Eos Energy Enterprises, Inc. (EOSEW) Q2 2026 Earnings Call Transcript

Good morning, and welcome to Eos Energy Enterprises Second Quarter 2026 Conference Call. As a reminder, today's call is being recorded, and your participation implies consent to such recording. Operator provided instructions. With that, I would like to turn the call over to Liz Higley, Head of Investor Relations. Thank you. You may begin.

Good morning, and welcome to Eos's Second Quarter 2026 Conference Call. Today, I'm joined by Eos's CEO, Joe Mastrangelo; COO, John Mahaz; and CFO, Alessandro Lagi. Today's call may include forward-looking statements, including our expectations regarding future results and the outlook for our company. These statements are based on our current expectations and assumptions and are subject to risks and uncertainties that could cause actual results to differ materially. For more information on these risks and uncertainties, please refer to our SEC filings. These forward-looking statements speak only as of today, and we undertake no obligation to update them, except as required by law. Today's remarks will also include references to non-GAAP financial measures. A reconciliation of these measures to the most directly comparable U.S. GAAP measure is included in our earnings release. Non-GAAP measures should be considered supplemental to and not a substitute for financial information prepared in accordance with GAAP. In addition, these measures may not be comparable to similarly titled measures used by other companies. This conference call will be available for replay via webcast through Eos Investor Relations website at investors.eose.com. Joe, John and Alessandro will walk you through our business outlook and financial results before we proceed to Q&A. With that, I'll now turn the call over to Eos's CEO, Joe Mastrangelo.

