Ultra Raises $50 Million to Rent Packing Robots by the Month

The warehouse robot that pays for itself will not look like a movie android. It will look like a compact two-armed workstation wedged between conveyor belts, packing padded mailers at 3 a.m. while its maker collects a monthly fee. That is the bet a Brooklyn startup just convinced investors to fund, to the tune of $50 million.

Ultra, a New York City-based robotics company founded in 2024, announced on October 9, 2026 that it has raised a $50 million Series A round led by Framework Ventures, with Y Combinator participating, according to reporting by Fortune's Term Sheet. The round brings Ultra's disclosed funding to $62 million, following an earlier $12 million seed round led by Y Combinator and NextView Ventures. The company plans to use the money to scale manufacturing, expand its robot fleet, and accelerate commercial deployment of its fulfillment automation across domestic logistics networks.

A service business, not a machine business

Ultra's commercial model is the real headline beneath the funding figure. Instead of selling expensive robots to warehouse operators, the company charges an upfront integration fee to install its machines, then bills a recurring monthly fee for hardware and software support. It is robotics as a service, and it removes the large upfront capital check that has killed many warehouse automation deals.

The economics of that choice cut both ways. A rental model lowers the barrier for customers and turns a one-time sale into a long revenue tail. But it also means Ultra carries the hardware risk: if robots break, underperform, or sit idle, the company absorbs the cost while the customer simply keeps paying for a service. Chief executive Jon Miller Schwartz, who founded the company alongside Max Friefeld, Oliver Ortleb, and Chaitton Parthasarthy, told Fortune that customer demand has been strong enough for Ultra to raise its prices, a signal that, so far, the risk has been worth carrying.

The machine at the center of the business is called Operator, or OP1. It is a stationary dual-arm robot built for a narrow set of jobs: packing, sorting, and kitting (bundling multiple items into a single package). It stands on locking caster wheels so it can be repositioned between workstations, occupies a five-by-five-foot footprint, handles payloads of up to 10 pounds per arm, and runs on a standard 120-volt outlet. Ultra reports deployments in New York, Georgia, New Jersey, and Texas.

Renting instead of selling removes the big upfront check that kills most warehouse automation deals.

The numbers on the floor

Ultra says its robots have packed more than half a million orders. One notable deployment sits at Highline Commerce in Brooklyn, where the robots reportedly handle up to 30 percent of the company's fulfillment volume. The setup is designed for speed: installation is measured in hours, not the weeks or months that a full automation overhaul can demand.

Revenue figures were not disclosed, so independent verification of the business remains out of reach. What is verifiable is that the company is operating in live fulfillment environments rather than demo stages, and that venture firms are pricing that reality at a substantial premium. Framework Ventures led the Series A, and Y Combinator doubled down from the seed round, a pattern that usually signals the investors have seen operational metrics they find credible, even if those metrics are not public.

The broader funding climate supports that reading. This week alone, robotics data-infrastructure startup Mecka raised a $60 million Series B and RobCo crossed a $1 billion valuation, according to industry funding roundups. Capital is moving into physical AI, the convergence of machine learning and physical machines, and Ultra's raise is one of the larger early-stage bets in warehouse automation this year.

The brains are rented too

Perhaps the most strategically interesting part of the announcement is what Ultra does not build: the AI. The company's robots run on models from Physical Intelligence, a separate AI company developing foundation models for robotics, and the funding round coincides with a deepening of that partnership. Ultra focuses on hardware, installation, and customer operations; Physical Intelligence supplies the robot's decision-making.

The relationship predates the raise. In accounts shared earlier this year, Ultra reported running a full shift at 96.4 percent autonomy using Physical Intelligence's pi0.6 model, with throughput and reliability improving across successive model generations. When the model encounters something it cannot handle, human operators intervene remotely. Those interventions keep order processing correct while generating training data for the next model generation, a loop that converts every failure into future capability.

The division of labor is deliberate and, for the industry, instructive. Building a robot and building its intelligence are two different hard problems; Ultra is betting that specialization wins. Physical Intelligence gets something it cannot manufacture in a lab: hours of real orders with real messy items, the kind of data that separates a promising model from a reliable one. Ultra gets to concentrate on the unglamorous work of deployment, uptime, and customer operations without having to win the AI race itself.

What this says about the robot race

The Ultra raise belongs to a larger pattern in robotics this year: the money is shifting from spectacular demos to deployable machines. Humanoid robots generate headlines, but the robots closest to profitability are stationary arms doing repetitive fulfillment work, mobile robots moving totes, and palletizers stacking boxes. Ultra's machine does not walk, does not look human, and does not need to. It packs boxes, reliably, for a monthly fee.

There are fair caveats. The 500,000-orders figure is the company's own count, the 96.4 percent autonomy figure describes one reported shift rather than a fleet-wide guarantee, and undisclosed revenue means the unit economics of the service model are still a private matter between Ultra and its investors. Scaling from a handful of warehouses to dozens will test whether the human-intervention loop and the support costs scale as gracefully as the hardware.

Still, the structure of the deal is a sign of where the industry thinks the near-term value is. Ultra is not selling autonomy; it is selling completed orders with a support contract attached. If the next wave of AI investment lands anywhere physical, as analysts now expect, a deliberately boring robot that ships boxes on a subscription may be exactly the kind of machine that pays back a $50 million bet.