Bloom Energy (BE.US) Q3 Earnings Call: Surging AI Power Demand Brings Opportunities; Increased Capacity Expansion Investment, Yet Cash and Balance Sheet Remain Solid

Bloom Energy (BE.US) released its Q3 2026 financial results after the U.S. market close on Thursday. Data showed that the company's quarterly revenue reached $1.236 billion, up 138.1% year-over-year, setting a new quarterly record; GAAP net income attributable to common stockholders was $250 million, compared with a net loss of $23.09 million in the prior-year period, achieving a turnaround to profitability. On a per-share basis, Bloom Energy's Q3 GAAP diluted earnings per share were $0.78, compared with a loss of $0.10 per share in the prior-year period; non-GAAP adjusted earnings per share were $0.94, above the $0.15 in the prior-year period. On profitability, the company's Q3 GAAP gross margin reached 33.8%, up from 29.2% in the prior-year period; non-GAAP adjusted gross margin was 34.5%, up from 30.4% in the prior-year period. GAAP operating income reached $225 million, compared with only $7.8 million in the prior-year period; non-GAAP adjusted operating income reached $286 million, far above the $46.2 million in the prior-year period. Bloom Energy's substantial earnings growth reflects continuously expanding market demand for its on-site power generation solutions. With the rapid growth of power demand from artificial intelligence (AI) data centers, enterprises are placing increasing importance on reliable, scalable, and rapidly deployable power supply solutions, creating growth opportunities for the company's solid oxide fuel cell business. Looking ahead to full-year 2026, Bloom Energy raised its revenue guidance to $4.4 billion to $4.6 billion from the previous $3.9 billion to $4.2 billion, with the midpoint of the new guidance representing approximately 122% growth over full-year 2025 revenue. At the same time, the company raised its full-year non-GAAP operating income guidance to $960 million to $1.03 billion from the previous $800 million to $900 million, and raised its non-GAAP adjusted earnings per share guidance to $3.10 to $3.40 from $2.55 to $2.85. The full-year non-GAAP gross margin target remains at approximately 34%. Bloom Energy stated that the company will continue to invest in capacity and business expansion to meet growing market demand. As AI infrastructure buildout drives rising power demand, the company expects its on-site power generation technology to play an increasingly important role in enterprise energy supply. The core view of Bloom Energy management is: the surge in power demand brought by AI is driving on-site power generation from a supplementary solution to an important power infrastructure choice, and Bloom hopes to become a core beneficiary of the industry's architectural transformation. The CEO mainly elaborated on the company's long-term strategy and market opportunities: traditional power grids struggle to quickly meet the growing power demand of AI data centers and other industries. Bloom's fuel cell on-site power supply solution, with its deployment speed, reliability, and environmental advantages, is expected to expand coverage among hyperscalers, commercial and industrial enterprises, utilities, critical infrastructure, and international markets. CCS, 800VDC power supply, and combined heat and power provide the company with possibilities for further market expansion. The CFO demonstrated from a financial perspective that growth is translating into profitability: Q3 revenue grew 138% year-over-year, operating income grew more than 500% year-over-year, and operating margin rose to 23.1%; the company raised its full-year guidance for the third time while continuing to invest in capacity expansion, and maintained positive free cash flow year-to-date and $2.5 billion in cash reserves. Key variables worth watching going forward include: whether backlog can continue to convert into actual deliveries and revenue, whether capacity expansion can keep pace with customer demand, whether margins can continue to improve as business scale expands, whether working capital and cash flow can remain stable, and whether new business opportunities such as CCS, 800VDC, and utility partnerships can gradually convert into actual orders and revenue. The following are the specific contents of Bloom Energy's Q3 earnings call. I. Executive Remarks 1. K. Sridhar Co-Founder, Chief Executive Officer and Chairman Hello everyone. At the end of September, as we once again set a quarterly performance record, I celebrated Bloom's 25th anniversary and the milestone of the company's inclusion in the S&P 500. At that time, I rang the opening bell at the New York Stock Exchange together with many manufacturing team members. Over the past quarter century, our mission has remained unchanged: to provide clean, reliable, accessible, and affordable power for everyone. After all, in modern society, electricity is essential to economic and physical health, human dignity, and a happy life. The last time Bloom rang the opening bell at the New York Stock Exchange was 8 years ago, to celebrate the company's initial public offering (IPO). At that time, the market viewed us as a niche power provider. Today, the market views us as the standard solution in on-site power generation and a key force driving the AI revolution. I am confident that when we look back on today 8 years from now, we will recognize that Bloom led