Cuộc họp công bố kết quả kinh doanh quý 3/2026 của Eos Energy Enterprises (EOSE): Doanh thu kỷ lục và tiến triển về biên lợi nhuận

Điểm nhấn chính

• Doanh thu quý 3/2026 đạt mức kỷ lục 82,7 triệu USD, tăng 171% so với cùng kỳ năm trước và 20% so với quý trước, nhờ lượng hàng cube xuất xưởng tăng 23% so với quý 2.
• Lỗ gộp điều chỉnh thu hẹp xuống còn 36,8 triệu USD, trong khi EBITDA điều chỉnh ở mức âm 63,8 triệu USD với biên lợi nhuận âm 77%. Ban lãnh đạo ghi nhận quý thứ tám liên tiếp biên lợi nhuận gộp được cải thiện.
• Đơn hàng tồn đọng kết thúc quý ở mức 765 triệu USD, tương đương khoảng 3,2 GWh. Con số này tăng 19% so với cùng kỳ năm trước nhưng giảm so với mức 807 triệu USD trong quý 2 do lượng hàng xuất xưởng vượt quá lượng đơn đặt hàng mới.
• Chi phí sản xuất giảm ở hầu hết các hạng mục chính: chi phí nguyên vật liệu trên mỗi cube giảm 9%, chi phí nhân công trực tiếp trên mỗi cube giảm 17%, và chi phí sản xuất chung trên mỗi cube giảm 8% so với quý trước.
• Eos tái khẳng định triển vọng doanh thu năm 2026 từ 300 triệu USD đến 325 triệu USD. Đối với quý 4, ban lãnh đạo dự kiến biên lợi nhuận gộp điều chỉnh ở mức âm 25% đến âm 20% và biên lợi nhuận sản phẩm ở mức âm 15% đến âm 10%.
• Tổng lượng tiền mặt đạt 344 triệu USD, bao gồm 271 triệu USD tiền mặt không bị hạn chế sử dụng. Ban lãnh đạo dự kiến chi tiêu vốn sẽ hạ nhiệt khi các khoản đầu tư lớn vào Dây chuyền 2 và Thorn Hill sắp hoàn tất.

Dữ liệu tài chính cốt lõi

Hiệu quả kinh doanh và hoạt động

Lượng hàng xuất xưởng tăng 23% so với quý trước, trong khi sản lượng pin tăng 25% và sản lượng cụm lắp ráp phụ tăng 26%. Quý 3 là quý trọn vẹn đầu tiên cả hai dây chuyền sản xuất đều vận hành thương mại.

Dây chuyền 2 đóng góp 17% sản lượng pin, tăng từ mức khoảng 1% trong quý 2, dù chỉ hoạt động theo lịch trình ca làm việc bán phần. Sản lượng của Dây chuyền 1 cũng tăng 6%. Thorn Hill sản xuất nhiều hơn khoảng 20% lượng pin mỗi ca so với Turtle Creek, nhờ thời gian chu kỳ nhanh hơn, giảm thời gian ngừng hoạt động của thiết bị và kiểm soát môi trường chặt chẽ hơn.

Eos đang hợp nhất sản xuất tại Thorn Hill để giảm chi phí chung và đơn giản hóa hoạt động vận hành. Thiết bị tại Turtle Creek đang được di dời và nâng cấp dựa trên các cải tiến được phát triển cho Dây chuyền 2. Công ty đã bắt đầu vận hành 24 giờ tại Thorn Hill vào tháng 10 và dự kiến lượng hàng xuất xưởng sẽ tiếp tục tăng trong quý 4.

Trên toàn bộ hệ thống đã lắp đặt, các hệ thống của Eos đã xả 6,9 GWh qua hơn 4 triệu chu kỳ. Đội pin Z3 đóng góp gần 1 GWh, với hiệu suất chu trình khứ hồi trọn vòng trung bình tăng lên 79% từ mức 78% trong quý trước.

Công ty cho biết hơn một nửa danh mục dự án tiềm năng có thời lượng lưu trữ từ tám giờ trở lên. Hoạt động gần đây bao gồm dự án 10 giờ tại Tây Virginia cho Google. Trong thử nghiệm được thiết kế để mô phỏng biến động phụ tải nhanh, hai hệ thống đã hoàn thành 147 chu kỳ liên tiếp trong hơn 735 giờ, duy trì hiệu suất chu trình khứ hồi 81% và không ghi nhận mức tăng nhiệt độ phát sinh.

Theo ban lãnh đạo, mọi hệ thống xuất xưởng trong quý đều sử dụng pin do Eos chế tạo và các cell pin chứa hơn 90% nguyên vật liệu từ Mỹ.

Định hướng của ban lãnh đạo

Eos tái khẳng định định hướng doanh thu cả năm 2026 từ 300 triệu USD đến 325 triệu USD. Ban lãnh đạo dự kiến giá bán trung bình trong quý 4 sẽ quay trở lại các mức của nửa đầu năm 2026 khi cơ cấu dự án bình thường hóa.