Thanks, Liz. Good morning. Thanks, everyone, for joining us. This quarter comes down to three simple things. We ship more product than we have in any prior quarter. We grew our backlog and we committed to consolidating our manufacturing footprint, a strategic decision that trades near-term revenue to lower our cost base as we exit 2026. Now let me walk you through all three of these. We're tightening our 2026 revenue outlook range to $300 million to $350 million. And this is a business decision, not an operating surprise. Let me address this change directly. We are accelerating the consolidation of operations into our modern Thorn Hill facility because of what it has begun to deliver. Line 1 will be down during the move and upgrade to the operational improvements we've implemented on Line 2. The volume that would have produced is the difference in the upper end of our guidance range. We're doing this so that 2027 is not only a volume growth year, but also a margin expansion year. John and Alessandro will take you through the operational and financial expression of the decision in a few moments. The low end of this range is roughly 2.5x our 2025 revenue and more than 19x in 2024. We delivered just under $126 million in the first half, which already exceeds all of last year's revenue. Let's frame the range itself. While we are still finalizing the detailed schedule, the shape is very clear. The second half exceeds the first half, the fourth quarter is higher than the third. The bottom of the range takes roughly $50 million of second-half growth over the first half. That is just maintaining the run rate that we exited June with on revenue already secured through backlog in Frontier Power USA. The top end of the range is achievable and it comes down to how quickly we scale Thorn Hill operations in a 24/7 production facility like we have today in Turtle Creek. We're planning for that, and we'll report against it every quarter. Moving to Slide 5, our second quarter operating highlights. We achieved record backlog, record revenue, record cube shipments and a significant improvement in adjusted EBITDA margin. We're starting to see the operating leverage we've been talking about. As volume increases, fixed costs are spread across more cubes, which drives margin improvement and closes the profitability gap. John and Alessandro will take you through the details behind those numbers and our path forward. But before they do, I'd like to spend a moment on fleet performance and cash. First, discharge energy. The number we're looking at on the page is up nearly 0.5 gigawatt-hour since our last call. The fleet now cumulatively has discharged 6.5 gigawatt-hours of energy and the Z3 fleet continues to perform, operating at an average round-trip efficiency of 78%. Let me be precise about that number because precision is what matters here. 78% is the average across a 20-100-20 state-of-charge window. It includes units running on DawnOS and units that have not yet been upgraded to DawnOS. So the performance is what we've developed, but we're continuing to count the performance where we still have to improve to show the true number of what customers are experiencing out in the field. It's a fleet average under real duty cycles, not a laboratory result on a single unit. We're starting to scale here. We have more work to do, but there is a clear path to continue to improve performance. Turning to cash. We ended the quarter with $364 million in total cash. What's important is what sits behind that number. Our operational cash use this quarter closely matched our adjusted EBITDA loss. Cash on cash, there was very little gap between the P&L and cash flow. That burn rate needs to continue to come down and turn positive. The initiatives that John and Alessandro will walk you through are designed to drive that improvement. Now let's move on to Slide 6. Let's focus on what wins our next order: reference hours. This page shows the hours the fleet has already delivered and the continued growth ahead with more than 200 additional megawatt-hours expected to come online over the next 6 months based on current customer project schedules. Let's start on the top left of the page. The fleet has now run over 3.9 million cycles and discharged 6.5 gigawatt-hours I talked about earlier. On Z3 specifically, over 1.1 million cycles. We're moving towards 1 gigawatt-hour of discharge energy. Every hour of cycling makes the next project easier to finance because customers can now evaluate a track record, not a promise. Round-trip efficiency tells a more interesting story. Note how the performance range is narrowing. The bottom is rising towards the fleet average. That variation is coming out of the system and reducing variation is what makes performance bankable. At the same time, the top of the range has crossed above 90%. In manufacturing terms, that is entitlement. It is what this technology delivers when everything runs as designed. It is not a ceiling we hope to reach. It is a level the fleet has already demonstrated, and now our work is to deliver it consistently across every cycle, every cube in the field. The duration tile shows the range our systems are operating in the field: from 2.5 hours to 14 hours, one product, one SKU dispatched however the market needs it. The photo on the right is a project that was added into our backlog in November of 2024. I want to use it to show you how a pipeline opportunity becomes an asset operating in the field. The units were built and shipped by November 2025. They went on foundations in May or June of this year, and the project is expected to come online by year-end. Order to operations is roughly 2 years and notice where the time went. The product was ready in 12 months. The second year was everything else from site readiness to third-party equipment delivery outside of our scope and the site construction schedule. That is one of the industry's key bottlenecks and it's exactly why Frontier Power USA was built to simplify the process and streamline the customer experience. The next page highlights how that strategy is translating into results. On Page 7, the U.S. storage market is changing in ways that favor our technology: load growth from data centers and electrification is pulling capacities forward faster than new generation can interconnect. In PJM, the grid operator for 65 million people and the largest power market in the country, prices have hit the ceiling in three consecutive capacity auctions. And the way the market now counts a resource towards capacity favors those that hold output to the system full system need rather than the first two hours of it. Virginia has written the same logic into law this spring. The statute carves out 4.5 gigawatts for