the first major architectural transformation in the power industry's 150-year history. Fuel cell-based on-site power generation will become the preferred choice for power supply across industries and regions and an important complement to the grid. Countries with inadequate grid infrastructure will leap directly to distributed on-site power generation, just as some developing countries skipped landlines and adopted mobile phones directly. Hyperscalers are the earliest customer group to embrace on-site power generation. They need to obtain large-scale power supply quickly, and some individual data centers consume more electricity than some small states combined. However, the grid was designed to supply power evenly to a large number of users spread across a vast area. Faced with the highly concentrated and massive power demand of data centers, the grid is already overwhelmed. Mandatory cost-plus rate design and utility franchise regulations further exacerbate this problem. These systems were not designed to incentivize regulated utilities to innovate or improve efficiency. In the 20 years before the rise of AI, power demand remained relatively flat, leading to underinvestment in power infrastructure and leaving many urgently needed upgrades delayed. The speed and scale of the AI revolution have fully exposed the grid's shortcomings. Hyperscalers rushed to find any available grid power and quickly realized that deploying on-site power themselves was the way of the future. Bloom quickly became the standard solution. Our solution is friendlier to surrounding communities and safer. A redundant, fault-tolerant, and distributed architecture provides higher availability, ensures continuous power supply, and enhances the system's resilience against natural disasters, cyberattacks, and physical attacks. At the same time, our solution requires no backup generators. We have advantages in total cost of ownership, safety, power access, deployment speed, and sustainability, and we have a clear path to net zero emissions. Today, the same factors driving hyperscalers to adopt on-site power generation are driving broader segments of the power market toward on-site power generation. According to our recent power survey of more than 200 commercial and industrial enterprises across 15 different U.S. segments, by 2030, one quarter of all U.S. commercial and industrial facilities are expected to rely entirely on on-site power generation; by 2035, that proportion is expected to rise to 35%. Why is this trend emerging? In the future, electrification, digitalization, and population growth will drive factories and enterprises to consume large amounts of electricity while requiring extremely high reliability of power supply. In fact, the International Energy Agency (IEA) expects global power demand to grow 40% by 2035. Existing grids are ill-equipped either to rapidly scale supply or to meet concentrated loads with high reliability; the difficulties they face today are much the same as their difficulty in meeting the power needs of AI factories. Bloom will become the preferred solution in this vast market. We will become the leader in enabling the power market to meet new global demand. Over the next decade, the global power market is expected to expand from approximately $3 trillion currently to approximately $6 trillion, growing continuously each year. So, 25 years after our founding, our journey has only just begun. Today, our growth is accelerating. The market has a deeper understanding of the value we create, and the Bloom brand has gained broad recognition. On our Q1 2026 earnings call, we expected full-year revenue growth of 60% year-over-year. Today, we have raised that guidance and expect full-year revenue growth of more than 120% at the midpoint of the guidance range. Our customer project pipeline continues to grow, and backlog continues to expand faster than revenue; at the same time, our customer base, the industries we serve, and the geographic regions we cover are becoming increasingly diversified. We are delivering on our commitments to customers. Just 15 months ago, we announced our first project with Oracle and delivered power to its Utah site in less than two months. Today, we are achieving the same goal on the Nebius project in New Jersey. From signing the contract to obtaining planning commission approval for the fuel cell project, it took us less than three months. Two months after approval, the first power modules were ready for customer use. Our business foundation in this important segment is solid and growing rapidly. Community members and policymakers recognize that our solution is clean, quiet, and the best choice for providing stable on-site power. For example, in Virginia's 2026 energy plan, Governor Abigail Spanberger explicitly mentioned non-combustion fuel cells and said and I quote her: "Compared with traditional gas-fired generation, these technologies produce significantly lower local emissions, thereby reducing impacts on the health of nearby communities." This is significant because Virginia is not an ordinary jurisdiction. Most of the world's data centers are located in that state. At the same time, Virginia is also at the center of tensions between data centers and surrounding communities. At a critical moment for both the data center industry and the communities where it operates, we appreciate Governor Spanberger's recognition of the value of Bloom's non-combustion