Asma Sharafi's career started on solid-state batteries. The work was the lithium metal anode, and the effort was to make a solid electrolyte carry it.
She now manages cell design at Cuberg, working on the same anode with a liquid electrolyte.
That reversal is the most interesting thing about the company, and Sharafi is willing to list the reasons for it rather than gesture at them. Solid state has trouble with material compatibility against different cathode materials. It has trouble at the other interface too, between the solid electrolyte and the lithium metal, where short circuits and dendrite penetration remain a risk. And then there is the problem she puts above the others, which is not a chemistry problem at all: getting from lab scale to commercialisation, where defect densities appear and manufacturing complexity multiplies.
Complexity, in her account, resolves into cost. Solid state asks for new equipment and a new supply chain, and both land in the price of the cell.
Cuberg's answer is a liquid electrolyte with a formulation specialised for lithium metal. The payoff is not a performance claim. It is that the company can use current lithium-ion equipment and processing, and draw on the process knowledge the industry has built over the past ten or twenty years.
The company was acquired by Northvolt in 2021 and is now around 200 people. Its stated focus is stabilising the product, keeping development on track against internal milestones, and working out how to reach large-scale manufacturing volumes using what its parent already knows about scaling.
Silicon has the larger volume problem
Several developers who set out to build solid-state cells have moved to a silicon anode instead. Sharafi does not treat that as a defeat for one side or a win for the other.
Both chemistries are pursued for the same reason, high energy density, and both are aimed at broadly the same customers: aviation, eVTOL and eCTOL aircraft, high-performance vehicles, with the wider EV market behind them. What differs is the problem each has to solve.
For lithium metal, the problem is safety.
For silicon, it is volume change, which Sharafi says is far larger than anything lithium metal goes through. That expansion has to be absorbed somewhere, which means module or pack integration has to accommodate it, and there are side reactions that can eat into cycle life on top.
Rizwan Dard, who works on corporate strategy at Cuberg, frames the comparison as a portfolio question rather than a contest. Electrifying the hard-to-electrify segments will take several types of solution, each with its own advantages and disadvantages. There is no animosity in it, he says, because the world needs a whole host of next-generation chemistries to hit its carbon targets.
A very good cell can still lose at system level
Cuberg builds modules and systems in-house rather than selling cells to integrators, and the argument for that is the most transferable thing in the conversation.
Lithium metal behaves differently enough from lithium-ion that several of the consequences never show up at cell level. Aviation regulation asks for propagation resistance and containment, which are system design questions. Stack pressure requirements are a system question. So is the volume change.
Sharafi's warning is specific. A very good high energy density cell can be designed in isolation, and then, in her phrase, "you're going to get hit with the integration factor" that brings the number back down once the cell sits in a system.
So Cuberg keeps cell design and system architecture close together, and changes each against the other. The cell is adjusted to suit what the system wants, and the system architecture is adjusted to suit what the cell does. Doing both in-house means the learning arrives faster than it would through a customer relationship.
The other half of that work is educational. Automotive and aviation customers have designed their modules and packs around lithium-ion, and Cuberg spends time showing them where a lithium metal cell diverges and what has to change to accommodate it.
Dard adds the commercial version of the same point. The system layer is where the company can optimise energy management and thermal management, and where services at pack and system level could eventually sit. The know-how matters less than what the know-how enables over a long contract.
Aviation is chasing the battery, not the aircraft
Electric aviation is the segment Dard talks about most, and his reason for it is that the industry's ceiling is the battery.
Aircraft developers are focused on power and gravimetric energy density because those determine how much payload the aircraft can carry and how far and how long it can fly. That is an unusually direct line between a cell property and a customer's business case, and Dard thinks it puts Cuberg in a strong position at both the cell and the system level.
The second target is high-performance automotive and motorsport, where the same properties convert into drivability, range and performance, and where weight savings and space consumption carry the same weight they do in an aircraft.
The cycle count came from someone else's lab
Asked where the product sits on the automotive scale of A-sample through D-sample, Dard declines to answer, noting that aviation stage-gates development differently anyway. What he will say is that the cell has moved through multiple generations of improvement and that module and system development is hitting both internal and customer milestones.
Then he offers something firmer, and the reasoning behind it is worth more than the number.
The battery industry, on his account, produces a great many claims about energy density, power and cycle life that nothing supports. Cuberg's response has been to buy third-party validation rather than assert. A cell-level validation report was published in 2022. Two more were in progress at the time of recording. A module validation report was close to release, tested by an external party, and in that test the module reached 692 cycles at what Dard describes as a very high energy density at pack level.
The figure is a module figure, not a cell figure, which is the point. It is the number that survives integration.
What 200 people get from a parent company
Both come back to Northvolt, and they arrive at it from different directions.
Sharafi's version is about constraint. A startup is resource constrained by definition, and a parent with infrastructure, expertise and people removes some of that, which compresses development time on exactly the things a small team finds hard. At around 200 people, Cuberg is developing a new technology, taking it to scale, listening to customers, adapting the technology quickly, and running cell-to-system integration and testing at the same time. All of that sits inside an infrastructure the company would otherwise have to build.
She points to recycling as an example of work already under way with Northvolt, and to the chance to feed into its advanced battery technology platform rather than only draw from it. Cost reduction runs the same way.
Dard's version is commercial. Northvolt has spent on renewable electricity, on domestic supply deals, and on both scaling and improving lithium-ion recycling. Cuberg does not simply have exposure to those capabilities, it has roadmaps to integrate against them: the equipment supply chain, the domestic supply base, and the route to improving the sustainability footprint of its own product without building the expertise twice.
Then the argument he makes to customers. A customer that partners with Cuberg is never only partnering with Cuberg. An aviation or automotive OEM is by default partnering with a large battery manufacturer in the Western hemisphere, with operations across Sweden, Montreal, Germany and now the United States.
Seen from the customer's side, Dard says, what is on offer is a long-term partner stable and bankable enough to survive the ups and downs of the macroeconomic environment.
This piece draws on the full conversation, which is available with a complete transcript on the episode page.