Công ty dự kiến biên lợi nhuận gộp điều chỉnh quý 4 sẽ cải thiện thêm 19 đến 24 điểm phần trăm, đạt mức âm 25% đến âm 20%. Biên lợi nhuận sản phẩm dự kiến sẽ đạt âm 15% đến âm 10%.

Mức cải thiện dự kiến sẽ đến từ việc gộp đơn mua vật liệu, tự động hóa Dây chuyền 2, cách bố trí nhà xưởng tại Thorn Hill, công suất xử lý cao hơn và việc loại bỏ các chi phí chung bị trùng lặp tại Turtle Creek. Ban lãnh đạo cũng kỳ vọng chi tiêu vốn sẽ bắt đầu giảm bớt khi các khoản đầu tư lớn vào Dây chuyền 2 và Thorn Hill hoàn tất.

Eos đã xác định các nguồn thanh khoản tiềm năng bổ sung gồm 20 triệu USD thuộc đợt 2 từ DOE, khoảng 24 triệu USD tiền giải ngân tài trợ từ bang Pennsylvania, và khoảng 39 triệu USD tín dụng thuế sản xuất ròng. Công ty có kế hoạch đưa ra định hướng năm 2027 trong cuộc gọi công bố kết quả kinh doanh quý 4.

Rủi ro và các yếu tố cần theo dõi

Đơn hàng tồn đọng giảm so với quý trước do lượng hàng xuất xưởng của Eos nhiều hơn lượng đơn hàng mới nhận. Ban lãnh đạo cho rằng đây là do tiến độ dự án chứ không phải nhu cầu suy yếu, nhưng thừa nhận rằng công ty phải chuyển đổi nhiều cơ hội tiềm năng hơn thành các đơn hàng chính thức.

Eos ghi nhận 3,5 GWh các cơ hội ở giai đoạn cuối, vượt quá lượng đơn hàng tồn đọng hiện tại. Hơn một nửa đang chờ thu xếp tài chính, trong khi phần còn lại đang trong quá trình đàm phán hợp đồng. Ban lãnh đạo lưu ý rằng không phải mọi cơ hội đều sẽ được chốt thành công.

Không có dự án nào của Eos được chọn trong đợt đầu tiên của chương trình lưu trữ quy mô lớn NYSERDA của bang New York. Ban lãnh đạo kỳ vọng một đợt khác sẽ diễn ra vào cuối năm tùy thuộc vào tiến độ chương trình, đồng thời đang theo đuổi các chương trình lưu trữ thời lượng dài ở các bang khác.

Công ty tiếp tục phát sinh các chi phí liên quan đến nâng cấp phần mềm điều khiển cho các hệ thống đã lắp đặt. Ban lãnh đạo dự kiến công việc này sẽ tiếp diễn trong 2 đến 3 quý tới. Việc hợp nhất sản xuất cũng tạo ra rủi ro thực thi khi thiết bị của Dây chuyền 1 được chuyển giao và tái vận hành tại Thorn Hill.

Mức độ tập trung khách hàng vẫn ở mức cao. Doanh thu từ các bên liên quan chiếm 83% sản lượng theo quý, dù ban lãnh đạo dự kiến tỷ lệ đó sẽ giảm và tiến gần hơn đến cơ cấu đơn hàng tồn đọng theo thời gian.

Nội dung nổi bật trong phần hỏi đáp với chuyên gia phân tích

Tiêu hao tiền mặt và thanh khoản: Ban lãnh đạo cho biết dòng tiền hoạt động vẫn bám sát EBITDA điều chỉnh và dự kiến hồ sơ tiền mặt sẽ cải thiện khi biên lợi nhuận mở rộng. Vốn lưu động về cơ bản đi ngang trong quý 3 nhờ các khoản đặt cọc của khách hàng và các đợt thanh toán theo tiến độ dự án. Ban lãnh đạo không đưa ra dự báo chi tiêu vốn cụ thể theo quý cho năm 2027.

Tiến trình biên lợi nhuận: Ban lãnh đạo cho biết việc giảm chi phí mang tính cấu trúc chứ không chỉ phụ thuộc vào sản lượng cao hơn. Môi trường vận hành tại Thorn Hill đã cải thiện thời gian chu kỳ của mô-đun pin khoảng 10%, trong khi việc tinh giản sản phẩm, giảm linh kiện và tự động hóa vẫn là các đòn bẩy tiết kiệm chi phí bổ sung.

Chi phí liên quan đến vận hành thực địa: Ban lãnh đạo cho biết chi tiêu thực địa hiện tại gắn liền với việc nâng cấp hệ thống điều khiển hơn là do yêu cầu bảo hành tăng. Dự phòng bảo hành tăng theo khối lượng xuất xưởng, nhưng việc sử dụng các khoản dự phòng đó không thay đổi đáng kể theo thời gian.

Frontier Power USA: Doanh nghiệp này đang phát triển danh mục cơ hội riêng đồng thời đánh giá các dự án bắt nguồn từ Eos. Đơn vị này đang thực hiện huy động vốn bằng nợ cho từng dự án và xây dựng các cấu trúc tài trợ danh mục đầu tư rộng lớn hơn.

