- Tutor Intelligence has raised $75 million in a Series B round led by Andreessen Horowitz.
- The company rents out robots by the hour or per pick, and customers don’t need to make any upfront purchases.
- Almost one-third of customers who got robots in the past year came back for more.
Josh Gruenstein, who once caused elevator outages as a kid by welding in his apartment building’s basement, now leads a company that rents robots to factories by the hour. His startup, Tutor Intelligence, just raised a $75 million Series B led by Andreessen Horowitz.
Tutor Intelligence, based in Watertown, Massachusetts and spun out of MIT’s CSAIL, has raised $ 117 million in total. Union Square Ventures and Neo, which led the Series A and seed rounds, also joined in.
Two MIT tinkerers and a data problem
Gruenstein co-founded Tutor in 2021 with Alon Kosowsky-Sachs, whom he first met at MIT and once called the “guy with the DIY Segway robot” in the company’s Series A letter. In 2020, Gruenstein joined Kosowsky-Sachs in an MIT lab focused on building robot AI with deep learning.
While others saw robot learning as an algorithm problem, Gruenstein and Kosowsky-Sachs saw it as a data problem. Gruenstein pointed out that no one had collected robot data on the scale of computer vision or language datasets because it “would cost millions, if not billions.”
Their answer was to send out a fleet of robots to do paid work, letting each task create training data. To do this, they had to start a business. By December 2025, Tutor had over 60 employees. Its $34 million Series A, led by USV, brought total funding to $42 million.
Robots on the clock
Tutor lets businesses hire robot labor the same way they hire people: by the hour or by the pick. Customers don’t have to pay upfront or sign long contracts. The company handles deployment and maintenance and keeps the robots’ software up to date.
Tutor’s robots, Sonny and Cassie, work in big US manufacturing and logistics networks, handling things like breakfast cereals, snacks, candy, cosmetics, home goods, and toys. Remote operators, called “tutors,” give input to the robots, and every finished task becomes training data for Tutor’s foundation models.
This ongoing learning loop sets Tutor apart from competitors with larger funding rounds. Eka Robotics raised $150 million in a Series A at a $1.45 billion valuation on October 21, led by a16z and F-Prime, and trains its robots in simulation before moving skills to hardware. Skild AI raised $1.4 billion in January at a valuation over $14 billion, led by SoftBank, to build a single “omni-bodied” foundation model for many robot types. Apptronik raised $520 million in Series A in February for humanoid robots. While these companies focus on simulation, unified models, and humanoid hardware, Tutor trains its robots on real-world paid jobs.
“It’s not only that Tutor’s models let robots do jobs that robots have never been able to do before. This type of intelligence also makes robots affordable on a whole new level. We’re unlocking much broader access to the solutions that businesses need to survive and that make basic goods cheaper for everybody. And the demand is there. We can hardly build robots fast enough,” said Gruenstein, co-founder and CEO of Tutor Intelligence.
Recent data backs this up. Tutor’s robots now work in almost four times as many places as last year, and about a third of organizations that got robots in the past year came back for more. The company expects to have over 1,000 robots under contract by the end of 2027, all built and deployed in the US.
The new funding will help train Tutor’s advanced robot foundation model and support the production and deployment of Sonny and Cassie. By renting out robots, Tutor covers the cost of building each one instead of charging customers upfront.
Why a16z is backing deployment over simulation
a16z led Eka’s funding round 12 days before Tutor’s. While Eka tackles the missing-data problem with simulation, Tutor does the opposite, as an investor pointed out in the announcement.
“Tutor’s insight is that deployment isn’t just the destination for robotics research—it’s how you overcome its limitations. Rather than asking the real world to fit the research, Tutor starts with what real-world deployments demand and works backward through distribution, hardware and research. Continual learning across an expanding range of deployments is the path from specialized capabilities to general-purpose robotics,” said Raghu Raghuram, general partner at a16z and former CEO of VMware.
The firm has a track record of investing in hardware-focused companies. It also backed Northwood Space, which raised $375M in a Series C at a $2B valuation. Tutor’s flywheel builds toward a generally capable robot workforce and a safer, more affordable and more abundant world for the American consumer.
Tutor is joining a crowded market. Generalist, started by former DeepMind researchers, hit a $3 billion valuation with $200 million in funding. Germany’s RobCo became Europe’s newest robotics unicorn at a $1 billion valuation. Agility Robotics went public at a $2.5B valuation.
Mordor Intelligence predicts the warehouse robotics market will grow from $10.96B in 2026 to $24.55B by 2031, a 17.5% CAGR. Since Tutor also works in factories, this number is just part of its potential market.
The physical-AI field is still split on the best way to build robot intelligence: simulation, a universal model, or real-world deployment. a16z has backed two of these approaches in just 12 days. Tutor’s aim to build 1,000 robots by the end of 2027 will help show whether fleet data is most effective.
