Humanoid Daily

Agility Robotics is going public with $465 million in new capital, $300 million in pending robot orders, and a factory capable of producing 10,000 humanoid robots annually.

In this exclusive interview, Agility Robotics CEO Peggy Johnson joins John Koetsier to discuss the company's Nasdaq debut, its next-generation Digit 5 humanoid, and what it really takes to scale humanoid robotics.

We dig into:

- $465 million in capital and $300 million in pending orders

- 30+ customers in the pipeline, targeting 100+ by the end of 2027

- Digit 5's ability to work outside safety cages

- 90 minutes of work, nine minutes of charging, and 20-hour workdays

- 65,000+ hours of real-world robot training data

- 50-pound lifting capacity and interchangeable hands

- RoboFab's capacity for 10,000 robots per year

- Humanoids moving beyond warehouses into manufacturing, retail, and more

Johnson also explains why real-world deployment data is critical to robot intelligence, and why Digit 5 is the platform Agility believes it can finally scale.

Subscribe for more conversations about AI, humanoid robotics, autonomous systems, and the future of technology.

More humanoid robotics news: https://www.humanoidmegahub.com

#HumanoidRobots #AgilityRobotics #Digit5 #Robotics #PhysicalAI

00:00 Agility Robotics goes public

01:35 $465M in capital and $300M in orders

03:45 The race for robot training data

05:45 65,000 hours of real-world robot data

07:10 Why Digit 5 is ready to scale

09:35 Digit 5: 20-hour workdays and fast charging

10:50 50-pound payloads and interchangeable hands

12:20 Building 10,000 humanoids a year

13:55 Life as a publicly traded robotics company

15:15 The labor shortage driving robot demand

16:15 Humanoid robots in retail and beyond

Can a robotics world model run at 250 frames per second while NVIDIA's Cosmos runs at roughly 1 FPS? Saturn Dynamics CEO Elisa Seghetti says her company's ARC technology can do exactly that.

In this episode, we explore how world models could transform robot training, why high-quality industrial data might outperform massive datasets, and what it will actually take to get humanoid robots working reliably in factories.

We discuss:

- 250 FPS vs. 1 FPS: Saturn's claimed performance advantage over competing world models

- World models vs. simulators: How ARC differs from NVIDIA Cosmos and Isaac Sim

- Reinforcement learning: Moving robots from 70–80% success rates toward industrial reliability

- Edge AI: Why trillion-parameter models don't make economic sense for robot deployments

- Humanoid robots: Why wheeled humanoids and mobile manipulators may make more sense in factories

- The robotics supply chain: Why today's vertically integrated robot companies may eventually resemble the automotive industry

- Robot training data: Why Saturn is betting on high-quality industrial demonstrations instead of egocentric data

Seghetti also shares her predictions for the next 12–18 months, including why the robot data provider market could shrink as better models learn from fewer demonstrations.

00:00 Introduction: the race to train robots

01:00 Meet Saturn Dynamics CEO Elisa Seghetti

01:45 The different approaches to robot training

03:45 Can frontier AI models control robots?

05:20 How world models work

07:30 Making robots easier to train

09:15 Saturn ARC vs. NVIDIA Cosmos and Isaac Sim

11:10 Where humanoid robots fit in

13:45 Are robotics companies wasting money on training?

15:00 Why robotics needs a supply chain

17:20 Vertical integration and robot AI security

19:00 Safety, guardrails, and industrial deployment

21:20 The next 18 months for Saturn Dynamics

22:40 Will the robot data market shrink?

23:40 Closing thoughts

#Robotics #HumanoidRobots #PhysicalAI #WorldModels #NVIDIA #RobotLearning

What if you could program an industrial robot just by talking to it?

That's what Standard Bots is making possible with a new generation of AI-powered robot training tools. In this episode of Humanoid (almost) Daily, I talk with Evan Beard, co-founder and CEO of Standard Bots, about making industrial robots as easy to use as an iPhone ... and why he believes this could help bring about the ChatGPT moment for physical AI.