resources that run 10 hours or more, inside its total state storage target above 20 gigawatts. At the minimum duration, that carve-out alone is 45 gigawatt-hours of energy. Buyers are no longer procuring just a storage system; they're procuring hours. Inside of this, we see four customer types: energy providers and regulated utilities who generate revenue from assets; energy consumers and assurance buyers who carry them as a cost of operations. The largest energy providers are independent power producers who need to deliver multi-hour and multi-cycle day after day because those capacity payments reward duration and energy margin rewards throughput. Utilities need assets that regulators will allow them to earn a return on over a 20-year life. Think about that for a moment: an energy provider, an IPP, uses that discharge window I talked about earlier. When we've always talked about the degradation of our product over time and having a 25-year life, that helps a utility with its regulator and its rate base. And if you move over to the largest energy consumer, that's high-speed computing, where power is just the cost of goods sold. Think of a data center as a factory and think of energy as an input for them to produce. So storage is judged on delivered costs, how fast the site can energize, and how reliably it will operate. The assurance segment is made up of defense or critical infrastructure customers, where storage is priced against the cost of failure and the rapidness of being able to perform. Two book it as revenue, one book it as cost of goods sold and one book it as insurance: all four buy hours and all four screen for supply chain origin. We manufacture in Pennsylvania with the domestic supply chain. That is a commercial advantage today, not a future one that we're planning on. The pipeline on this slide is built from all four of these customer types and the composition is where we are focused. I talked about backlog earlier. But what's important to note is that six customers placed orders this quarter, four new and two repeat. Our pipeline of $24.6 billion, nearly 112 gigawatt-hours, is up 31% year-over-year. Fifty-one percent of the pipeline is eight hours or longer. That is the duration band where our economics separate from incumbent technologies. Thirty-two percent is data center related, which two years ago was a de minimis amount. Three commercial developments framed the second half. After the quarter closed, we were awarded a strategic partnership agreement under the Golden Dome America program with the U.S. Department Award. During the quarter, we signed a 750-megawatt-hour master supply agreement with CAPAC covering Germany, Austria and Switzerland. And Frontier Power USA holds a 2-gigawatt-hour capacity reservation agreement. Under that agreement, we are now seeing purchase orders convert into projects, beginning with the Bimergen project and most recently with the $100 million purchase order we announced this morning for Phase 1 of the Blanquilla project in ERCOT originally developed by Stella Energy. I'll highlight the obvious with nearly $25 billion of pipeline against an $807 million backlog. Our job is conversion, not origination. Capital availability is one of the critical opportunity conversion factors. Two slides ago, I mentioned we built something to improve it, and that now brings me to Frontier Power on the next slide. Frontier Power USA is working as we intended. We have started execution on our first project because our priority is to get more projects into the field, begin generating returns and begin the operating references that help turn the investment flywheel. Frontier Power delivers pipeline conversion I was talking about a moment ago. We saw that strategy beginning to play out in the second quarter. A pre-existing project that will ultimately be part of Frontier Power USA was executed prior to the closing of the joint venture, using financing provided by a service affiliate. That project accounted for roughly 80% of second-quarter revenue. It demonstrates how this structure can help us get projects into the field sooner and build the reference hours that support future growth. Adding this project is an asset that we believe will deliver mid-teen returns and accrete the value of the joint venture in which we hold the minority interest. I'm putting that on the table first because I want you to understand it is a strategy rather than just a footnote. Our pipeline has historically experienced delays closing project financing, not technology acceptance. We saw qualified projects with real offtake sitting unbuilt because developers could not close their capital stack. So we've built the vehicle. Frontier Power USA supplies the capital, Eos supplies the technology, and we hold a minority interest in the entity. Walk the left side of the page, $263 million of gross proceeds initially raised, supporting an estimated $1 billion project deployment, the funnel behind it: 16 gigawatt-hours of opportunity pipeline, 5 gigawatt-hours acquired, selected or under active due diligence, and 1.8 gigawatt-hours under construction are approaching full notice to proceed. First projects under this vehicle are expected to be online by the third quarter of 2027. That is the project journey I showed you two slides ago, running at platform scale with capital waiting for projects instead of projects searching for capital. Now, in the middle of the page, because this is a long-term operating asset and it creates value in three ways. Frontier Power USA operates projects for recurring revenue. It can sell projects and recycle the capital with new ones and at scale, the platform itself becomes highly valuable. Eos participates in all three. We are the long-term service agreement counterparty across the small fleet with up to 25% to 30% of total CapEx over a 20-year life. We hold economic ownership in the platform so we share in the recurring cash flows, the project sale proceeds and any future monetization of the platform. In every project Frontier Power USA puts into operations, it adds reference hours to the installed base and to that chart I showed earlier, which will accelerate the next order and backlog growth and conversion of pipeline into orders, orders into assets operating in the field. APAC, the U.S. Department Award and the customers who placed orders this quarter are growing. Both engines are running, and they compound as we execute our strategy and projects become operational. So we have a strong demand signal. We are building installed-base operating hours as a capital partner that unlocks accelerated growth. Strong execution delivers profitable growth. And let me turn it over to the man responsible for all that, John, for an operational update.

