Hardware is hard... to churn away from. Suck on that, software and AI!
Plus takeaways from the IEEE Hard Tech Venture Summit and Robotics Manufacturing Workshop, Chef Robotics dishes up 100 million meals, Ultra Robotics launches Operator
Apparently we are now living in the post “Software is eating the world” era, in spite of being in the AI bubble. In fact, while AI is largely credited with the eating of software that alone doesn’t explain why so many hard core software investors are suddenly becoming hot hardware only fans. The physical AI cool-aid is just an apéritif for the main course of physical industry and infrastructure, all being rebuilt with robots.
And hardware is stickier than software or AI right now. All the things that can make hardware a slow sale cycle are the same things that make it much much harder to move away from. Suddenly this looks like a feature, not a bug.
The market opportunity of rebuilding the world’s physical infrastructure layer is beyond anything ever imagined. Each previous technology revolution was largely limited to a few domains - even computers, with the joke being that we can see the productivity figures for the computing revolution everywhere but in the statistics.
All the world’s compute layers are still built on physical infrastructure. This is why I first came to Silicon Valley almost twenty years ago. I had a thesis that the emergence of robotics would need the convergence of software, startups AND silicon.
Hardware? Hardware is here until it rusts or busts.
IEEE Hard Tech Venture Summit @ SRI - Menlo Park
100+ people (investors and startup founders) attended the first of the 2026 IEEE Hard Tech Venture Summits with our Manufacturing for Robotics and Hardware Startups Workshop. It’s so encouraging to see years of volunteer hard work and startup feedback scaling globally in a successful format. We know it’s successful because even investors are paying to attend and many attendees are returning for the next event.
“Most useful startup event I’ve ever attended!”
Thanks to our venue hosts, partners and organizers: SRI, IEEE Entrepreneurship, IEEE Systems Council, Silicon Valley Robotics, Federal Laboratory Consortium, TVLP Institute Silicon Valley.
Plus core organizers: Joanne Wong, Bruno Jafelice, Cassandra Carothers, Sabrina Sasaki, Caitlin Zubrowski, Paolo, Daniel, Malcolm Knapp (and many others I haven’t included - sorry!)
More summits coming up this year in Boston and Toronto: https://entrepreneurship.ieee.org/venturesummits
Here are some takeaways from our speakers and mentors, Bilal Zuberi, Troy Helming, Rui Xu and Ted Larson
Bilal Zuberi, Red Glass Ventures and keynote speaker:
“AI is rapidly changing the robotics field but we are still early, and in development. Although progress is happening really fast across all layers of the stack, some people may have unrealistic expectations in the short term. But the long term arc has definitely changed, and when this new tech gets deployed at scale, it will create a revolution!”
Troy Helming, CEO of EarthGrid and keynote speaker
“Get the prototype right.”
Initially EarthGrid wanted to build two machines, an excavator and a tunneler. Talking to customers confirmed demand for the excavator which was a one year build, instead of a four year build. But the excavator allowed the team to gather data and test assumptions for the development of the tunneler in stage two.
“And so that was a way that we could get something out to market faster, cheaper, with near term revenue potential. Then you can work on either a second product or more advanced version of the first from the lessons that you’ve learned.”
There was a heck of a lot more great information from Troy, but if I told you, I’d have to use the plasma torches on you.
Rui Xu, Manufacturing Mentor
“If you’re an engineer or founder in Silicon Valley building consumer hardware — smart home, wearable, AI companion, anything with a mic and a screen — and you’re still defaulting to Android: the economics have already moved on. The founders who go embedded-first will ship products at BOM costs their competitors cannot touch. The engineers who go deep on RTOS and MCU firmware are building the scarcest, most valuable skill set in hardware for the next decade.
The embedded shift already happened. It looks like a $2.50 MCU with a good microphone and a cloud connection. The question is whether you’re building with it or still waiting for memory prices to come back down. They’re not.”
Ted Larson - Manufacturing Mentor
“Common mistakes in hardware design often appear early in the process and quietly compound over time. As a result, teams can face missed deadlines, rising costs, or products that fail to meet expectations.
OLogic frequently works with teams that build early prototypes using hobby-grade components or parts sourced from places like Amazon. While these parts may work for testing, they often create serious challenges during manufacturing.
Instead of waiting until late-stage development, incorporate manufacturing considerations early. By partnering with engineers who have taken products from concept to the factory floor, teams can select realistic components, materials, and assembly methods. This approach reduces risk, cost, and delays later on.”
Ted has a whole post on Common Mistakes in Hardware Design on the Ologic Blog
Want to host an event or reach the robotics community?
Many companies are reaching out about hosting events, outreach or hiring and we have a great community network, of people looking for work, of investors interested in deal flow and of technologists interested in learning about the latest advances.
Would you be interested in hosting the next Bots, Bevs and Devs? Or something customized specifically for your goals? Reach out here!
Chef Robotics dishes up 100 million servings of food
According to CEO Rajat Bhageria, Chef Robotics now has an order-of-magnitude more servings in production than all other food robotics companies combined, and more real world deformable-material training data than any other physical AI company.