Trung tâm dữ liệu và phụ tải công nghiệp: Ban lãnh đạo nhận thấy cơ hội sử dụng cùng một tài sản lưu trữ để vừa điều hòa phụ tải nhanh vừa xả năng lượng trong thời lượng dài. Tuy nhiên, họ nhấn mạnh rằng độ tin cậy nhất quán tại hiện trường và việc chuyển đổi danh mục tiềm năng giai đoạn cuối thành đơn hàng vẫn là điều thiết yếu.

Toàn văn biên bản cuộc gọi công bố kết quả kinh doanh

Toàn văn cuộc gọi công bố kết quả kinh doanh

Phần trình bày của ban lãnh đạo

Operator

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

Elizabeth Higley

Good morning, and welcome to Eos' Third Quarter 2026 Conference Call. Today, I'm joined by Eos' 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.

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 measures 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. Further, 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.eofb.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' CEO, Joe Mastrangelo.

Joseph Mastrangelo

Thanks, Liz, and good morning, everyone, and thanks for joining us today. Our team delivered a strong operating quarter. We delivered the most cubes in our history while continuing to drive product cost out. At the same time, our backlog came down from last quarter, and I'll come back to that on a later slide. There are 3 critical industry shifts that are shaping our performance. First, the Power business has gone from managing retirements to adding capacity as fast as it can. For 15 years, U.S. electricity demand barely grew. Today, grid planners expect 166 gigawatts of new demand by 2030. While data centers are more than half of it, manufacturers and commercial customers have the same challenge. They need power faster than the grid can deliver. .

Storage helps close that gap. Every grid connection at a gas plant, the solar farm, a substation, a steel mill or a factory has hours when it runs below its capacity. Storage puts that power to work faster than waiting for new generation or transmission. Second, where batteries made is now part of a project financing model under the foreign entity rules that took effect in January, a project's tax credit depends on how much of its cost traces back to prohibited foreign suppliers. And the cell is about half of that requirement. Every system we shipped this quarter, we used Eos made batteries and cells with more than 90% American Materials, that's what we mean when we say Americas battery.

Third, reliable performance besides who wins, and it has 2 key parts. First is delivery. A system delivered late can push a project past its tax credit deadline or delay system revenue. That's why we're consolidating operations into Thorn Hill, 1 factory running at scale at lower cost. And it's already working. Cost per cube came down again and margins improved for the eighth consecutive quarter. The second is performance in the field. Our system is a 20-plus year asset, and owners are using its flexibility to do more and meet new demand requirements. The same system might run a short cycle in the morning and a long discharge at night. Other systems today are built for 2 to 4 hours.

Longer, more flexible duty cycles need a different design, and that's where Eos competes. Our installed fleet runs from 2 hours to 14. We now have 12 sites operating in commissioning, up from 7 a year ago. More than half of our pipeline is 8 hours or longer, and recent deals include a new 10-hour project in West Virginia for Google. I've worked through several booms and busts in this industry. The winners were never the companies with the best technology on paper. They were the ones this equipment showed up on schedule, resolve issues and then ran year after year, day after day. Customers and shareholders are asking us for the same, be reliable.

Now let's look at fleet performance on the next page. Across our installed base, Eos systems have discharged 6.9 gigawatt hours over 4 million cycles, nearly a gigawatt hour of that from the Z3 fleet. Z3 round trip efficiency averaged 79% on a full cycle, up from 78% last quarter. And that includes systems not yet operating [indiscernible]. We didn't get here in a straight line, and the biggest lesson is at the bottom of this page. In our early systems, we controlled each cube as 1 unit. So a weak module set the performance of the whole cube.

We changed how we control the system from 1 control point per cube to 3 to 6 to 1.68. Donal -- last manages each of the 672 modules on its own, tracking its state of charge and state of health in real time. No single module holds back system performance and with the improvements that John will talk about in operations, modules going out have a much smaller deviation or distribution of how they perform. The industry standard is to plan for capacity losses and add batteries over time to recover.

We manage every module so the system holds performance, and we see a module that needs service before it impacts the system's output. And because we know exactly what's installed at every site, down to the module, down to the software version down to the firmware and improvement implemented in 1 site can be rolled out everywhere else in applies. The steepest learning curve for the company wasn't the battery itself. It was the system, the battery plus power electronics, the controls and the grid working as one. Our early Z3 projects took longer to commission than we planned, and we learned from each project and improved our fleet performance.

Now let's go and look at a specific customer use case. Here on Slide 6, we are showing what the Dan OS architecture makes possible. New loads coming on to the grid aren't just large. They're dynamic, and AI training cluster can swing tens of megawatts in under a second over and over and over again and our furnaces and large industrial processes do the same. I spent much of my career around turbines, and I know the risk. Swings like that where on turbine generator shafts over time. Regulators are seeing it, too.

FERC has directed NER to write mandatory standards for these loads by year-end, and grid operators are proposing ramp rate limits. Here, we are showing how an density core absorbs the swing within milliseconds and the grid sees a fraction of those swings and the utility sees a steady customer, not a volatile one. The questions that customers ask is what does that do to the battery itself. Fast swings create heat, and heat is what were traditional battery is out. We ran 2 intensity cores available for customers today, charging and discharging into each other on a fast fluctuation profile for 147 straight cycles for over 735 hours.

