Nidus Technologies, a New York startup, has raised a $30 million seed round led by Kleiner Perkins, the company announced on November 10, 2026. The round follows a pre-seed led by Costanoa. Nidus trains robotics AI on teleoperation data collected from its own rigs, aiming to automate manufacturing tasks in the defense-industrial sector, starting with wire harnesses. The company says it has booked more than $1 million with defense industrial customers in four months.
The round is a notable bet on a specific approach to robotics: rather than programming robots task by task, collect demonstrations from skilled human operators and train AI models to reproduce that work.
What Happened
Nidus Technologies announced on November 10, 2026 that it raised $30 million in seed funding. The round was led by Kleiner Perkins, one of the venture industry's best-known firms. It follows an earlier pre-seed round led by Costanoa.
The company's first target is wire harnesses — the bundles of wires and connectors that carry power and signals through vehicles, aircraft, and other complex machines. Wire harness assembly is famously labor-intensive: wires must be cut, stripped, crimped, routed, and tested in precise sequences, and much of the work is still done by hand. It is the kind of task that is difficult to automate with traditional fixed programming, because every harness is somewhat different and the work requires dexterity and judgment.
Nidus says it has booked more than $1 million with defense industrial customers in four months. For a seed-stage robotics company, early commercial bookings of that scale are significant, because robotics startups often spend years in development before earning meaningful revenue. The figure, as reported, suggests customers are paying for real work rather than funding experiments.
What It Is
Teleoperation is a method of controlling a robot remotely, with a human operator guiding its movements in real time. In the Nidus approach, skilled operators perform manufacturing tasks through teleoperated rigs — the company's own hardware — and every movement, grip, and adjustment is recorded as data. That data is then used to train AI models that can perform the same tasks autonomously.
The idea belongs to a broader shift in robotics often described as learning from demonstration. Traditional industrial robots are programmed with explicit instructions: move here, grip this, rotate that far. That works well when the task never changes, as on a car assembly line. It works poorly when the task varies, when parts are irregular, or when the work requires adapting to what the operator feels and sees. Demonstration-based learning aims to capture the judgment of a skilled worker, not just the motions of their hands.
Defense-industrial manufacturing is a demanding first market. Parts must meet strict specifications, production runs are often small and varied, and the workforce of skilled technicians is shrinking.
Why It Matters
Robotics has long promised to automate physical work, but progress has been uneven. Robots excel in structured environments and struggle in unstructured ones. The demonstration-learning approach is one of the most closely watched attempts to close that gap, because it borrows the most powerful tool in modern AI — training on large amounts of real-world data — and applies it to physical tasks.
The defense-industrial focus also matters. Defense manufacturing combines high willingness to pay, urgent demand, and tasks that are difficult to offshore or automate with conventional methods. It is a market where a startup can earn revenue while its technology is still improving, rather than waiting for a perfect product. That revenue, in turn, funds more data collection, which improves the models — a cycle that, if it works, compounds over time.
At the same time, the approach faces real questions. Demonstration data is expensive to collect, skilled operators are scarce, and a model trained on one task does not automatically generalize to another.
What This Means for Founders and Investors
For founders in robotics and applied AI, Nidus offers a template worth studying: pick a narrow, painful, well-funded problem; build the data pipeline yourself; and earn revenue early enough that customers fund your learning. The wire harness starting point is deliberately specific. A company that can automate one hard manufacturing task convincingly earns the credibility to attempt the next one.
For investors, the round reflects a shift in how robotics startups are evaluated. The question is no longer only whether the robot works in a demonstration, but whether the company has a data engine — a repeatable way to collect demonstrations, train models, and improve with each deployment. Early bookings matter as evidence that the engine is producing something customers value, though a single figure over four months is a beginning, not proof of scale.
The competitive field is crowded. Many robotics companies are pursuing demonstration learning, and large technology companies are investing heavily in the same ideas. Nidus will need to show that its rigs, its data, and its focus on defense manufacturing give it an edge that general-purpose approaches cannot easily match.
Angeled.com connects founders with investors.
Company announcement, November 10, 2026
Read Next
General Translation Raises $27 Million Series A Led by a16z to Build Agent-First Localization Infrastructure
LangChain Raises $300 Million Led by CapitalG at a $3 Billion Valuation to Build the Agent Stack
Mount Thor Raises $45 Million to Run Cloud Fleets of Macs for AI Agents
OpenAI Reportedly Raises $15 Billion at a $1.25 Trillion Valuation
Q-CTRL Raises $290 Million Series C Led by Salesforce Ventures, Reaching Unicorn Status
SE3 Raises $30 Million Series A Led by Accel