Their key discovery was that simulation didn’t work well for variable, deformable and unpredictable materials so getting into real world production rapidly was the only way forward. This kicked off a great data/capability flywheel.
https://chefrobotics.ai/post/100-million-servings-and-counting
TechCrunch link: https://techcrunch.com/2026/04/17/chef-robotics-escaped-the-robot-cooking-graveyard-and-says-its-thriving-heres-why/
The Information gives Gemini faint praise
I suppose it’s difficult to find headlines for each new model release, but I think it was kind of snarky to say “Google’s New Model Makes Robotic Brains Slightly Smarter”.
Carolina Parada, Senior Director and Head of Robotics at Google DeepMind was a little more enthusiastic.
Love this video from Boston Dynamics using #GeminiRobotics’ embodied reasoning to help #Spot complete a to-do list with its new 🦿🧠!
#GeminiRobotics-ER 1.6 is a significant upgrade to our reasoning-first model that enables robots to understand their environments with unprecedented precision, enabling Spot to:
☑️Read the whiteboard
☑️Plan long horizon tasks
☑️Tidy up the scene
And then take Koda for a walk 🐕 🩷
See more information and the video on the Boston Dynamics blog.
Path Robotics adds welding torches to quadrupeds
Fortunately this isn’t for cleaning up the house, but for very constrained and useful use cases, ie. welding in shipbuilding where traditional automation has struggled, access is difficult and the labor force is retiring. Autonomous maritime vessel company Saronic Technologies is an early adopter. Path Robotics says production will be limited to 50 units, with shipments planned for 2027, and companies can sign up now to explore deployment. More information at Path Robotics
Pi claims emergent behavior in latest model
We’ve trained a new model, π0.7, that exhibits a step-change in generalization. π0.7 is a general-purpose model that can perform a wide range of dexterous tasks with the same performance as fine-tuned specialists, but even more importantly, it can follow new language commands and perform tasks that were never seen in its training data. In our experiments, we see π0.7 exhibiting the first signs of compositional generalization, recombining skills from various tasks to solve new problems, like using new kitchen appliances and even enabling a new robot to fold laundry for which there is no laundry folding data. More information at Pi
Ultra Robotics launches novel humanoid form factor
Ultra Robotics got their start in warehouses with dual armed manipulation using more traditional robot arms with vision. Operator, their new iteration reminds me of Robonaut on the Canadarm and looks very effective in terms of load, reach and access. Basically it’s half a humanoid as an end effector on a robot arm.
Skild.AI acquires Zebra, the AMRs formerly known as Fetch Robotics
This actually creates an easy entry point into warehouse automation, with the goal of setting up a multirobot, multideployment data flywheel. It’s a big bet which is dependent on Skild continuing to grow rapidly.
Logistics operators no longer need to engineer their warehouse around their robots and deal with several partners — each providing one fragment of the automated solution. Skild AI’s acquisition of Zebra’s Robotics Automation business will create the first organization that can provide a full end-to-end automation solution for warehouses: humanoids for pick-place, robotic dogs for inspection, robotic arms for packing, AMRs for material movement and an orchestration layer to control them all.
Researchers at RAI look into what DDD really means
This really resonates with me. Coming from a background of HRI, trying to streamline the adoption of good robots into society in order to solve global problems, I’ve long been aware of two big roadblocks. One is we really don’t have enough data or research about robots in the real world. The second is that economists, politicians, roboticists, investors and other experts engaged in real world deployments don’t pay attention to what research does exist, or fact check the research that they use.
Most commonly accepted wisdom about humans and robots is just fearmongering or wishful thinking. Kudos to the RAI team, who released this paper for HRI 2026: Nozomi Nakajima, Pedro Reynolds-Cuéllar, Caitrin Lynch, and Kate Darling.
When roboticists think about where robots can best fit into our daily lives, there is often a focus on "Dull, Dirty, and Dangerous" (DDD) jobs. But what if those terms aren't as straightforward as they sound?
Before we automate, we need to understand the whole picture for this type of work.
We explore the social science behind DDD work and offer a framework to better understand the context of DDD jobs.
2026. Dull, Dirty, Dangerous: Understanding the Past, Present, and Future of a Key Motivation for Robotics. In Proceedings of the 21st ACM/IEEE International Conference on Human-Robot Interaction (HRI ’26), March 16–19, 2026, Edinburgh, Scotland, UK. ACM, New York, NY, USA. Read more at RAI Blog
Instawork gets into the dirty, dull and dangerous
Instawork is transforming gig work with a new robotics business. The company launched Instawork Robotics Lab on Thursday and a robotics certification program, and Instacore, a wearable camera system, designed to collect detailed robotics training data.
Announcing the Open Robotics Technology Strategy for 2026!
As robotics continues to evolve at a breakneck pace, Open Robotics is committed to providing the trustworthy, agile, and open-source software our worldwide community needs to build the future while sustaining our existing users.
