I’m writing you from O’Hare International Airport after a two-day survey of the upper Midwest – some A3 business in Detroit on Monday, followed by a talk at HAND ERC Discovery Day on the Northwestern campus in Chicago (killer views of the water by both). Thanks to Rebecca and the gang for holding things down and giving me the confidence to disappear for weeks at a time in the future. I’m feeling good about the things the team has been building as we’re staring down the barrel of a new year. The newsletter, the podcast, the tantalizing things I can’t talk about.
As I mentioned during my Tuesday talk, I had some misgivings about immediately following a Nobel Laureate, but you’re talking to the host of a 2x Signal Award-winning podcast. So, you know, basically the same thing, right?
Good stuff lined up for you this week. On Tuesday, I sat down with Tutor Intelligence CEO, Josh Gruenstein, to discuss their recent $75 million fundraising round led by Andreessen Horowitz. And yesterday, Jonathan Hurst and I caught up a bit to chat about Agility’s big NASDAQ debut.
Speaking of big rounds, this week’s interview with Boris Sofman follows Bedrock Robotics' recent $600 million Series C. The convo was recorded a bit ago, so we don’t dive into the round specifically, but there’s a lot ground to excavate. Also, I always love a good excuse to chat about Anki.
Now that we're back, I would love to kick-start our jobs board into high gear. Getting people hired has always been the most rewarding part of doing a newsletter, so anything your company can do to help us fill this out every week is greatly appreciated.
What does $75 million mean for Tutor Intelligence? “More robots,” says CEO Josh Gruenstein, matter-of-factly. Fair enough. Piloting, scaling, and building more robots – both existing and new. Most importantly, however, it means delivering.
There’s no universal benchmark for success in this market. The waters have muddied of late, amid lengthy timelines, huge funding rounds, and flashy demos. In business, ROI gets us a bit closer. This is, after all, the eventual goal of even the longest runway, fullest-stack automation startup. Increasingly, it seems, these teams are calculating return on investment using human labor as a baseline.
The rising popularity of RaaS helps with these calculations. Instead of factoring in huge upfront costs, companies effectively rent the systems making for a clearer one-to-one comparison at the end of the month. Boston-based Tutor takes things a step further with a unique framing.
Customers “hire” by the “robot hour.” It’s an even more straightforward way of juxtaposing automation and human labor expense. With its current structure, clients pay between $14 and $18 per robot per hour.
“We think that the closer you can get to charging per unit of work, the more aligned you are with the customer incentive,” says Gruenstein. “If you're charging per month, maybe that robot is sitting down a portion of that month. As the robot vendor, we should be screaming. We should be incredibly motivated to get that fixed. If you're a factory that sometimes runs 40 hours a week and sometimes runs 120 hours a week, your robot's cost should be dynamic to that.”
Flexibility has always been a key RaaS selling point. The model enables factory and warehouse bosses to adapt fleet sizes to seasonality and changing demand. Tutor currently produces two robots: the palletizing/depalletizing/case picking Cassie and the bimanual Sonny. Continue reading >
The Anki/Bedrock cofounder shares lessons connecting consumer robots, self-driving vehicles, and autonomous construction equipment. Watch on YouTube.
Rodney Brooks (Robust.AI): Live from Silicon Valley, the robotics pioneer makes his triumphant return to help us mark one year of Automated.
Dr. Jan Zizka, (Brightpick): Discusses the tension between engineering quality and startup speed, and why robotics companies need to balance both.
Jagdeep Singh, (Rhoda AI): Believes the industry needs a much larger and more diverse source of training data to build robots that can handle the unpredictability of real work. At Rhoda AI, that starts with internet video.
Agility picked a rough day to go public. The Nasdaq Composite dropped 5.52%, the S&P 500, dipped 4.36%, and the Dow declined 1,833. As CNBC notes, the performance of the markets marks a reversal from post-midterm election gains, as the deflating AI bubble once again begins to take hold. It will be impossible to uncouple the first-day performance of a company increasingly focused on physical AI. The SPAC merger approved earlier this week, valued the humanoid firm at $10 per share, and it debuted as AGLT yesterday at $11.26. The stock was up $0.23 at close, and opened at $11.22 this morning – so not far from where things began.
As I write this, the figure is hovering right around $9. One is tempted to compare the first U.S. “pure-play humanoid” stock to Unitree’s IPO, which saw massive gains on day one (followed by a sharp decline and subsequent reasonable rebound), but as cofounder Jonathan Hurst told me this week, it’s difficult to compare performances in dramatically different markets. “It's such a different environment in China,” he noted. “The Chinese government is supporting this industry and identifying humanoids as a national priority, because they recognize the value. That's very different from U.S. venture capital investor culture, around providing value for customers and requiring market pull and product market fit.” My full interview with Hurst publishes soon.
ABB Thursday issued a statement confirming that China’s Office of the Working Mechanism for Foreign Investment Security Review has blocked the planned $5.4 billion sale of its spin-off robotics division to Japanese investment giant, Softbank. “ABB and SoftBank are in close dialogue and are working constructively on the transaction,” the Zurich-based automation giant wrote in a statement issued this morning. The Financial Times, which first reported the news, notes that China represents Softbank Robotics’ largest market, employing more than 12,700 in the region. It adds that the country’s regulatory bodies were largely expected to approve the deal, which was announced last October. Reports suggest regulators blocked the deal over concerns around losing manufacturing capability and the security of robotics data. Softbank's investment in U.S.-based OpenAI reportedly raised scrutiny, as well.