Standard Bots is announcing three new capabilities that dramatically simplify robot deployment and training:

- Agent: Vibe code your robot using natural language. Tell it what you want to automate, and AI helps build the program.

- Locate: Train computer vision models to recognize and locate objects in seconds, without expensive custom engineering.

- Inspect: Working with Roboflow, train AI models to identify defects, scratches, and other quality issues.

We also discuss Showpath, a beta capability that allows robots to learn new tasks from just a few demonstrations. Evan explains why traditional industrial automation is so expensive and brittle, why most robots still require specialized engineers to program them, and how AI could make automation accessible to smaller businesses.

The bigger vision?

Building the Anthropic of the physical world — and making robots intelligent enough to adapt when real-world conditions change.

In this conversation:

- Why deploying industrial robots traditionally costs so much

- How natural-language programming changes robot automation

- Training robot vision models in seconds instead of weeks

- Teaching robots new tasks through demonstrations

- Why flexible AI-powered automation could deliver better ROI

- The possibility of a ChatGPT moment for robotics

About Standard Bots:

https://standardbots.com

More humanoid robotics and physical AI news:

https://humanoidmegahub.com

Subscribe for conversations with the people building the future of robotics, AI, and automation.

#robotics #PhysicalAI #ArtificialIntelligence #StandardBots #RobotTraining #Automation #VibeCoding

The future of humanoid robots might not look human at all. In fact, the "humanoid robots" of the future could have wheels instead of legs, four arms instead of two, or even look more like spiders than people.

That's the provocative argument from Haomiao Huang, founding partner of Matter Venture Partners, which just raised a $450 million fund to invest in robotics, physical AI, semiconductors, and other deep technologies.

His perspective: "Humanoid is a state of mind, not a specific type of robot."

What matters isn't whether a robot walks on two legs or looks like a person. It's whether that robot can perform multiple tasks, manipulate objects, navigate human environments, and adapt to new situations without being completely reprogrammed.

In other words, the real revolution isn't humanoid robots. It's general-purpose robots.

In this episode of Humanoid (almost) Daily, we explore why Huang believes we're entering a golden age of robotics, what's still holding the industry back, and what the next generation of robots might look like.

We discuss:

- Why humanoid robots don't actually need to look human

- How general-purpose robots could transform manufacturing, logistics, and eventually our homes

- Why today's humanoids resemble the clunky wearable computers of the 1980s

- Whether robot manufacturers are scaling before finding product-market fit

- Why AI models, compute, and demand are converging to accelerate robotics

- The hardware and training-data bottlenecks still holding robots back

- Why robots like Dexterity's Mech and Path Robotics' welding systems might represent the near-term future

- Where Matter Venture Partners is investing its $450 million

- Why the most capable robots of the future might resemble spiders more than humans

And we get Huang's predictions for what robotics will look like in five to ten years.

00:00 – The golden age of robotics

00:17 – Meet Haomiao Huang: The $450M physical AI fund

00:55 – If AI eats the world, hardware is its teeth

02:05 – Why physical AI is replacing the metaverse hype

03:49 – Bringing AI agents into the physical world

05:44 – Three forces creating the robotics boom

08:22 – Humanoid is a state of mind, not a robot shape

11:40 – Are humanoids scaling before product-market fit?

12:59 – Why robot reliability still isn't good enough

15:11 – Will we need humanoid-safe products?

16:01 – Hardware, AI, data: What's the biggest bottleneck?

16:39 – Why advanced robotics training data is scarce

17:24 – Today's robots are like early personal computers

20:40 – Why robot makers are building custom hardware

22:19 – Robotics, geopolitics, and military applications

24:07 – Where Matter will invest $450 million

26:24 – Predicting humanoid robots in 2031

27:25 – General-purpose industrial robots will scale first

28:31 – Why the ultimate robot might look like a spider

30:03 – Figure, 1X, and the race toward useful home robots

32:06 – Final thoughts

Schaeffler plans to put at least 1,000 of Hexagon's Aeon humanoids to work across its global factory network by 2032. In an industry where most deployments are still counted in single digits, that's a very big number.

How is that going to work?