To guide our projects and amazing open-source contributors over the next year, we are thrilled to announce our three core strategic priorities for 2026:
🤖 Enhance Support for Physical AI: In the next year, expect to see our projects enable Physical AI deployment with GPU-optimized data transfer workflows, measurable performance metrics, and versatile deployment mechanisms that bring Physical AI to a diverse set of hardware platforms.
📚 Expand Accessibility & Ease of Use: We want you to build and test robotic systems faster. Look forward to streamlined installation via modern package managers (like prefix.dev's Pixi and modern CMake) and a massive overhaul of our documentation.
⚙️ Adopt Modern Tools for Production Systems: Reliability is key for robotics. We are bringing in modern practices to meet contemporary industry and legal standards for robotics deployments, including support for Rust, Bazel, and Software Bills of Materials (SBoMs).
We’re always looking for new contributors and maintainers, and we would love your help building the future!
Read our full 2026 Technology Strategy here: https://osralliance.org/open-robotics-technology-strategy-for-2026/
Silicon Valley Robotics Investor/Startup Database
Are you a startup fundraising atm? Fill out our “For Investors Eyes Only” sheet.
These investors write checks anywhere from angel and seed stage to $5-50 million.
Feedback from the startups - high quality inbound interest from investors.
Feedback from the investors - incredible depth and range of robotics startups.
If you’re an accredited investor and haven’t already signed up for the spreadsheet - Sign up here for great robotics deal flow!
Wayve raises $60 million in a Series D extension round with investment from AMD, Arm Holdings and Qualcomm Ventures.
Jane Street Capital invests $1 Billion in Coreweave, and commits to spending $6 Billion on the cloud platform.
Like Nvidia investing in OpenAI, do you detect a theme?
Investors as customers = guaranteed ROI
FoundryRobotics has raised a $19M seed round. The San Francisco startup is building robotic systems for manufacturing assembly.
Anvil Robotics has raised a $6.5m seed round led by hard-tech veterans Matter Venture Partners and Humba Ventures for a modular Physical AI platform, accessible to anyone on a grad student budget.
Build’s Egocentric-1M - reportedly the largest egocentric video dataset ever released: 10K hours of first-person video captured across 2k factory workers, totaling 1B+ frames.
Vincent Vanhoucke’s video series "AI in Motion" received 2 honorary mentions at the 2026 Webby Awards, including Best Interview or Talk Show in the branded entertainment category.
Robotics Events
AMBayArea Summit 2026
Association of Manufacturers Bay Area at Chabot Space & Science Center - April 21
Keynote Speaker Andra Keay on Humanoids, Humans, and the Workforce
Revolution No One Is Talking About
Physical AI Industry Night: San Francisco
April 23 - 5.30-8.30 - SF - presented by Foxglove and Eclipse
Join us for an invite-only evening for developers and engineering leaders building the future of robotics and physical Al in San Francisco.
Topic: From Automation to Autonomous Machines: The convergence between traditional industrial robotics and modern autonomy.
Panelists:
[Host] Adrian Macneil - Co-Founder & CEO, Foxglove
Behrad Toghi - AI & Robotics Lead, GM
Kevin Peterson - CTO, Bedrock Robotics
Pete Florance - Co-Founder & CEO, Generalist
Ruijie He - Founding Technical Member, Mind Robotics
Expect a casual atmosphere, great conversations, and a chance to connect with others pushing the frontier of real-world Al systems.
Robots on Ice 2026
April 25 at Sharks Ice San Jose
Get ready for one of the coolest tech-meets-sport events of the year! Robots on Ice 2026 brings together robotics, creativity, and skating for an unforgettable experience on the rink.
Join us for Robots on Ice 2026—where engineering meets the ice rink.
🤖⛸️
Are you looking to register a bot? Here’s the link:
Bots, Bevs & Devs
May 5 - 6pm to 8.30pm at Circuit Launch Oakland
Speakers and details TBA
RSVP: Bots, Bevs & DevsBots, Bevs & Devs
More Robotics Events:
Robotics Invest - Boston - May 26-27
ICRA 2026 - Vienna, Austria - Jun 01 - Jun 05
IEEE Hard Tech Venture Summit - Boston - June 10 - June 11
Automate - Chicago - June 22 - Jun 25
Deep Tech Week - SF - Jun 21 - Jun 25
Robot Fringe Festival - London - Jul 2
Robotics: Science and Systems - Sydney, Australia - Jul 13 - Jul 16
Open Sauce - San Francisco - Jul 17 - Jul 19
ROSCon Global - Toronto - Sep 22 - Sep 24
IROS 2026 - Pittsburgh - Sep 27 - Oct 1
IEEE Humanoids 2026 - Silicon Valley - Dec 6-9
The annual IEEE-RAS International Conference on Humanoid Robots is the internationally recognized prime event of the humanoid robotics community. The conference is sponsored by the IEEE Robotics and Automation Society and coordinated by the Technical Committee on Humanoids.
Send me your 2026 progress, plans or posts for potential inclusion in Robots&Startups. Or join the organizing team at Silicon Valley Robotics - bots&beer@svrobo.org