Thanks, Joe, and good morning, everyone. Q2 was about focus, disciplined operation and increased efficiency. Turtle Creek delivered on all three. Cube output increased 20% sequentially, reaching an annualized production rate of approximately 1.5 gigawatt-hours in June. More importantly, we achieved that while keeping labor costs essentially flat. On materials, we're beginning to see the work we've been doing translate into lower cost. Material costs improved by 10% sequentially, with the benefit of tariff-free base paid in prior periods on imported components. Excluding that, material cost per cube improved 1% sequentially. We expect further improvement in the third quarter as inventory balances are worked through production, and we realize the benefits of our cost reduction initiatives. More broadly, there is a continuous learning cycle in our business where we take feedback from the field and incorporate those learnings into the design. While that can add cost in the short term, it ultimately drives meaningful cost reductions over time. When we launched DawnOS in the third quarter of 2025 material costs increased as we noted on our last earnings call. Since then, we have reduced material costs by 12.5% in less than a year. At the same time, we invested in product enhancements throughout 2026 based on the field learnings. Had those enhancements not been incorporated, material costs would have been down 14.5%. We achieved this despite elevated inflation, a volatile geopolitical environment and a continually evolving product design, which reflects the strength of our continuous improvement process. Labor productivity also improved during the quarter. Direct labor cost per cube declined 20% sequentially, while production increased, reflecting better execution and increasing efficiency across the factory. Manufacturing overhead per cube improved 4% sequentially. However, if you look at Turtle Creek on a stand-alone basis, overhead per cube improved approximately 16%, reflecting the productivity gains delivered by the team at the Turtle Creek plant. The consolidated result was temporarily impacted by the underutilization of Thorn Hill as we brought Line 2 into commercial production. That's exactly what we would expect at this stage of the ramp. As planned, we have been operating on one partial shift while we validate the line's performance. This phased approach allowed us to add labor incrementally as we ramp up that line. Over the next several months, we'll continue to add shifts and increase utilization. As volumes ramp, we expect utilization to improve, fixed costs to be absorbed across greater production and the operating leverage built into Thorn Hill to become increasingly evident in our results. While we're pleased with the progress at Turtle Creek, our focus is not simply on incremental improvements. Our focus is on achieving the cost structure we've always envisioned for the business. That's where Thorn Hill comes in. The biggest opportunity ahead of us is not just the continuation of what we've already accomplished, it is the earnings power we unlock as we fully utilize a purpose-built, highly automated manufacturing platform. During the first half of 2026, we produced 17% more cubes than we did in all of 2025, and we matched last year's total production volume in just 164 days. What's most important is that scrap dollars on that same volume were down 63%, validating that our manufacturing platform is scaling as planned. With Line 2 contributing only 1% of second-quarter production, we have yet to realize the full benefit of Thorn Hill, leaving significant operational upside ahead. Thorn Hill is already delivering the performance we intended. Initial Line 2 battery cycle times are 10% faster and bipolar cycle times are 11% faster than Line 1, with additional redundancies built in to improve line availability. That drives a near-term increase in overhead per cube and an improvement as we scale production, and we will continue to improve performance from here. As we move through the third quarter, we're evaluating the timing of consolidating Line 1 into Thorn Hill. There is never a perfect time to make a move like this. You have to balance execution, customer commitments and operational continuity. That said, after years of operating manufacturing facilities, I've learned that the sooner you pick a path, the sooner you begin realizing the benefits; waiting rarely creates value. Consolidating the footprint will allow us to upgrade Line 1 to the same single-piece flow design, while at the same time implementing redundancies to remove single points of failure. The focus becomes much clearer: one building, multiple production lines, one overhead structure and more volume going through the same footprint. Based on our current analysis, we believe this initiative alone could deliver an additional 10% to 15% reduction in conversion costs on top of the improvements already embedded in our current operating plan. Achieving those savings would require a modest investment to relocate and integrate Line 1 into Thorn Hill. But even after accounting for that investment, we currently estimate a payback period of approximately nine months. This is what positions us for 2027. It is the foundation for the margin improvement Alessandro will walk you through on the next page. Thanks, everyone. With that, I'll turn it over to Alessandro.