How does that change things from pilots with 1 or 10 or 50 robots?

How does that change the jobs robots do?

What does that change about the jobs humans do?

In this long-form episode of Humanoid (almost) Daily, we do a deep dive with Hexagon Robotics president Arnaud Robert.

Enjoy!

00:00 What does it take to deploy 1,000 humanoids?

01:31 Why 1,000 robots changes the problem

03:02 Why humanoids need to do multiple jobs

03:55 The race for physical AI training data

04:51 Simulation, synthetic and real-world data

06:20 Hexagon's sim-to-real-to-sim loop

08:10 What robot-generated training data captures

10:21 Pilot vs. production-ready humanoids

12:14 The industrial experience advantage

13:12 The two metrics that really matter

15:39 What Hexagon learned from failure

17:35 The humanoid actuator problem

18:47 Why Aeon uses wheels instead of legs

19:59 Yes, a wheeled humanoid can climb stairs

21:28 What the factory of the future looks like

23:42 Using humanoids to eliminate bottlenecks

25:17 Automation exposes the hidden 3%

26:07 Schaeffler's path to 1,000 humanoids

27:18 Train, deploy, scale

What if deploying a new robot didn't take weeks or months ... but hours?

Ashish Kapoor, founder and CEO of General Robotics and former head of Microsoft's Robotics and Autonomous Systems Group, joins Humanoid Daily to explain why robotics still has a massive research-to-deployment gap.

And he shares how AI agents could help close it.

General Robotics is building GRID, a unified robotics platform designed to bring together robots, simulation, AI models, training, deployment and engineering tools in one place where AI agents can work across the entire stack.

We talk about:

• Cutting robot onboarding from a month to as little as 2 hours

• Using AI agents to engineer and deploy robotic systems

• Why robotics is much harder than agentic software development

• Creating custom robot simulators in minutes

• Why future enterprise robot fleets will include multiple OEMs and form factors

• The three layers of robot training

• Why foundation models could reshape the robot-data race

• Why post-training may ultimately capture much of the value in robotics

Kapoor's big thesis: robotics doesn't just need better AI models. It needs a new engineering architecture that can actually get those models onto real robots in the real world.

00:00 The robotics deployment problem

02:15 Auto Engineering: virtual robot engineers

04:20 Why AI agents are harder in robotics

06:35 What General Robotics built with GRID

09:40 AI-generated robot simulations

11:40 Onboarding almost any robot

14:05 The future of mixed robot fleets

16:15 Where General Robotics fits in the stack

19:10 Why robotics needs a different startup playbook

22:05 The three layers of robot training

25:05 Will foundation models change the robot data race?

28:10 Where the value in robotics will be captured

30:00 Closing thoughts

Boston Dynamics just unveiled new hands for its Atlas humanoid robot: four fingers, 13 degrees of freedom, and no pinky.

That's a deliberate choice. 1X, Tesla and Genesis AI are building hands with 20 or more degrees of freedom. Boston Dynamics went the other way, betting on simplicity, strength, ruggedness, and hands that can be accurately simulated for reinforcement learning. The team even taped their pinkies to their ring fingers for a day to decide whether Atlas needed one.

In this video I break down what's new, how it compares to other humanoid robot hands, what Boston Dynamics hasn't told us yet, and why simulation might matter more than finger count.

Get the full story on HumanoidMegaHub

https://www.humanoidmegahub.com/news/boston-dynamics-new-atlas-hands

Chapters

0:00 No pinky?

0:12 What's new in the Atlas hand

0:30 How it compares to 1X, Tesla and Genesis AI

0:45 The pinky experiment

1:25 Built to survive impacts

1:55 Why simulation matters

2:25 Built for industrial tools

2:35 What we still don't know

Agility Robotics is launching Digit 5, its newest humanoid robot. Digit 5 is designed to work in close proximity with humans without the physical safety barriers that have traditionally separated robots from workers.

In this episode of Humanoid Daily, we chat with Agility Robotics CEO Peggy Johnson.