We saw no added temperature rise and the round trip efficiency held at 81% and in density runs the same module level controls we just covered on the previous slide. We're continuing to test and are using this data to show customers our operating flexibility. One asset can do both jobs, smooth the load milliseconds and shift energy over hours. For the customer, that's a faster path to power for the utility that becomes system reliability for the community, that's growth about higher electricity bills. That's what we mean by assets that meet today's grid requirements.

Turning to Slide 7. Let's go into some details on the backlog. We ended the quarter at $765 million, about 3.2 gigawatt hours. That's up 19% from a year ago and down from $807 million in Q2. We shipped more than we booked last quarter. This is project timing, not a fundamental change in demand. We're currently working 3.5 gigawatt hours of late-stage opportunity more than today's entire backlog. More than half is waiting on financing the rest is in contract negotiations. Our pipeline reflects where the market is moving, longer duration storage, growth across multiple end markets and stand-alone projects that will make existing energy infrastructure more efficient.

While Frontier Power U.S.A. is now about [indiscernible] of the backlog, we're also widening the customer base. We closed orders with 6 customers, 4 of them new last quarter. We don't expect every late-stage opportunity to close and the NYSERDA bulk storage program is a good example. one of several state programs we are pursuing. No EOS projects were selected in the first round that awarded less than 1/3 of the program's 3 gigawatt target. Those projects tend to reapply for the next round, which is expected by the end of this year. Focusing now on the middle of this slide, you see why the capabilities on the previous page are so important.

The biggest driver of storage economics isn't routes of efficiency. It's how many cycles a day can you run to generate revenue. adding just half a cycle per day reduces lifetime storage costs by approximately 10x more than a 5-point round trip efficiency improvement. On top of that, the intrinsically lower operating cost of an EOS density system further increases project returns. Policy at the state and federal level is making long duration a critical requirement with a growing number of states setting procurement targets with defined duration mandates.

California is eliciting up to 2 gigawatts of 12-hour and multi-day storage. While Massachusetts has launched a 5,000-megawatt energy storage program, for projects with durations of 4 hours or longer, including dedicated longer duration storage tranches. And in PJM, Virginia passed one of the largest storage mandates in the country this spring, calling for approximately 21 gigawatts a by 2045. The market is requiring American-made long-duration energy storage. We must turn more of our pipeline into backlog, and that's where Michelle and the team are focused.

With that, I'd like to turn it over to John to walk through operations, including our progress at consolidating into Thorn Hill.

John Mahaz

Thanks, Joe. The team delivered a strong quarter and everything we did operationally in Q3 moved us closer to profitability. We increased output, reduce unit costs and improve productivity across both Turtle Creek and Thorn Hill. That execution continues to strengthen our manufacturing platform and positions us well to add future capacity. A couple of key highlights. Shipments were up 23% from the second quarter battery production was up 25% and subassembly was up 26%. Now Line 2 contributes a big part to that. Q3 was our first quarter with both lines running commercially. Line 2 went from about 1% of our battery output in Q2 to 17% in Q3, and it did that on a partial shift schedule.

Line 1 increased 6% as well. That's productivity coming from the same asset base -- now here's a number I really like. On average, Stonehill is producing about 20% more batteries per shift than Turtle Creek. That comes from faster cycle times and less equipment downtime and it's structural. Is the facility layout and the equipment enhancements designed into Line 2, same shift, more batteries, more efficiency. Yield tells a similar story. Line 2 reached the same first pass yield as Line 1 in just 3 months. At Total Creek, it took 13 months for the bipolar line and 26 months for the battery line to reach that level.

There are further opportunities to improve and the accelerated ramp to those levels highlight the benefits of the improvements we've made and provides a blueprint for scaling future lines more efficiently. This is where the economics show up. higher output and better yields drive lower costs, and we saw improvement across every major cost line in the quarter. Our materials cost per cube came down 9% from Q2 that several supply chain initiatives taking hold and those savings are structural, not onetime. On labor, direct labor cost per cube declined 17%. That's our teams getting more productive at both sites with additional volume and on overhead. We reduced manufacturing overhead per Q by 8% through higher facility utilization and better fixed cost absorption. That's the operating leverage built into our manufacturing model.

The more we build, the less each cube cost us. Now the next big lever is consolidation. That move is already underway and on an accelerated time line. We began this assembly earlier in the quarter and equipment is moving now. running 1 site instead of 2 means lower overhead, lower conversion costs and simpler operations. Our manufacturing process is simple but precision matters. While we don't need clean rooms, Down Hills infrastructure provides much tighter control over air pressure, humidity and temperature all critical to process control. That means more consistent output, better yields and all of that contributes to our profitability road map.