Thank you, John, and good morning, everyone. Before I start to discuss the quarter, let me say that it's a privilege to be here, and I want to thank the entire U.S. team for the company they have built and the progress that they've made over the last few years. I followed Eos for several years, first as a shareholder and now for the past two months as CFO. Over the last 25 years, I've led finance organizations across global energy and industrial businesses. What brought me here was the combination of a unique vision and differentiated technology, an expanding market and a business at an operational inflection point. Before I joined, I visited our manufacturing facilities. Having spent most of my career around industrial operations, the level of automation stood out. Eos designed this platform for the volume the business is growing into rather than the volume it was at, and that decision is now paying off. I strongly believe that Eos is at a real inflection point. Looking at what the team has built gives me tremendous confidence in the opportunities ahead. I believe I bring an operational mindset that complements the team with a particular focus on execution and margin expansion. I'm excited to be part of the next phase of growth and to help translate the scale we've built into stronger profitability and long-term shareholder value. I want to finally thank Nathan for his partnership through the transition and for the financial foundation he has established. With that, let me turn now to the second quarter. Revenue increased to its highest of $68.8 million, which is up 351% year-over-year and 21% sequentially, with cube deliveries increasing 207% year-over-year and 20% sequentially. Now turning to margins. Gross loss totaled $48.8 million. Margin improved 132 points year-over-year and 7 points sequentially. Excluding stock-based compensation and depreciation and amortization, adjusted gross loss was $42.9 million, and an adjusted gross margin of negative 62%. This marks our seventh consecutive quarter of gross margin improvement and reflects the operational progress we're making across the business. The second-quarter results also reflect the continued scaling of our manufacturing operations with some expected cost pressure as we invest in supporting that growth. In these regards, two items impacted the quarter. First, Thorn Hill. As you heard from John, a newly commissioned line operates below its long-term utilization targets, which weighed on fixed asset absorption. As throughput increases and the line matures, absorption improves. Second is field cost. Our installed base expanded, which drove higher deployment and commissioning activity, and we accelerated the DawnOS upgrades across a portion of the legacy fleet. The improved field data that Joe discussed earlier is directly related to this work. Both reflect investment in supporting a growing asset base rather than a structural increase in our cost profile. We flagged these pressures last quarter, and we continue to expect them to diminish significantly by the fourth quarter. Operating expenses totaled $35 million, increasing 6% year-over-year while remaining essentially flat compared to the first quarter. While revenue increased 351%, we reduced SG&A by 4% and increased R&D by 46% to invest strategically in the future software capabilities and product development. This clearly demonstrates the diligence around cost and cash management from the team. Net loss for the quarter was $276 million, with an adjusted EBITDA loss of $71.4 million, a margin of negative 104%, which is improving 235 points year-over-year and 16 points sequentially. Reported net loss continues to be driven primarily by noncash fair value adjustments related to our capital structure. Specifically, changes in our share price result in mark-to-market revaluation of warrants and derivative liabilities. As an example, when our share price increases, the value of certain warrants also increases, which can result in a higher account liability and a corresponding noncash expense. Those adjustments create volatility in reported earnings and do not reflect operating performance. Now turning to the balance sheet and cash flow. We are encouraged by the continued improvement in how operating cash flow tracked adjusted EBITDA during the quarter with almost 100% free cash flow conversion from operations. Working capital did not consume incremental cash even as revenue grew 21% sequentially, and we continue to invest in the Line 2 build-out at