbers clean without human intervention and are monitored remotely as fleets. Robots clean the windows of a 45-storey tower in New York.

02

Hardware gets cheaper and more general

Humanoids rent for a few hundred dollars a day and will do more over time. The bottleneck moves from hardware to planning and know-how, which is what we sell.

03

Outcome-based robot services are accepted

Formic, Robonnement and Skyline Robotics sell robot work by the hour, the month or the result. Customers already buy this way; supply is what limits it.

06 — Risks and stop rules

If fewer than 3 in 10 customers would buy, or fewer than 3 in 10 fleets have capacity to sell, we change course.

The fallback is manufacturing tasks such as palletising and machine tending (D2). Otherwise we pivot.

Risk	How we handle it
Supply is not free when it is needed; cleaning fleets already work at night	Ask about idle windows first; look for off-peak supply; match on capability × time × place
Chicken and egg: neither side comes without the other	Start with fleet operators who want more work; one city, one job type; free job plans and quotes for customers
First-time setup is slow: mapping, elevator integration	Focus on recurring jobs; save each building's plan
Transport and logistics	Local supply only at first
Liability, insurance, night access, labour rules	Insurance partner and standard contracts; safety always stays with the robot and its operator
Contractors see us as a competitor	Position as overflow capacity and a planning partner for contractors

07 — Validation plan

Twelve weeks: 30 conversations, one job planned by hand, three paid pilots, then a choice of first market.

Planned	Actual	Milestone	Output	Status
Weeks 1–2	—	30 conversations: 10 facility managers and building owners, 10 cleaning contractors and fleet operators, 10 manufacturers	Pains ranked; idle windows; willingness to pay; outcome pricing	Planned
Weeks 3–4	—	One real job planned by hand: scope, fit check, logistics, sourced fleet, quote	First job plan and planning rules	Planned
Weeks 5–10	—	3 paid pilot jobs, at least one recurring	Planned against actual time; proof accepted or not; margin; commission tolerance	Planned
Weeks 11–12	—	Choose the first market and pitch	Evidence-based plan	Planned

The planned column stays frozen at the date of this plan. Slip will be shown next to the actual date, never rewritten.

Robot work sells. Its supply is capital-bound.