We are using the relocation to incorporate Line 2 enhancements into Line 1. Regarding time line, the first bipolar lines from Turtle Creek are targeted to begin production in the fourth quarter. On the battery line, we're completing the upgrades we're seeing from Line 2 performance and targeting a full line returning to production near the end of the first quarter. Antelman, Line 2 will carry production. In mid-October, we transitioned 4 shifts to Turtle Creek personnel to Dorn Hill and began 24-hour operations. We feel really good about the progress we've made. We expect another step-up in shipments in Q4, remain on track to deliver the revenue outlook and remain focused on delivering profitable growth.

With that, I'll turn it over to Alessandro.

Alessandro Lagi

Thanks, John, and great to be back with you all this morning. The key points from Q3. Revenue hit a record margin posted its biggest improvement of the year and we absorbed that sequential growth with less operating cash flow usage. Revenue closed at $82.7 million. That's up 171% year-over-year and 20% sequentially. Volume drove Cube shipment rose 23% sequentially. Pricing was lower this quarter, and that was a project mix, not a market trend. As mix normalizes, we expect average selling prices to return to the first half 2026 level in Q4.

Overall, we delivered 210 projects across a diverse set of end markets and use cases. Related party revenue represented 83% of the quarter of the quarterly volume. That includes 2 projects that were originated by independent third-party developers and subsequently acquired by F USA. This number moves with delivery shadow and project size, and we expect this percentage to more closely align with a backlog mix over time. declining already in Q4.

Importantly, we are beginning to see that revenue growth is translating into better operating performance. Gas loss was $44.2 million with margin improvement 58 points year-over-year and 18 points sequentially. On an adjusted basis, gross loss was $36.8 million with 18 points of improvement in margin percentage. This marks our 8 consecutive quarters of gross margin improvement, and we expect that trend to continue.

Below the gross profit line, we continue to demonstrate cost management discipline. Sequentially, operating expenses were flat with SG&A and R&D overall at $34.7 million of which about 20% is noncash. So overall, revenue up 20%, 18 points improvement in adjusted gross margin and OpEx flat. That's the Eos operating model working. Now turning to the bottom line. Net income was $125 million. That's driven by noncash fair value adjustments related to a capital structure. This mark-to-market revaluation of warrants and debt are driven by changes in our stock price, not operating performance. The line we track here is adjusted EBITDA, which was negative $63.8 million, with margin improving to negative 77%. That's 27 points of improvement sequentially and a 96 points improvement year-over-year.

Two drivers there. First, Volume growth contributed approximately 16 points and 12 net of price. This shows how our business model can quickly create strong operating leverage as volume expands to the plant. Over the past several quarters, we have invested ahead of expected growth by building the team and capabilities required to support a significantly larger revenue base. As revenue continues to grow, we expect a greater portion of each incremental dollar of gross profit to flow through the adjusted EBITDA and cash flow. We also expect SG&A to become a smaller percentage of revenue over time clearing meaningful operating leverage throughout the organization.

Second, operational improvements contributed another 15 percentage points, including 12 points from product cost reduction and 3 points related to project execution. As John discussed earlier, cube costs improved again this quarter, and we expect that progress to continue in the coming quarters. The project execution benefit reflected the timing and mix of activity during the quarter. including our continued work to transition installed systems to downs and simplify operations for our customers, as Joe already discussed earlier.

We expect to see those benefits flow not only through profitability but also through cash flow and liquidity. Now turning to liquidity. We ended up the quarter with $344 million of total cash including $271 million of unrestricted cash. That's down $20 million sequentially. We used $106 million of net cash, offset in part by $87 million of DOE reimbursement received during the quarter. There are 2 key points here. First, we supported a meaningful increase in production and shipment without increasing quarterly operating cash consumption.

Operating cash flow tracked closely with our adjusted EBITDA loss again this quarter. Working capital was essentially flat, supported by customer deposits and project milestone payments and we collected another $11 million of cash by monetizing our remaining 2025 production tax credits. Given the pace of our revenue growth maintaining neutral working capital reflect disciplined execution and effective cash management across the business. Second, we are nearing completion of the major investment associated with Line 2 and the Port Hill facility. As those projects are completed, we expect CapEx to begin moderating. The way to think about cash is that a lot of the heavy lifting on line 2 and the tunnel facility build-out is already behind us and we still have meaningful sources of liquidity available to us.

$20 million under the DOE tranche 2, approximately $24 million of Pennsylvania grand reimbursement and roughly $39 million of net production tax credits. Those resources provide additional liquidity as we complete Line 2. Quarterly cash flow will continue to be influenced by production volumes, working capital and customer payment timing. But as Line 2 ramps, margins improved and capital spending moderates, we expect a stronger cash flow profile over time.

Now turning to our margin improvement. Since Q1, we have made steady progress. Adjusted gross margin improved 25 points and adjusted EBITDA improving by 42 points. We expect that momentum to accelerate in the fourth quarter with adjusted gross margin improving another 19 to 24 points to between negative 25% and negative 20%. Execution for the improvement is what we discussed already. A couple of key drivers support that outlook. First, we are seeing average selling prices move back towards first half of 2026 levels. That's a function of what's in macular today. Second, savings from bundled material purchases as volume continues to grow.