Thorn Hill. As a result, we ended the quarter with $364 million in cash. We remain focused on disciplined cash management, and we believe we are well positioned as we continue to improve margin. We are preparing the advance request for the second-year retranche and expect to close it by quarter end, subject to the conditions outlined in the loan agreement. Now let me close today's prepared remarks with the critical drivers to deliver positive adjusted gross margin. When you look at our history, product adjusted gross margin moved from approximately negative 983% in the second quarter of 2024 to negative 40% this quarter. That is more than 940 points of improvement in two years. Two years ago, we carried the cost of building a manufacturing platform well ahead of volume. We invested in automation, expanded the footprint, qualified suppliers and built an organization for the business we believe we could become. Over the last year, those investments began translating into performance. We increased production, improved yields, reduced manufacturing costs and benefited from supplier economics as volume grew. This quarter continued our progress. Some of the manufacturing gains were offset by the project execution investments I described earlier. Those were deliberate: we chose to strengthen our ability to execute as deployments scale, and that choice creates near-term margin pressure. The heavy lifting of building the platform is largely complete and now the work is leveraging it. That is what the right side of this page shows. There are four drivers of cost-out that we're pushing with detailed plans in place that John just walked us through. First, we anticipate a roughly 25 percentage-point reduction in material cost as a percentage of revenue. As backlog conversion becomes more predictable, we move from transactional purchasing to longer-term supply agreements. Additionally, the team is executing against more than 90 active cost reduction initiatives, focused on simplifying design, reducing material content and improving manufacturability. Second is conversion costs, approximately 20 points of reduction from the framework that John just walked us through. The step change comes from running under one cost structure. Supervision, planning, quality, maintenance and production support stay relatively constant, whether we run one line or four. Materials management is one of our largest labor costs today as we move product between floors, buildings and warehouses with label-intensive processes using temporary labor. At Thorn Hill, the majority of that movement is automated with conveyance and automated mobile robots. Third is project and field services, contributing another 20 points or so. As we said, we invested in the field this year through DawnOS upgrades while leveraging third-party resources. As that work is completed and we bring execution activities back to internal teams, project productivity improves and our reliance on external support declines. We are applying the same operational discipline we have established inside our manufacturing facilities. We actually view the field as a factory within our walls and lean principles apply. As projects become operational, we expect this cost to increase; however, we view this as an attractive opportunity over time and believe it can become a profitable area of the business as we continue to build and scale our internal capabilities. Finally, we expect approximately 8 points from continued yield improvements in our subassembly processes. We have already made solid progress here and as tooling and equipment upgrades are completed and tighter component tolerances are implemented, we expect to further reduce scrap and improve first-pass yield. So combined, these initiatives provide what we believe is a clear path to over 72 points of adjusted gross margin improvement over the next 12 months, assuming we execute our plan and achieve expected production volumes. We have demonstrated that we can build the capability. The next chapter is converting that capability into earnings, and I believe we have a clear plan to get there. As we continue the base cost control, we expect adjusted EBITDA to improve with increasing operating leverage, with execution and volume growth determining the pace of our improvement. While there is still work ahead, I'm confident in this team, I'm confident in the road map and excited about the opportunity to drive margin expansion. With that, I'll turn it back to the operator for questions.