solution. The following are five business developments this quarter that demonstrate how we are expanding our addressable market. First, the U.S. Army has conditionally granted our partner Indelible a long-term lease allowing it to build and operate a commercial power generation project at a U.S. Army facility using Bloom Energy Server. The project aims to enhance power security using our distributed on-site power architecture. We believe that providing secure and reliable power for military and civilian critical infrastructure, such as military bases, water treatment plants, hospitals, and airports, is an important growth opportunity. Second, AEP disclosed that it will deploy 700 megawatts of Bloom Energy Server equipped with carbon capture and storage (CCS) capability to provide Google with near-net-zero emissions power. This will be a groundbreaking project. Unlike turbines and engines, Bloom's technology makes large-scale carbon capture and storage feasible. We believe CCS is critical to achieving decarbonization goals and could represent a $1 trillion market opportunity for Bloom. Third, as we have previously shared with you, we are expanding our presence in the Rocky Mountain region, the Midwest, and the Great Lakes region. Today, even in U.S. states with the lowest electricity prices, our value proposition is already competitive. For example, in Texas, we will deliver our first equipment this quarter. Fourth, we view the utility partnerships we are building as an important channel to help the United States continue moving forward amid power shortages. A growing number of utilities recognize that deploying clean baseload generation assets such as Bloom Energy Server into their distribution networks is a faster and more economical way to meet customers' growing power demand while enhancing the stability and reliability of the distribution network. I expect significant progress in utility-related business, often referred to as front-of-the-meter arrangements. Finally, fifth, we see strong interest in many international markets in deploying Bloom systems to meet rapidly growing power demand. We are in discussions with potential local partners in some of these markets. We look forward to announcing some new partnerships to you soon. In summary, Bloom is the preferred solution in community-friendly on-site power generation, capable of rapid deployment while being clean, reliable, and competitive. Our on-site power solutions have been applied in multiple scenarios, including powering gigawatt-scale AI factories, megawatt-scale inference data centers, factories, commercial enterprises, and mission-critical infrastructure such as hospitals, telecommunications hubs, and essential public services. Demand for Bloom's on-site power business is growing both domestically in the United States and internationally, and we are building strong partnerships. Combined with future growth opportunities such as CCS and 800-volt DC power supply technology, this growth potential means Bloom is well positioned to capture a substantial share of the expanding power market. This judgment gives us greater clarity on our future direction. To seize this once-in-a-century opportunity, we must scale and execute efficiently. I have full confidence in the team and know they are ready to meet this challenge. Next, I will hand the call over to Simon to present the specific financial data. 2. Simon Edwards Chief Financial Officer Thank you, KR, and good afternoon everyone. As a reminder, I will focus today on non-GAAP adjusted financial metrics. Reconciliations between GAAP and non-GAAP metrics are included in our earnings press release and supplemental materials. Bloom once again set performance records in the third quarter. Revenue reached $1.24 billion, up 138% year-over-year; operating income reached $286 million, growing even faster at more than 500% year-over-year. With this quarter's strong performance, we have raised full-year revenue, operating income, and earnings per share guidance for the third time this year. Before diving into this quarter's details, let's review what we have achieved year-to-date. Year-to-date revenue has reached $3.05 billion, 50% higher than our full-year 2025 revenue. Operating income in the third quarter alone has already exceeded last year's full-year operating income. This shows that the operating leverage embedded in our business model is now manifesting in our financial results. Operating margin expanded to 23.1% this quarter from 8.9% in the prior-year period. Revenue grew 138% over the same period, while operating expenses grew only 26%. As KR said, to seize this once-in-a-century opportunity, we must both scale and execute efficiently. We will continue to invest in our innovation and engineering R&D roadmap, growth initiatives, and the infrastructure needed to support business scale expansion. Year-to-date capital expenditures were $231 million. We are investing in expanding capacity to support future growth and funding these investments through operating activities. Even though our investment amount has reached roughly four times full-year 2025 spending, our cash balance remains roughly at the level at the beginning of the year. Our margin structure enables us to invest in scaling the business while maintaining a strong balance sheet. This is a financial model we intentionally designed and implemented, and it is working as planned. Now let's look at this quarter's