Third, Automation on line 2 continues to reduce labor cost per cube. Fourth, the optimized turn Hill layout improving through output and material flow. And finally, the location of Line 1, eliminating overhead costs from Charter Creek. That puts product margin in the range of negative 15% to negative 10% in Q4, and the profitability road map doesn't stop at year-end. There are additional initiatives to execute in the first part of 2027, along with the volume leverage and 2 lines in production at Tom Hill. Now on full year outlook. We are reaffirming 2026 revenue of $300 million to $325 million. As John discussed, accelerating the Line 1 transition cost us some revenue in the near term, but it also sets us a leaner, more productive footprint for 2027.

Currently, Line 2's performance gives us the confidence in that hole. We are encouraged by the progress we are seeing. Margins are improving, operating leverage is showing through and much of the heavy lithium on line 2 and terminals behind us. With that, thanks, everyone, for joining us, and I'll turn the call back to the operator for questions.

Operator

[Operator Instructions] Our first question is from Christopher Souther with Truist.

Phần hỏi đáp

Christopher Souther

Congrats on the progress, particularly on the margin front here. Can you -- I think we can kind of do the math around with that nice continued sequential improvement in 4Q for margins and kind of the OpEx like implied EBITDA and $50 million to $60 million loss for 4Q. Can you talk through any kind of cash deviation from those levels that we should think into [indiscernible] as we're continuing to ramp and then shifting the production over to Thorn Hill and starting to ramp that 1 up? Just anything from a CapEx perspective or working capital perspective that we should be watching over the next like 2 quarters?

Alessandro Lagi

Chris, it's Alessandro here, and -- thanks for the question. So couple of points here. First of all, I'm really pleased with the progress that we have done on margin. As you said, great third quarter, we see the continuation of margin expansion in the fourth quarter. When it comes to cash, it's pretty much the same story, right? We are pleased with the fact that our cash performance today have been tracking on the operating side, the EBITDA. So we have basically no absorption in the working capital. And we see for the future this trend to continue, right? If you think about our cash profile, it will go align with the EBITDA improvement and it will improve over time with the EBITDA until we get profitable.

We feel pretty comfortable around the position in cash that we have on our balance sheet. And you heard that we also have a few other sources of cash between grants and the DOE reimbursement. So we feel good about it until we get profitable.

Joseph Mastrangelo

And Chris, I would just add on the CapEx side, like I think John and the team are running to a predetermined plan of how we get there. So the only thing that would change any CapEx profile in the short term is going to be the timing of those near-term projects coming in and how that relates to how we lay out the volume for 2027 as we're working through here over the next 5 or 6 weeks. .

Christopher Souther

Got it. Okay. That's super helpful. And then maybe just on the Frontier Power USA team now kind of getting assembled. Can you talk through -- we're starting to see some of the stuff that's been in the backlog flow through there, obviously, but like where are they starting to kind of set up their development process kind of parallel to your pipeline or within your pipeline. Can you just speak to their process is starting to kind of get out into the market and look for opportunities that may be incremental or closing within kind of the pipeline would be great.

Joseph Mastrangelo

Yes. I think Chris, like you've seen -- Jay Bellos has been announced as the CEO of Frontier Power USA, which I think is a great addition to the team, not only from the standpoint of he brings his own contact list of customers that he can go out and has started to build pipeline and opportunities that we're working on closing, but at the same time, he's got decades of experience operating systems. So things we're talking about around software and Donal s and having somebody that's been a real-time asset operator is helping us improve ourselves, but like from the pipeline itself, we've got projects that we put in the pipeline from our opportunity set that they're looking at.

They've got their own pipeline. It's a nice list of opportunities. And at the same time, working through and securing the debt financing. And that's been progressing really well as far as packaging up the aerial brine insurance and going out and looking both at specific project financing and also broader portfolio of financing to be able to pursue projects. But we'll keep everyone updated as it progresses. But I really like the team that Jay is assembling underneath them. I think it's going to make us stronger in the market and stronger as Eos.

Operator

Our next question comes from the line of Stephen Gengaro with Stifel.

Stephen Gengaro

A couple for me, and just maybe back to the cash burn and balance sheet question. When you think about the guidance you've laid out for I think, 10% cash gross margins in the second quarter next year. If you get there, can you just talk -- what would cash burn a quarterly cash burn look like? I know it's not the easiest question to answer, but that is -- it's clearly on top of investors' minds. Just kind of curious your views on that and your views on your current liquidity and sort of the runway you currently have before more capital might be needed.

Joseph Mastrangelo

Yes. So Stephen, just -- I'll start off and then Alessandro can talk second half of the -- of your question. The first thing I would say is like I think Alexander coming in here for the last couple of months, like you've seen 2 quarters in a row where operating cash is tracking to the operating results that we have in the company. And actually, the growth rate we saw last quarter, along with the cash burn being slower than that growth I think that's the impact of him as our CFO and the experience that he brings to the job. And I would expect that model to continue.

And then I think as we get into laying out 2027. John has his CapEx plan. We've built all those things together to have the runway to get there. But Alessandro can talk more details. But what I would say on operating cash -- you've seen the impact of having an operating CFO, and I expect that to continue.