She explains what has changed with Digit 5, including a 50-pound payload capacity, faster charging, improved vision and human detection, swappable end effectors, and the ability to move between multiple tasks inside logistics and manufacturing facilities.

We also dig into one of the biggest challenges facing humanoid robotics: safety.

Peggy explains how Agility is working with ANSI and ISO standards, how Digit detects and reacts to nearby humans, and why removing safety cages could unlock much broader deployment of humanoid robots.

We discuss where humanoids fit alongside traditional automation and autonomous mobile robots, why Agility chose legs instead of wheels, what the company has learned from more than 65,000 hours of real-world robot operation, and how customers including major logistics and manufacturing companies are helping shape Digit’s development.

Peggy also shares how Agility trains new robotic skills using a combination of real-world operational data, simulation tools such as NVIDIA Isaac Sim, and learning from demonstration.

Plus, we talk about the increasingly competitive humanoid robotics market, Agility Robotics’ U.S.-based manufacturing strategy, RoboFab, and what Digit 5 could be capable of a year from now compared with day one.

Topics include:

• Digit 5 and human-robot collaboration
• Humanoid robot safety
• 50-pound payload capacity
• Fast charging and multi-shift operation
• Robotics standards and regulation
• Legs versus wheels
• Warehouse and manufacturing automation
• Agility Robotics’ RoboFab
• Physical AI and robot training
• NVIDIA Isaac Sim
• Learning from demonstration
• The growing humanoid robotics market

00:00 Digit 5 can now work around humans

00:32 Introducing Agility Robotics’ Digit 5

01:17 50-pound payloads and longer operating time

02:03 Fast charging vs. swappable batteries

03:02 Why Digit can move differently than humans

04:03 The growing industrial labor gap

05:01 Breaking the humanoid safety barrier

05:41 How Digit detects nearby humans

06:18 Building safety standards for humanoid robots

07:05 When Digit 5 will begin deploying

07:30 How fast should a humanoid robot walk?

08:02 What happens when a human gets too close?

09:04 Why Agility is taking a cautious approach

10:03 Swappable grippers and robotic end effectors

11:05 Using the existing robotics ecosystem

11:42 Where humanoid robotics is today

13:00 Moving from one task to end-to-end workflows

14:17 The factory of the future

15:30 Humanoids vs. wheeled robots

16:41 Why legs make sense for Digit

17:36 Connecting Digit with other automation

18:03 The exploding humanoid robotics market

18:42 Agility’s real-world advantage

19:17 What Agility has learned from customers

20:28 How U.S. policy affects Agility Robotics

21:05 RoboFab and Agility’s U.S. operations

22:04 Expanding Digit 5 into Europe

23:00 How Digit 5 will gain new skills

23:39 The race to train humanoid robots

24:31 Real data, simulation and Isaac Sim

25:02 Learning from demonstration

25:48 Final thoughts

Humanoid robots could become one of the largest technology markets in history, potentially worth tens of trillions of dollars.

In this episode of Humanoid Daily, I talk with Andrew Kang, CEO of Robostrategy, about the rapidly accelerating humanoid robotics industry, the companies he thinks are best positioned to win, and why the market could explode over the next few years.

We dig into Tesla Optimus, Figure AI, Apptronik, Agility Robotics, Unitree, Agibot, 1X, Genesis AI, Neura, Humanoid AI, and other emerging robotics players in the US, Europe, and China.

We also explore the picks-and-shovels opportunity behind humanoid robots: actuators, motors, components, AI systems, and the infrastructure required to manufacture robots at massive scale.

Plus: why Chinese robotics companies have helped accelerate the entire industry, what new FCC restrictions could mean for US robotics companies, and why Andrew believes millions of humanoid robots could be operating in the world before 2030.

Topics include:

• The massive jump in humanoid robotics valuations

• Tesla Optimus vs Figure AI vs Apptronik

• Why not every humanoid robotics company will survive

• The most promising robotics companies in the US

• Europe’s emerging humanoid robotics ecosystem

• The opportunity in actuators, motors, and robotics components

• US vs China in humanoid robotics

• The impact of FCC restrictions on Chinese robots

• Low-cost humanoid robots and developer platforms

• Why Unitree helped accelerate robotics research

• How large the humanoid robotics market could become

• When millions of humanoid robots could arrive

• What the robotics adoption curve could look like

If humanoid robotics follows anything close to the trajectory of AI, GPUs, smartphones, or autonomous vehicles, the next few years could be extraordinary.