Roughly half-a-year after emerging from stealth, autonomous trucking firm, Humble, just announced a $51 million round. Obvious Ventures led the funding, which also featured Eclipse and Energy Impact Partners. Humble was founded in July of last year by Waabi’s head of hardware engineering, Eyal Cohen, who also spent time at Apple and Uber. “Our latest funding will advance customer pilots, testing and validation, early manufacturing, and expansion of our platform into a broader family of vehicles built for different regional freight needs,” Humble writes in a post announcing the funding. "We’ve spent years proving that a truck can drive itself. The harder question is how to make autonomy something a freight operator can depend on every day. The Hauler showed what a purpose-built autonomous vehicle can be. Now we’re bringing unified autonomy to commercial freight, one dock-to-dock route at a time.”
Axon is pumping another $1.2 billion into U.S. drone maker Skydio. The deal ups the Taser/bodycam manufacturer’s stake by another 20% at a $5.2 billion pre-money valuation. The deal gives Axon exclusive “public safety” distribution rights in the U.S. and Canada. Per Skydio CEO, Adam Bry: “This investment strengthens our ability to accelerate our technology roadmap, scale our impact in public safety and continue advancing autonomous flight across every critical industry we serve."
Good week to own a drone company. Zipline just raised a cool $1.5 billion to expand its last mile delivery service, coast to coast. The round, which values the firm at a staggering $21.5 billion, was led by Paradigm, WCM Investment Management, and Layer Global. Zipline and Walmart currently have an additional 300 sites planned, while Uber is hoping to hit one million deliveries via the service by 2029. Per Paradigm's Matt Huang, “Zipline is now one of the largest autonomous robotics companies in the world – operating with scale, safety, and a magical user experience for consumers and merchants. It's no wonder that some of the largest merchants in the world are partnering with Zipline to use their delivery network.”
Boston Dynamics this week unveiled the sequel to its trailer-unloading Stretch robot. The aptly named Stretch Gen 2 features a more durable design with swapable modular parts, designed to increase maintenance accessibility on the job. The gripper gets an upgrade, with refined sensing for better placement and grasping, along with improved rotating joints. BD also notes that it has reduced the system’s internal cabling by 81% or “about a half-mile’s worth.” The system can charge from zero to 90% in 75 minutes and is designed to unload up to 1,000 cases an hour, “a rate highly dependent on factors like case weight and how neat or disorderly cases are stacked, with a multipick feature that pushes performance to peak rates.”
Robot delivery firm Serve just launched its Physical AI Services division with three new offerings: Serve Link, Serve Data Curator, and a new foundation-model data catalog. The San Francisco startup looks to leverage the data collected by its 2,000+ robots spread across 40 cities (and 27 hospitals, factoring in the Diligent acquisition) for “other growing Physical AI fleets to help turbo charge their flywheels.” Construction robotics firm (and this week’s podcast guest), Bedrock Robotics, is Physical AI Services’ first publicly announced customer.
Disney Accelerator has worked with 60 startups since its launch a dozen years ago, including some current top name like Epic Games and ElevenLabs (as well as some personal favorites like Sphero and Littlebits). Over the summer, the division announced a 2026 cohort entirely focused on AI. The list includes Promise, OpenArt and Simile, along with two very familiar physical AI startups, FieldAI and Physical Intelligence. When the companies showed off what they’ve been working on during a Demo Day earlier this week, it was FieldAI’s work, in particular, that really captured the public’s imagination.
A short video showcased characters from "Cars" and "Wall-E" directing traffic in a Disney park. “FieldAI is exploring how robots can support parking operations across complex terrain and unpredictable weather to improve the work environment and provide a more consistent experience for guests,” the company writes. “Walt Disney Imagineering R&D is also collaborating with FieldAI to explore how advanced situational awareness and navigation capabilities may inform the future development of character-based robotic experiences in Disney parks.”
The Imagineering team is also working with Physical Intelligence. “As part of their collaboration, the teams are exploring how Pi’s models can power the practical task of handling and arranging 3D glasses for cleaning between shows,” writes Disney. “The collaboration pairs Pi’s advances in robot learning with Imagineering’s operational expertise to explore practical assistance for Disney Cast Members, and allow them to focus on guest interactions with the highest impact.”
This week Liam Critchley takes a look at some research out of ETH Zurich that combines soft robots with lab grown muscles. The latter were created using mouse cells. When combined with a hydrogel structure, they form bioactuators capable of jumping, swimming, and even grasping objects. Researchers note that the systems can walk at speeds up to 6.2 mm/s, "faster relative to its body length than any walking biohybrid robot reported so far."
Syncere, makers of the of the housekeeping lamp Lume, just announced an $11 million pre-seed, led by a16z Speedrun, and featuring Draper Associates, Physical World, SeaX Ventures, and Sequel. Lume seems to operate on a similar principle as Frog’s coat rack-esque reference design, Nome. Both aim to be home robots that don’t really look like robots, though Lume’s ambitions extend beyond picking laundry up off the floor. The system is also designed to fold laundry and make the bed.
AI2 (the Allen Institute) just announced a partnership with Google designed to improve climate projections. Google’s three-year, $5 million investment in the project goes toward expanding the AI2 Climate Modeling team behind the open-source NeuralGCM+. The existing Google.org and Google Research teams will also be lending a hand. “Much of the progress in Earth system AI modeling has focused on weather forecasting, in the range of days to months,” says AI2’s Oliver Watt-Meyer. “But communities also need to understand what could happen over decades as they plan for a changing climate. That longer view needs sustained research, and it’s where we see AI2 making a contribution. We’re grateful for Google.org’s support that will allow us to push even further in this direction, adding to the work we’re already doing like the ongoing development of the ACE climate model.”
The Association for Advancing Automation (A3) is North America’s largest automation trade association representing more than 1,500 organizations involved in robotics, artificial intelligence, machine vision & imaging, motion control & motors, and related automation technologies.