Our first question comes from Christopher Souther with Truist.

So just to kind of unpack the updated revenue guidance and the path here: the low end is essentially 1.5 gigawatt-hours for the rest of the year, just Thorn Hill and the high end? Are we assuming that Line 1 comes back online at Thorn Hill and is producing as well?

Chris, so the lower end is basically continuing the run rate of June throughout the rest of the year to get to $300 million. The higher end of that is to get Thorn Hill into full 24/7 operation by the end of the fourth quarter, rather than specifically getting Line 1 moved and operating at Thorn Hill immediately.

Got it. Okay. So if Thorn Hill is just 1% of 2Q production, what kind of throughput are we seeing today? And how close are we to ramping that up towards the 1.5 gigawatt-hour rate we need for the low end there?

So we'll continue to run Line 1 and Line 1 is running really well, and John has actually got Line 1 to its nameplate performance. The team continues to bring Thorn Hill up into operations. But Thorn Hill operations right now are more about training and staffing the people to run the line. We'll follow the approach we learned at Turtle Creek: get one turn up and running, add a second shift, get that up and running and go from there versus trying to do it all at once. What John and the team have done with the second line is nothing short of phenomenal when you look at the results and how the line has been performing initially.

Got it. Okay. And then I appreciate all the gross margin walk drivers to get to the 10% gross margins by second quarter of next year. Can you provide a bit more detail on some of the material costs and project drivers? I think the conversion and scrap are pretty clear, but I'd love to get a better sense on what the cost-out initiatives are and the DawnOS and third-party labor drivers look like and how those progress over the next year?

So Chris, I think, and John and Alessandro can jump in, when you look at what the team is doing, having a clear revenue conversion plan now allows us to go out to suppliers and drive down costs, and we're seeing those costs come down. We talked about this in the presentation itself where in 2Q there was a little bit of timing, and in 3Q we're seeing costs come down as we get into July. John will keep driving that with the team on supplier cost-out. Then the second piece is part simplification: as we move through, simplifying DawnOS and how we run the firmware and hardware, taking cost out on that, and scaling up with suppliers. We start with suppliers who can move quickly to get us through prototype to initial production, and then John, with his relationships and background with contract manufacturers like Jabil, allows us to scale into a lower-cost solution to continue to drive that down. And then the third piece is labor: as we've ramped and installed more megawatts in the field, we started off with temporary labor to bring up Eos capability and then eventually phase out that temporary labor and have Eos personnel perform those roles. You see that in the seven quarters of improved margin that we've delivered. We're doing the same thing out in the field: starting with supervision and third-party help and then selectively transitioning to Eos employees to lower labor input cost for installation and commissioning.

Our next question comes from Stephen Gengaro with Stifel.

Two for me. The first one is, can you talk about the customer concentration that we see in the 2Q revenue and the backlog that you mentioned in the press release? How should we expect and what do you expect to see from a diversification of the customer base going forward?

Yes. When you look at 2Q, we had an opportunity to take a project that had a strong return profile and add it as an initial asset into Frontier Power, and we capitalized on that working with Cerberus, which was a great move for us. What we're trying to drive with Frontier Power is returns on the basis of individual project returns, and this project delivers that. At the same time, we create a pool of assets that can be monetized later and spin the flywheel. Owning a minority interest in the entity gives us the opportunity to build valuation around the platform. Over time, there will be a blend of projects executed through Frontier Power and projects sold into the market with third parties, and that blend will evolve. The important thing for us is having a baseline of backlog conversion that allows us to load the factory and sign longer-term supply agreements because we know what we're going to need to deliver. It also gives our commercial team the ability to sell slots in the factory to deliver revenue and accrete margin.

We have a question from the line of Stephen Gengaro.

Sorry, Joe, I was muted. Just a follow-up. In your pipeline of opportunities and the customers you're talking to, how should we expect that now 50% of your backlog is from that single entity to evolve? If we're sitting here 12 months from now, how should we think about that from a nonaffiliated entity in the backlog?

Steve, if Frontier Power is adding assets that deliver returns, that's great. What I'd like to see is less percentage change and more of the size of the pie growing. That's what we're focused on. Getting Frontier Power executing and getting projects referenced in the field will grow the other half of that pie too. We continue to be selective and execute transactions at arm's length. We're looking at projects that deliver returns and support long-term asset growth into Frontier Power. At the same time, there is a large pipeline that needs to convert and convert faster. I'm encouraged by the Germany, Austria and Switzerland master supply agreement, which lands real volume into Europe and allows us to expand there. I'm also encouraged by the U.S. Department strategic partnership award, which aligns with government energy consumption needs and our American-made manufacturing advantage. Having a 50/50 split between Frontier Power and third-party executed projects would mean Frontier Power is doing its job and we're growing a developer-like platform, and growing the pie from the $807 million backlog is the team's goal.

Great. And when you think about the mobilization of Line 1 over time and consolidating at Thorn Hill, two questions behind that. One is, are you doing it now versus waiting because of the timing of backlog and delivery obligations? Or is it because you've seen such higher efficiency out of Thorn Hill and it's critical to driving margin expansion?

