Impossible Metals wants to mine the deep sea one rock at a time

There is a strange fact at the center of the clean-energy transition: the metals we need to build batteries are lying loose on the floor of the deep ocean. Nickel, cobalt, copper and manganese, bound together in dark potato-sized rocks called polymetallic nodules, scattered across abyssal plains four kilometers down. The hard part was never finding them. It was picking them up without wrecking the ecosystem you are standing in. Impossible Metals, a company out of Y Combinator's Winter 2022 batch, decided the answer was a robot that works like a careful hand rather than a bulldozer.

Most approaches to deep-sea mining involve dragging a heavy collector across the seabed and vacuuming up everything in its path - nodules, sediment, and whatever animals happen to live there. Impossible Metals refused that design. Its autonomous underwater vehicle, the Eureka Collection System, does something harder to engineer: it hovers above the seafloor using a buoyancy engine, scans the ground with AI computer vision, identifies individual nodules, checks each one for visible marine life, and picks up only the rocks that are safe to take. One at a time.

01 / What it doesA robot that hovers, looks, and selects

The whole company runs on a single distinction: selective collection versus bulk dredging. A dredge is faster and simpler, but it kicks up sediment plumes that can drift for kilometers and it takes everything indiscriminately. Impossible Metals bet that the version regulators, scientists and the public will actually tolerate is the one that leaves most of the seabed untouched. That bet shaped every part of the machine.

How the Eureka Collection System harvests a nodule. Each rock is a decision, not a scoop.

02 / Why the seabedFour metals in a single rock

A polymetallic nodule is an unusually convenient object. It takes millions of years to form, growing atom by atom around a shell fragment or shark tooth, and it ends up containing four of the metals the battery economy depends on at once. That is why the deep sea keeps pulling in engineers and geopolitical strategists alike: the resource is concentrated, it sits on the surface rather than locked in rock, and it is outside the handful of countries that dominate today's supply.

Indicative share of the battery metals found together in a single nodule. One rock, four supply chains.

The customer is, ultimately, the battery. Impossible Metals is a business-to-business supplier: it plans to collect nodules with its own fleet of vehicles, process and refine them, and sell nickel, cobalt, copper and manganese into electric-vehicle and clean-energy supply chains that would rather not depend on Indonesian nickel or Congolese cobalt. In 2026 the company framed its Pittsburgh robotics hub explicitly around a "China-free" critical-mineral supply chain - a phrase that reads differently in a congressional hearing than it does in a pitch deck.

There is a second reason the seabed keeps drawing attention, and it is less about geology than about geography. Today's supply of battery metals is concentrated in a small number of jurisdictions, each with its own labor, environmental and political baggage. Land-based nickel and cobalt mining often means clearing rainforest, moving enormous volumes of overburden, and processing ore with a heavy carbon footprint. The nodule pitch is that the resource is already sitting on the surface, pre-concentrated, in international and domestic waters that sidestep some of those dependencies. The counter-argument - the one Impossible Metals has to answer - is that the deep ocean is one of the least understood ecosystems on Earth, and that "less bad than rainforest mining" is not the same as "harmless."

---

03 / The rivalryTwo philosophies, one seabed

The clearest way to understand Impossible Metals is to put it next to its loudest rival. The Metals Company, or TMC, is the industry's most visible name and has moved to mine international waters at scale, working with large collector vehicles. Impossible Metals took the opposite path on almost every axis - smaller vehicles, selective pickup, and a lease application inside US federal waters rather than the contested international commons.

	Impossible Metals	Conventional dredging
Method	Selective pickup, one nodule at a time	Bulk collection across the seabed
Contact	Hovers above the floor	Machine dragged along the floor
Marine life	Avoids nodules with visible life	Indiscriminate
Sediment plume	Minimized by design	Larger, can drift
Speed	Slower per pass	Faster per pass

The trade is explicit: Impossible Metals gives up raw throughput to buy regulatory and ecological headroom.

That trade is the entire company. Selectivity is slower and harder to build, which means fewer competitors are willing to attempt it. It also means the environmental argument and the engineering are the same thing - you cannot bolt "responsible" onto a dredge after the fact. Whether it pays at commercial scale is the open question, and it is the one investors, regulators and critics are all circling.

The industry backdrop makes the stakes sharper. Deep-sea mining has had a rough couple of years: a Norwegian firm once positioned to be the world's largest seabed operator filed for bankruptcy, and the international rules that govern mining in the high seas remain unsettled. In that environment, two survival strategies emerged. One is to press ahead into international waters and argue the rules out later. The other, Impossible Metals' route, is to stay inside a national regulatory framework, apply for a proper lease, and make the case in public that a lower-impact method deserves a permit. Both are bets on how the politics of the ocean resolve, and neither is guaranteed.

04 / The peopleFounders who had already exited hardware

The founding team did not need to do this. CEO Oliver Gunasekara spent years at ARM, where he helped grow the mobile-chip business from roughly nothing to dominant market share, then founded the cloud video-codec company NGCodec, which was acquired by Xilinx (now AMD). CTO Jason Gillham built and sold 2G Robotics (Voyis), an underwater laser-scanning company - which is to say he had already shipped robots that work at depth. Renee Grogan brought years of mining sustainability, permitting and stakeholder-engagement experience, and scientist Kenneth Nealson rounded out the founding group. The pattern in the team is people who build hardware for hard environments and pick problems with few competitors.

05 / Where it sitsBetween the lab, the lease office and the protest line

Few startups are attacked and courted at the same time. In 2025 Impossible Metals became the first company to apply for a critical-minerals lease in US federal waters, off American Samoa, under the Outer Continental Shelf Lands Act - and its CEO was invited to testify before the House Natural Resources Committee. In the same period, Greenpeace publicly called the lease bid desperate. Both reactions are signs of the same thing: the company sits on a genuine fault line in the world, where the demand for battery metals runs straight into the question of whether the deep ocean should be touched at all.

The technology has been climbing a ladder of proof. Eureka I demonstrated selective harvesting underwater in 2022. Eureka II completed deep-water testing in 2024. Eureka III, the full-size production vehicle, is the version meant to prove the economics - and the reason the company has been raising and building toward economic-scale readiness. In 2026 it chose Pittsburgh, a city that has reinvented itself from steel to robotics, for an Advanced Marine Robotics Hub.

Eureka roadmap · proof, one rung at a time

2022Eureka I - selective harvest proof of concept

2024Eureka II - deep-water test complete

2026Eureka III - production vehicle, scale readiness

The partnerships hint at how the eventual business could knit together. An arrangement with BGR, the German federal geoscience institute, gives access to a licensed contract area in the Clarion-Clipperton Zone for testing. A memorandum with metals-processing company Aqua Metals points at the downstream half of the problem - turning collected nodules into refined, sellable metal - and a 2026 memorandum with Deep Sea Minerals Corp. extends the resource side. None of these are revenue at scale, but together they sketch the shape of a company that wants to own collection, processing and supply rather than just build a robot and license it.

For now, Impossible Metals is pre-commercial. Its revenue is small and its biggest customers are still prospective. What it has is a clear thesis, a working robot that keeps getting bigger, partnerships with a German geoscience institute and metals processors, and a front-row seat in one of the most contested resource debates of the decade. The company's whole argument fits in a sentence: leave the ecosystem, take the metal. The next few years - the Eureka III build, the lease decision, the first real economics - will decide whether that is a viable business or a very well-engineered idea.