Seattle agriculture-tech startup Carbon Robotics raised $20 million in new funding to support the creation of another piece of AI-powered machinery for farms.

With its signature LaserWeeder and relatively new Autonomous Tractor Kit (ATK) already being used by hundreds of customers, Carbon founder and CEO Paul Mikesell told GeekWire that “a brand new AI robot” is coming.

Mikesell said the machine, which is at least nine months away from being revealed, will leverage the same AI system used in Carbon’s other equipment but perform tasks beyond weeding.

“It’s very flexible, capable of doing a lot with the world around it, understanding what it’s seeing, what’s happening,” Mikesell said of Carbon’s system that uses an array of AI, computer vision and machine learning technology. “We see our ability to reinvest in that platform and double down on what it can do in some new activities.

“It’ll blow your mind,” he added.

Founded in 2018, Carbon Robotics made its name across ag-tech with the LaserWeeder, a machine which can be pulled behind a tractor and uses its tech to detect plants in fields and then target and eliminate weeds with lasers. The latest iteration, the LaserWeeder G2, was released in February.

In March, the company unveiled the Carbon ATK, previously called the AutoTractor. That autonomous platform is designed to fit on and control existing farm equipment and serve as an answer to labor shortages and increased productivity in farming.

Both platforms are continuing to grow and scale, and “things are moving really fast,” according to Mikesell, a longtime technologist and entrepreneur who previously co-founded data storage company Isilon Systems.

LaserWeeders are active on farms across the U.S. and in 14 countries around the world. Mikesell said revenue continues to grow every year, but Carbon is not yet profitable.

Ranked No. 9 on the GeekWire 200 list of top privately held startups based across the Pacific Northwest, Carbon has previously been backed by NVIDIA and Seattle-based Voyager Capital.

The Series D-2 extension round attracted Giant Ventures as lead investor. The UK-based VC invests across a variety of “purpose-driven” startups, and Mikesell said, “They got what we were trying to do.”

Giant previously invested in a $140 million round for Tidal Vision, a Bellingham, Wash.-based company turning discarded crab shells into a valuable industrial chemical called chitosan.

Beyond the secretive new machine, Carbon is revealing more about the “large plant model” at the heart of how it does computer vision through its AI systems.

Mikesell said the company is at the point where it has enough training data and labeled images that it can teach its AI to learn about the basic structure of the plants it’s seeing. This allows Carbon to run one model on every machine in the world.

“If new weeds pop up in an onion field in France, and those are eventually going to show up in a carrot field in the U.S., the first time we see that weed anywhere it can be part of the model and be ready to go,” Mikesell said. “It also means that if we want to go into a new crop that we’ve never seen before, we can do it immediately.”

A LaserWeeder is designed to target the meristem of a weed to kill it as quickly as possible and the large plant model helps it understand where to precisely target its zap.

Carbon Robotics, which has raised $177 million to date, now employs about 260 people. The company runs a manufacturing facility in Richland, Wash., and added another in the Netherlands to offset some trade and tariff issues as well as speed deployment of machines in Europe.

Mikesell said as far as competition, there are some companies in Europe who claim to be building some version of a LaserWeeder, but he’s never seen one in a field or competed against one.

“It’s very hard to create a LaserWeeder,” he said. “The targeting system is so special, and the AI is so special. It’s not just about detecting where the weeds are. The trick to making it work is you need a targeting camera to be able to keep the lasers on target [while moving], and everybody I’ve seen that says they’re gonna build a LaserWeeder doesn’t understand that concept.”