Stephen, both factors matter. We want to hit January running. Turtle Creek is operating to nameplate, but consolidating simplifies operations and reduces complexity, and it positions us to scale next year. John's team achieved strong performance on Line 2 with redundancies learned from prior ramps, and you can see the difference in performance between Thorn Hill and Line 1. This is not about Turtle Creek not working; it's that Thorn Hill works better. We moved into Turtle Creek because it was what we could afford when we set up manufacturing. Thorn Hill gives a straight shot to improved efficiencies without the workarounds that Turtle Creek required. John has gotten the line running quickly, and while scaling any manufacturing operation has stumbles, we don't repeat the same mistake twice, and the results are showing that.

Great. And one quick one: how should we think about the profitability of Frontier Power USA? When does that business become profitable and how do you participate with your ownership?

Steve, remember this will be below-the-line profitability in our financials. Initial projects coming online in the second half of next year should begin to show returns and create other income as those assets operate. The timing will depend on project execution and when assets reach commercial operation.

Our next question comes from Joseph Osha with Guggenheim.

One of the things you talked about last year that we haven't heard as much about recently is the data center opportunity, and in particular some of the advantages you felt like you have in terms of the ability to cycle, the ability to locate close to the building and so forth. I'm just wondering if we might be able to get an update there. And I'm curious to the extent you're doing anything, what kind of durations you're seeing your customers ask for?

Joe, if you look at the pipeline, 32% of what's in it is data center related. We have discussed two segments within data centers. Segment one is co-locating with the data center. Segment two is having a storage asset in a generating area where data centers are going to be installed. We feel really good about the work with Mac and his team on projects in PJM and Pennsylvania that support demand from data centers. There is qualification work to do for locating directly on site, and we are working through those requirements. Our technology can meet power-quality needs including millisecond response times, and we've said repeatedly that we can perform on inference and fast response. It's a matter of working with suppliers and customers to reach firm contracts for co-siting energy storage alongside a data center. Both segments are moving and we will announce firm contracts as they close.

Okay. So it sounds like at this point it's more grid-level resilience for the near term, at least, than it is necessarily the on-site power quality, although that's evolving. Is that a correct way to think about it?

I wouldn't strictly characterize it that way. We're making progress in both areas. We're working through the timelines of PJM auctions and backstop mechanisms, and both approaches are moving well. I'm proud of the team's work and the customer conversations. Both on-site and grid-support opportunities are developing.

And on duration: one thing we hear is data center operators ask for rapid response but shorter durations of one to two hours. Are you seeing requests in the market for shorter-duration devices?

You see multiple cycles and shorter durations that add up to longer-duration discharge across the day. That plays to a strength of Eos: cycling multiple times per day without the thermal risks seen in other chemistries. There's a diversity of technologies required across the market and our technology serves a large set of use cases.

That concludes today's question-and-answer session. I'd like to turn the call back to Joe Mastrangelo for closing remarks.

Thanks, everyone, for listening today. We continue to make progress. One comment that came out: we are very clear on the goal of the company, which is to become EBITDA-profitable. Gross margin is a signpost and a journey to becoming profitable and generating free cash flow. That's what we'll keep everybody updated on as we move forward. It's the focus of John, Alessandro and the team. We continue to see strength on the commercial side, and we'll have to keep working through the opportunities in the pipeline, but we're really excited about the ability to create and accrete value for our shareholders through Frontier Power USA through multiple avenues: by building up the potential returns of Frontier Power itself, and by Frontier Power giving us the opportunity to better plan factory production to get assets into the field faster and get more reference hours. We're encouraged by the 6.5 gigawatt-hours we've discharged and how DawnOS is evolving and the performance we've seen. As we cycle, we learn from every cycle, we update our software and we get more performance out of the system. As I mentioned, the bottom end of our round-trip efficiency distribution is creeping up to the median, skewing the distribution higher. We can run cycles at entitlement of 91%. We run cycles as short as 2.5 hours and as long as 11 to 14 hours on Z3 technology. It's a flexible technology that can meet multiple use cases. We've got to keep our heads down, execute and make the company profitable, and that's what we're focused on as a leadership team. We'll keep everybody updated on the progress. Thanks for listening today.

This concludes today's conference call. Thank you for participating. You may now disconnect.