A new report titled A Techno-Economic Assessment of Seabed Mining by Michael Barnard (TEAoSM) and Lyle Trytten, evaluating Impossible Metals’ deep-sea mining approach, attempts to model our system but relies on multiple incorrect assumptions, misapplied benchmarks, and calculation errors. These issues lead to conclusions that do not reflect our published data or current engineering progress.
In the coming weeks, we will publish a complete correction of all errors in Appendix 2 and Appendix 6 of the report, the sections of the document that address our technology. Below is a preliminary summary of this correction document. We value constructive dialogue and rigorous external review, and welcome opportunities to collaborate on a more accurate analysis. We hope that the authors of this document are willing to engage with us for a review of their document and our future economic analysis releases.
Preliminary Summary
At Impossible Metals, we publish our techno-economic analyses and unpolished technical progress because we believe that the responsible development of critical metals requires transparency and challenge from experts. Our goal is the same as most stakeholders in this space: to ensure that when seabed resources are collected, it is done with the best economics and with the lowest possible environmental impact.
Impossible Metals was founded with a mission to responsibly collect critical metals. This mission has led to our innovative approach, which has both economic and environmental focus. We selectively collect nodules to minimize the environmental impacts of habitat loss and plume generation, which are concerns with deep-sea mining. Additionally, our approach has robust economics with extremely low cost per tonne of material delivered to shore.
The recently published document by Michael Barnard and Lyle Trytten, dated August 2025 but released more recently, undertakes a deep dive into the techno economics of the Eureka Collection System concept of operations. We appreciate critical, well-informed engagement with our materials, as this benefits the industry and the planet. However, the document developed by these authors is ill-informed and contains many errors. In reviewing the document, six main types of errors are described below. Going forward, should these authors or others be interested in undertaking an analysis of our economics, we would be willing to engage with them directly to ensure that their critique of our plans is accurate. This collaborative approach will be most constructive for everyone as we make important decisions about responsible critical mineral collection.
In line with Impossible Metals’ transparency, we also encourage anyone engaging in public reviews of our materials to disclose any conflicts of interest. In the document, it is stated that “The authors have no financial or other interests in Impossible Metals, The Metals Company, or any competing ventures.” However, we believe this to be untrue. Lyle Trytten is a key advisor to Giga Metals, a large Nickel and Cobalt mine which will compete with Impossible Metals. The progression of our technology, along with its strong economic and environmental benefits, has the potential to be substantially disruptive to the traditional mining industry.
Types of Errors
1) Math Errors
Using incorrect assumptions that do not reflect our system, the authors state, “With 90 AUVs cycling batteries of ~200 kWh each, daily recharge demand approaches 400 MWh based on the energy projections performed above”. Given a 4-hour cycle time and assuming that the batteries are experiencing a full depth of discharge per dive (~200 kWh), also assuming a 90% charge efficiency, the total daily generation capacity required for one AUV is (0.2 MWh x (24h/4h) x (1 / 0.9) = 1.33 MWh. Assuming 90 AUVs, this would result in a daily recharge demand of 120 MWh, not approaching 400 MWh.
2) Architecture Errors
The authors present basic metrics and assumptions about how our fleet is operating that are inconsistent with our latest published documentation and demonstrate a lack of understanding of our system and our operations beyond a basic level. One example of the authors’ incorrect assumptions is the number of autonomous underwater vehicles in a full fleet. A full-scale Eureka IV fleet will operate with 314 vehicles, whereas the authors assumed it would operate 90 AUVs. A second incorrect assumption is the ship size, where the authors assume a Panamax-class vessel with a Deadweight Tonnage (DWT) of 65-80 kDWT. The ship size used in the latest published model is 144 kDTW, and our latest plan is a Newcastlemax-class vessel with a 200 kDWT capacity. There are numerous other errors like this, and there is no documentation indicating where the authors’ numbers for our system come from. You can find our latest published model at this link. An updated version will be published and presented toward the end of this year.
3) Inaccurate References
Possibly stemming from a lack of familiarity with marine technology, the authors contrast our technology with benchmarks that are not relevant in multiple instances. In numerous cases, this results in the authors assigning a component cost that exceeds the actual component cost for Impossible Metals, and in one case, the authors underestimate the component purchase costs. One example is the arms that the authors selected for comparison. Through their research, the authors found that “Commercial subsea manipulators such as Schilling’s Titan or Kraft’s Predator cost $150–250k each depending on depth rating.” While this is true, these are high-dexterity, seven-function manipulators designed for a high range of motion and a wide range of tasks. Our arms have substantially fewer actuators and are a very different style. The cost to us today for fabricating our arms is approximately $60k. Our TEA model assumes $30k, and we have vendor quotes confirming that this cost is achievable at production volumes. A second example is where the authors state that “industrial robotic pick-and-place systems in manufacturing rarely sustain <2 s cycles when operating continuously with environmental uncertainty.” However, the reference link used by the authors points readers to a serial-style robot arm optimized for dexterity rather than speed, so while this may be true for serial robots, it is not the case for our delta-style robotic arm, which achieves a 1.7-second cycle time today. Additional errors of this type in the document further highlight the authors’ potential lack of familiarity with robotics and offshore technology.
