Agibot ships 5,000 humanoids in 90 days

PLUS: Humanoid runs warehouse via cloud, Physical Intelligence eyes $11B valuation, and autonomous farming software lands €4M


Agibot ships 5,000 humanoids in 90 days

Welcome back to your Robot Briefing

Chinese robotics maker Agibot just compressed three years of production into 90 days, shipping 5,000 humanoid robots in a single quarter. The company says the milestone proves commercial humanoids aren't a future promise — they're a current workforce option.

If a relatively unknown Chinese firm can scale this fast, what does that mean for Western manufacturers still running pilot programs? The gap between testing robots and deploying them at scale might be closing faster than most businesses realize.

In today's Robot update:

Agibot ships 5,000 humanoids in one quarter
Cloud-controlled robot works automotive warehouse
Physical Intelligence chases $1B raise at $11B valuation
German startup puts autonomous farming in tractors
News

Chinese robotics firm ships 5,000 humanoids in 90 days — matching its prior 3-year total

Timeline chart illustrating Agibot's manufacturing acceleration, showing the first 1,000 robots took 2 years to build, the next 4,000 took 1 year, and the most recent 5,000 took just 3 months, representing a 4x increase in efficiency.

Image Source: There's A Robot For That

Snapshot: Agibot delivered 5,000 humanoid robots in three months, compressing what previously took three years into a single quarter and hitting 10,000 total units shipped. The Chinese manufacturer claims this acceleration proves humanoids are ready for workforce deployment today, not in some distant future.

Breakdown:

Agibot's production timeline shows dramatic compression: 2 years to reach 1,000 units, 1 year to hit 5,000, and just 3 months to double from 5,000 to 10,000 units — a 4x acceleration in manufacturing efficiency.
The company states it's deploying robots in logistics and industrial manufacturing sectors, targeting what it describes as global workforce demand rather than pilot programs.
Agibot claims 90% of the global humanoid market is manufactured in China, positioning this as validation that mass production economics have arrived for the category.

Takeaway: The production curve matters more than the absolute numbers — when a manufacturer can compress years of output into quarters, it signals supply chains and manufacturing processes have matured past the prototype phase. Operations leaders should watch whether these units stay deployed for 12+ months, which would indicate customers are finding actual ROI rather than running expensive experiments.

News

Humanoid robot executes warehouse tasks via cloud software at automotive plant

Snapshot: HMND's wheeled humanoid completed a live proof-of-concept at Martur Fompak, an automotive supplier, where it received task instructions from SAP's enterprise system over the internet and autonomously picked and delivered parts in an active warehouse. The robot worked within existing workflows, not in an isolated test environment.

Breakdown:

The robot handled standard warehouse picking tasks — navigating to pallet locations, picking up KLT boxes up to 17.6 pounds, and delivering them to a trolley — all while the production facility remained operational around it.
Integration happened through SAP's Joule agent layer, allowing the warehouse management system to send tasks directly to the robot over the internet without custom local controllers or middleware.
HMND's KinetIQ AI platform managed the robot's behavior and supports coordinating multiple robots that can shift tasks based on production demand, suggesting the architecture scales beyond single-unit deployments.

Takeaway: The business signal isn't the robot itself — it's that enterprise software vendors like SAP are building robot task management into their core platforms. When warehouse management systems can dispatch robots the same way they route human workers, the integration burden drops from months-long IT projects to configuration exercises, which changes deployment economics for mid-sized facilities.

News

Physical Intelligence in talks for $1B raise that would double valuation to $11B

Snapshot: The robotics foundation model startup is negotiating a funding round that would double its valuation to over $11 billion just months after raising more than $1 billion. Investors are showing increased appetite for companies building AI models that work in the physical world, with Physical Intelligence developing software that lets robots learn movements for tasks like folding clothes and assembling boxes.

Breakdown:

The company has raised over $1 billion to date from investors including Amazon founder Jeff Bezos, and this new round would bring total capital to approximately $2 billion at a valuation that rivals established robotics companies.
Physical Intelligence is building vision-language models that let robots perform any task a user requests, similar to how ChatGPT responds to text prompts — the software has run on robotic arms in test environments handling assembly and manipulation tasks.
Investor interest in physical AI companies is accelerating, with Skild AI recently valued at over $14 billion and Mind Robotics at around $2 billion, suggesting venture capital is betting on a winner-take-most market for robot foundation models.

Takeaway: Foundation model companies are raising capital at software valuations while selling to a hardware market — that gap matters for buyers evaluating vendor stability. The strategic question for operations leaders is whether to wait for dominant platforms to emerge or deploy current-generation systems, knowing the software layer underneath may get commoditized by billion-dollar model providers in 18-24 months.

News

German ag-tech startup raises €4M to put autonomous farming software in tractors

Snapshot: Nature Robots closed a €4 million seed round led by Climentum Capital and Bayern Kapital to scale its modular autonomy platform that lets agricultural machinery manufacturers add self-driving capabilities without years of in-house development. The software enables equipment to operate autonomously across large fields, vineyards, and specialty crops.

Breakdown:

The company's modular system allows machinery manufacturers to implement specific autonomous functions they need rather than building complete autonomy stacks from scratch, reducing the time-to-market barrier that has kept many ag equipment makers on the sidelines.
Nature Robots emerged from the German Research Center for Artificial Intelligence (DFKI) and operates in Osnabrück's Agrotech Valley, surrounded by global ag equipment leaders like Claas, Amazone, and Grimme — suggesting potential integration partnerships with established OEMs.
The platform handles applications from basic field navigation to precision tasks like laser weeding with millimeter accuracy, which reduces pesticide use to near-zero levels and enables what the company calls "spot-farming" economics.

Takeaway: Ag equipment autonomy is following the same pattern as automotive — suppliers building the software layer so manufacturers can add autonomous features without becoming AI companies. For operations leaders in industries with specialized equipment, this signals that autonomous capabilities will likely arrive as vendor-supplied modules rather than requiring ground-up system replacements, which changes the capital planning conversation from "replace everything" to "upgrade existing."

Other Top Robot Stories

Amazon acquired Fauna Robotics two months after the New York startup launched Sprout, a 3.5-foot humanoid designed for homes and schools rather than warehouses — adding social robotics to Amazon's 1-million-robot logistics fleet.

LimX unveiled its Luna humanoid at China's Taobao Influencer Festival, performing a catwalk and illusion turn with 33 degrees of freedom — marking a strategic shift from its industrial OLI platform toward lifestyle-oriented robots for public interaction.

Sharpa demonstrated autonomous apple peeling using dual dexterous hands powered by its MoDE-VLA system, which combines vision, language, force, and touch sensing to control 22 active degrees of freedom per hand — tackling one of robotics' hardest manipulation challenges.

Unitree deployed its G1 humanoid at Brooklyn Bridge Park, where the 4'2" robot ran through crowds of children for two hours on a single charge — drawing mixed public reaction between entertainment value and dystopian concern over AI-human interaction.

🤖 Your robotics thought for today:

Agibot went from 5,000 units in three years to 5,000 units in 90 days. That's not incremental improvement — that's a manufacturing curve hitting escape velocity. Meanwhile, Physical Intelligence is chasing an $11B valuation selling software that's still folding laundry in labs.

I'm watching which model actually puts robots on factory floors at scale.

Until Friday,
Uli

Agibot ships 5,000 humanoids in 90 days

Great! Check your inbox and click the link to confirm.
Please enter a valid email address.