Alessandro Lagi

Thanks, Jo. And Steve, just to complement here, right? When you think about our cash usage today, there is an operating part that track as Joe said, and we expect that to actually improve quarter-over-quarter in the next few quarters until we get profitable. From a CapEx standpoint, we start to see in the fourth quarter the CapEx spend moderating down versus the earlier part of the year. And then in the coming quarters, we will have like even a lower run rate from the plant. So overall, we feel comfortable with the cash that we have in our balance sheet.

And again, the other sources of cash that we cannot happen and until we're going to get profitable. And in regards of what the guide for next year, obviously, we're going to use next quarter call in order to provide more clarity around those numbers.

Stephen Gengaro

Can you give us a sense of what a quarterly CapEx number would look like next year?

Joseph Mastrangelo

Stephen, let us come back. Let us come back in 4Q and give -- and lay that out. Again, like as I said to Chris's question, some of that quarterly like I don't want us to get ahead of ourselves on a number as we go through. We've got near-term deals here that we'll know a lot more on here in the next 4 to 5 weeks that are going to determine what that looks like. But I don't want to say anything and then have to come back and change that from there and wind up coming off a number of what we see today. .

Stephen Gengaro

Okay. And then just 1 follow-up. When you think about the current assets that are deployed and you think about your gross margin profitability, are there any costs that you're incurring related to assets in the field that need work from a warranty perspective or anything of that nature that has kind of weighed down the gross margin?

Joseph Mastrangelo

No. So not warranty, Stephen. I think like the page when I walked through the Dawn OS upgrades, there's costs associated with doing those upgrades that will come in over time. Like we talked about and when you look at now send mentioned this in his comments, there's product margin, which is approaching breakeven. There's projects that there is a spend that's occurring that is the upgrade of the OS that is going to be occurring over the next 2 to 3 quarters as we implement that -- those systems on the field.

Operator

Next question is with Sean Milligan from Needham & Co.

Sean Milligan

I guess around NYSERDA, not being included in that first bulk storage solicitation. I was just curious if you had any feedback there. And then kind of can you revisit the timing of the second solicitation and maybe project sizes that you might have in play in NYSERDA?

Joseph Mastrangelo

Yes, not ready to talk specific projects with NiceR, we're working through with the customers that submitted before and other opportunities from -- look, from a selection, they're the customer, they picked on where they wanted to start, it's less than 1/3 of the announced amounts. We feel like the projects that we submitted last time were good projects that met the overall goals of the program. From what we understand, the next round is going to happen by year-end, but obviously, that's subject to timing when you look at how the timing fell on Round 1. There was a lot of work that I certainly did to get to the projects that they selected, and we'll just continue to work with them as we move forward. .

Sean Milligan

Okay. And then you mentioned the state level management...

Joseph Mastrangelo

The other thing I would just add is like we talk about NYSERDA because we submitted and it was public information, but there's other states with programs that are larger than what NYSERDA has that are progressing with multiple customers that are in that number that we talked about of the late-stage opportunities, and we feel really good about where other -- how other stay -- how multiple states are looking at long duration storage and the role it plays with the energy demand that we're seeing. .

Sean Milligan

Okay. Yes, that was going to be my follow-up on the state level side. I know you've been working in terms of long-duration mandates there. You mentioned Virginia specifically. It seems like a pretty sizable opportunity. Like are there call-outs for the technology specifically or duration periods? Like what's attractive to that and kind of where you see momentum in other states?

Joseph Mastrangelo

Yes. So if you look at every one of them, right, they all are adding in a duration mix of wanting to have the 12-hour to 24 hour and then beyond multiple -- beyond multiple days, which would not be for our technology, it would be for other technologies. But everybody is looking at and understanding that to get the market to shift, you got to have the incentives and not just incentives from a price standpoint, but incentives from being able to derive revenue from the assets. And that's what a lot of these -- a lot of these programs are building into this that there's a return on the investment to the owner.

We feel really good at working with the states as they start shaping the policy and explaining to them and explain them that there is a battery that can perform in a system that can perform to those duties and it's made in America, with 90-plus percent raw materials coming from the U.S. And we feel like we meet those requirements. Now there's nothing in there that says specific Eos or specific zinc bromine, but when you have a -- when you have a program that targets a discharge time and how they want the system to operate, we feel like our technology meets up with that. We just got to work through the process. that each state and each customer has to get to an order.

Operator

Our final question comes from the line of Julian [indiscernible].

Unknown Analyst

It's Carter on for Julien. You provided a lot of color here, but wanted to try to unpack the margin progression a little bit more -- you saw some nice progression here in third quarter and expected to get in fourth quarter, I guess I'm thinking about $27 a little bit. It sounds like you need some clarity on where some of the deals in the pipeline land to get the exact cadence there. But if I think back to your 72 points laid out in the last call, I'm wondering across the 4 major buckets you laid out of that 72 points, if anything has materially changed or the margin progression is really just contingent upon where some of these large deals land into the first half of next year. .