Subscribe for more conversations about AI, robotics, emerging technology, and the companies building the future.

Chapters

00:00 Humanoid robotics could be worth tens of trillions

00:27 Meet Andrew King and Robostrategy

01:08 The robotics IPO and valuation boom

02:11 Is Agility Robotics undervalued?

04:48 The US humanoid companies most likely to win

06:12 Tesla Optimus vs Figure AI and Apptronik

09:07 The most interesting European robotics companies

10:03 Picks and shovels: where to invest beyond robot makers

14:37 The actuator opportunity in humanoid robotics

15:43 FCC restrictions and the US-China robotics battle

17:32 The rise of ultra-cheap humanoid robots

20:38 The Chinese humanoid robotics companies to watch

22:40 How big could the humanoid robotics market become?

24:49 When humanoid robots become material to the economy

26:24 Could robotics adoption follow the GPU boom?

27:39 Final thoughts

Are humanoid robots really the future of the home, or are we looking at consumer robotics the wrong way?

In this episode of Humanoid Daily, John Koetsier talks with Amber Atherton, serial entrepreneur and investor at Patron, about why the next wave of home robotics may look less like a humanoid assistant and more like a collection of highly specialized, beautifully designed devices that each take ownership of a specific part of everyday life.

That could mean robots for skincare, hairstyling, gardening, cleaning, wardrobe management, or dozens of other repetitive household rituals.

Amber argues that consumer adoption often happens through specialization rather than generalization. Instead of one robot that does everything, she sees an opportunity for what she calls “enchanted objects”: intelligent devices that quietly automate specific routines and eventually disappear into everyday life.

John and Amber also discuss why Roomba remains such an important example in consumer robotics, how AI foundation models are changing robot development, why falling costs for motors, sensors, batteries, and compute matter, and what price consumers might actually pay for useful household robots.

They also debate whether humanoids will eventually dominate, how manufacturing and supply chains could shape the market, the importance of hardware ecosystems and recurring revenue, and whether the future operating system for physical AI will be controlled by companies like Google DeepMind, NVIDIA, or an entirely new generation of startups.

Finally, they imagine what the average home could look like by 2036, and why the biggest robotics companies of the future may not look like robotics companies at all.

Topics include:

• Humanoid robots vs. specialized robots
• Consumer robotics and embodied AI
• Robots for skincare and hairstyling
• The future of household automation
• Why Roomba succeeded
• AI foundation models for robotics
• Robotics manufacturing and supply chains
• Pricing home robots
• Hardware subscriptions and app ecosystems
• Physical AI and robot training data
• Google DeepMind, NVIDIA, 1X, Figure, and Apptronik
• What the smart home could look like in 2036

Chapters:

00:00 Robots for skincare, hair, and household rituals

00:19 Meet Amber Atherton

01:16 What chore should robots eliminate first?

02:48 The overlooked opportunities in consumer robotics

04:15 Specialized robots vs. humanoids

06:23 Why technology both consolidates and specializes

07:40 Could robotic hands braid your hair?

08:10 Why home robots may not need humanoid bodies

10:10 The kitchen proves specialization can win

12:02 Why Roomba is still the consumer robotics benchmark

12:24 Why consumer robotics could take off now

14:06 AI is dramatically accelerating hardware development

15:04 Manufacturing, China, and the future of U.S. robotics

16:44 Could regulation actually favor humanoid robots?

18:01 What price will consumers pay for a home robot?

19:38 What would a skincare robot be worth?

21:50 The business model for consumer robotics

22:29 What investors are getting wrong about humanoids

24:09 Building the ecosystem for home robotics

25:00 What will the average home look like in 2036?

26:25 Who will own the operating system for physical AI?

27:12 The opportunity in robot training data

28:21 Final thoughts