4) Mixing Industry Technology Readiness with Impossible Metals Technology Demonstration
At Impossible Metals, we are making incredible technological progress, but there is still work to be done to have a fully operational fleet of AUVs. When considering the technology readiness of our system, the authors often confuse the technology’s readiness with Impossible Metals’ demonstration state. For example, when discussing the technology readiness level (TRL) of our plans to spend 10% of CapEx on spares and maintenance, a TRL of 5 was indicated by the authors, implying that this is a technology in the prototype stage, where, in reality, the industry has achieved spending this much on annual maintenance upkeep. While we won’t be demonstrating this until we are in operation (as we can’t), this is not a risk, as the offshore industry has repeatedly achieved spending 10% of CapEx funds on spares and maintenance. It is questionable if TRLs apply for a maintenance regime, but if they do, the correct TRL is 9.
Another example is the “$40M ‘simple’ support vessel,” for which the authors indicate that retrofitting is at TRL 3-4. Since retrofitting a vessel with the required technologies has been achieved many times in the past, this is not a low TRL. The authors state that our “approach requires advancing a dozen technologies beyond the current state of the art”. However, a number of the technologies incorrectly listed as having a low TRL in the document do not represent a need to move the state of the art forward, but simply indicate that non-R&D engineering is required to implement our solution. Eureka II, which was successfully tested in the deep ocean last year, has achieved TRL 6.
5) Commercial Comparison
A manufacturer produces goods at a given cost and sells those goods at a price higher than cost. The difference between the price and the cost is called the margin. The authors confuse the distinction between cost and price by comparing the cost of production for our Eureka vehicles with the price of purchasing new underwater vehicles from their vendors. Typical margins for underwater technology (vehicles and sensors) range from 25% to 80%, meaning the product’s cost is between ¾ and ⅕ of the purchase price. Given the Eureka IV manufacturing cost of $3.4M and the range of margins typical for the industry, a reasonable purchase price for a new underwater vehicle is $4.5M to $17M. Based on the authors’ research, they found that the price for reference-class vehicles ranges from $3M to $20M. This suggests that, without the authors’ confusion of price with cost, the overall vehicle cost estimate from our economics is reasonable.
6) Selective or Incomplete Statements
In several places, the report’s narrative can leave a different impression than the underlying analysis. For example, the document states, “Time & energy balance fails at realistic pick rates”; however, in their analysis, the authors show that the energy balance passes; it is only the time that fails their analysis. “Battery energy is therefore sufficient; the limiting factor is mission duration, which exceeds IM’s 4-hour target.” Note that the math for both the batteries and the target duration is incorrect, suffering from errors of type 2 (Architecture Errors, where the wrong sizing for the batteries and power consumption is assumed) and type 3 (Inaccurate References, where the authors reference a vehicle with a very different buoyancy engine size and assume the speed of our vehicle will be equal to the speed of that vehicle). A second example is the statement “Despite automation, 40–60 crew are needed, with accommodations comparable to those on offshore service vessels.” In our model, we assume 57 crew, but the authors’ statement overlooks this. There are additional instances of this pattern, which, taken together, risk leading readers to more negative conclusions than the authors’ own calculations support.
There are multiple examples of all error types throughout the document, with only a selection highlighted in this summary. Some of the technical assumptions and benchmarks suggest stronger familiarity with terrestrial process industries and markets than with subsea robotics or offshore operations. For future assessments of this kind, we would recommend including contributors with deep experience in autonomous subsea systems and offshore project execution.
Appendix 6 of the document considers our published full-production v6.2 TEA values, and then applies adjustments to these costs based on an assumed reduction in productivity (derived from and based on the errors in Appendix A) and on scaling factors that are treated as linear in some cases and exponential in others. These scaling coefficients are not derived from the operation’s physics or our system architecture, but instead reflect the authors’ judgment about cost escalation. As a result, we view the specific numerical outcomes in Appendix 6 as illustrative rather than predictive: they demonstrate the general point that lower productivity or higher costs will reduce project economics, but they do not provide a reliable forecast for our system. That said, the general idea of showing economic sensitivity to both unit productivity and nodule value is useful. We plan to incorporate a more physics- and operations-based version of this type of sensitivity analysis into our Q1 2026 TEA update.
While we appreciate efforts to review and critique our approach, this document is too full of errors to provide any insights into the technical or economic viability of our solution. We strongly believe in our analysis of the economics of our system, but encourage more people to provide input and feedback to push us to be even better. For future projects that look into our work, we suggest two modifications to the approach. Engage with us to ensure that simple errors are not overlooked and that people qualified to speak on subsea and offshore technology are involved.