Alessandro Lagi

So thanks for the question. So the process dam so far are very encouraging. Again, third quarter was very strong. I think if you look at the fourth quarter Combined with the third quarter, we are well over the 50% of PAC to get to the 72 points. So we feel very good about it. And when you think about -- when you double click on the margin expansion, John talked about it before the progress that we've done on tubes cost right? So material -- material went down in the third quarter. We expect the same trend coming in the fourth quarter. And then obviously, you have the conversion cost there.

With the increased volume that we're going to have in Tornio, we expect that conversion cost to go down significantly because of the better layout of the plant, the better efficiency, the better automation that we have versus start to treat -- so these are going to be the key markets. And obviously, we're going to -- Joe talked about it before as we finished to upgrade the control in the field. We're also going to see the fuel cost coming down. So it's the same pretty much in the same buckets that we talked about in the last call. And we see the same trend in the third quarter coming in the fourth quarter in 2027.

Joseph Mastrangelo

Yes. And Carter, the only thing I would add to Alessandro's comments to just kind of clarity around your question like -- it's not we need volume to get the cost out. They're structural cost out that we're getting. And I think John's comments, and I want to make sure that we double click on a portion of this is just the building infrastructure that we have in Thornhill allows the equipment to run better, and we're seeing a 10% improvement on cycle time of producing a battery module. And a lot of that just has to do with the stability of the operating environment that we have. That takes cost out in and of itself and downtime and a bunch of other things, right? I think volume on top of that volume is a big lever for us, and like we are going to work through on where that comes out and how that structurally changes the cost profile of the project. But what I would say, again, with Alessandro and John have built is a good old-fashioned cost out pipeline that they're executing on with Francis on engineering to just get -- simplify the product, remove parts from the system, make the system more robust and take cost out.

The one other thing kind of echoing Stephen's question before and what Alessandro talked about. I know Stephen asked about warranty costs. I think the one clarity on clarifying thing I would give here is like our warranty reserve goes up with volume, doesn't mean there's warranty cost underneath it. Just as volume goes up, you reserve for the event of the warranty, but if you look at the usage of that reserve, it hasn't changed over time.

Unknown Analyst

Right. That's definitely helpful. Just a last 1 for me. If you could provide some more color on just the M&A Google announcement in West Virginia. -- seems like an exciting transaction for you guys. Curious how you would characterize just general interest from Google or other hyperscalers. Can we see a similar kind of transactions going forward here?

Joseph Mastrangelo

Yes, Carter, the Google M&A, this is something we worked on for a while to get through the tech validations of multiple parties, both M&A and Google, we feel that having long duration is going to help the project and help the return to Google, I think, as we talk to more and more people, and I think this comment I made when I talked about like the impact of these dynamic loads. And we like to talk about AI, but like if I take an AI load and I put that up against the load that comes from a steel mill, the load that comes from any industrial factory and the spikes and peaks and valleys that it derives, they kind of all look the same.

So you need something that's going to be able to buffer those loads against the traditional the traditional power generation equipment. And I think that's where we come in. And I think what's important, and this is where the pipeline gets excited -- and I know we've been talking about this and the proof is going to be with orders closing. And I think out of that 3.5 gigawatt of near-term opportunities, we're going to see orders closing that meet this use case. But we go into people and we talk about the fact that we can handle those short-term spikes and give you long-duration discharge with the same asset and the same technology and people like that.

The part that we need to do is get the reliability where it needs to be out in the field and operate day by day by day and get to like Five9's capability on the reliability of the product, and that's another big piece of what John and Francis and Alessandro are doing that drives not only performance to the customer but cost out and profitability for Eos.

Operator

This concludes the question-and-answer session. I'd now like to turn it back to Joe for any closing remarks.

Joseph Mastrangelo

Thanks, and thanks, everyone, for listening. Again, I'm really proud of the work that John and his team and Francis have been doing to really get through get Thornhill up right the moving of the lines out of Foreign Hill it's somewhat emotional moment where I went to Turtle Creek about a week ago and where we had started Z3 production with semi-automated with a semi-automated line is now nothing is there because that equipment is now up at Foreign Hill installed and being commissioned to start operating here.

So the speed at which they're working and the efficiency of which they're doing this is impressive, and I really look forward to just 1 roof in a state-of-the-art building to start producing because what we see out of the Florange Hill line, it's not that the line is better. It's just that mine is operating in an environment that is conducive to better performance and having Alessandro here as an operating CFO, you're seeing the results and how our cash is tracking to profitability, and I think that's going to continue and improve over time.

And with Michelle being in our job here for the past couple of months and bringing a lot of what you're seeing on those state mandates tied to the work that she was doing before. Feel really good about what the -- her and the team are doing to position us to win deals. And we just got to put our nose to the grindstone and close and grow that backlog to show and continue to show that the technology is what the market needs for the future. So we'll keep everyone updated on that and look forward to the 4Q call, where we'll also lay out guidance for 2027 and how everything lands here as we get through year-end.

Operator

Thank you, Joe. We appreciate you participating in today's conference call with EOS. This concludes our presentation. You may now disconnect. Thank you.