specific data. As mentioned, third-quarter revenue was $1.24 billion, up 138% year-over-year and 16% sequentially. Product revenue was $1.07 billion, up 179% year-over-year and 15% sequentially. This was our first quarter with product revenue exceeding $1 billion, accounting for 87% of total revenue. Product revenue is recognized when control transfers to the customer, so it directly reflects the scale of systems delivered and accepted by customers during the quarter. Service revenue was $67.6 million, up 15% year-over-year and roughly flat sequentially. Installation revenue was $84.9 million, up 29% year-over-year and 66% sequentially. Installation scope varies by customer and project, and related revenue is recognized progressively as work progresses. Therefore, installation revenue does not move in lockstep with product revenue each quarter, and installation revenue will fluctuate with the timing and mix of customer projects. Next, margins. Product gross margin was 38.4% this quarter, up 250 basis points year-over-year and 120 basis points sequentially. Notably, this quarter we were both ramping capacity on new production lines and accelerating delivery. This metric directly reflects our pricing power and the cost reductions we continue to drive in product manufacturing. Service gross margin was 21.4%, up 700 basis points year-over-year and above 20% for the second consecutive quarter. Total gross margin was 34.5%, up 412 basis points year-over-year. Margins expanded in both product and service businesses, but this growth was partially offset by the share of installation revenue in the revenue mix and its recognition timing. Given quarter-to-quarter changes in revenue mix, we choose to manage and provide gross margin guidance on a full-year basis. Operating expenses were $141 million this quarter, reflecting continued investment in our engineering R&D roadmap, growth initiatives, and infrastructure. Operating income was $286 million, up more than 500% year-over-year and 19% sequentially. Operating margin was 23.1%, up 60 basis points sequentially. Adjusted EBITDA was $299 million, equivalent to 24% of revenue, expanding 1,286 basis points year-over-year. Non-GAAP diluted earnings per share were $0.94, compared with $0.15 in the prior-year period. GAAP diluted earnings per share were $0.78, compared with a loss of $0.10 per share in the prior-year period. Next, cash flow. Operating activities used $27 million in cash this quarter. Net income was offset by increases in working capital, primarily from inventory, contract assets, and receivables, reflecting our support for significant business growth while building inventory in advance for future deliveries. As I mentioned earlier, product revenue is recognized when control transfers to the customer, while invoicing is based on contractual milestone nodes, which may fall in different quarters. Therefore, cash collection may lag revenue recognition; this timing difference is reflected in contract assets and receivables until invoicing is completed and cash is collected. The timing of large project deliveries and related cash collections will vary by quarter, but these differences will balance out over time. We expect to continue investing in working capital to support future growth and scale the business in step with customers. Capital expenditures were $153 million this quarter and $231 million year-to-date, primarily for investing in machinery and equipment and production capacity to support our expansion of existing Fremont and Newark facilities. By adding capacity in stages, we can align capital expenditures with observed demand rather than expanding significantly ahead of time and having capacity outpace demand. Free cash flow was negative $181 million this quarter, meaning $181 million in cash used; year-to-date free cash flow was positive $42 million. As of quarter-end, we held $2.5 billion in cash. Finally, Bloom was included in the S&P 500 this quarter. We believe this inclusion reflects the company's achievements in business scale, profitability, and performance stability. We also welcome the broader shareholder base that comes with it. Next, guidance. We are raising our full-year revenue outlook to $4.4 billion to $4.6 billion, corresponding to approximately 122% year-over-year growth at the midpoint of the guidance range. This increase reflects third-quarter performance and our improved visibility into fourth-quarter customer delivery timing. Our quarterly performance cadence depends on customer readiness and delivery timing, so we manage the business and provide guidance on a full-year basis. As with each of our previous guidance raises this year, our outlook is built on two layers. The first layer is backlog, where we assess site readiness and conversion timing for each customer project. The second layer is a probability-weighted view of our project pipeline, including customers that need power this quarter. With one quarter left in the year, the delivery timing of a small number of projects around year-end will determine where final results land within the guidance range. We are maintaining our previously provided non-GAAP gross margin outlook of approximately 34%. This level is above 2025's 30% and above the approximately 32% we initially expected at the beginning of this year. This judgment reflects our expectations for fourth-quarter revenue mix and margins across business lines. We are also raising our non-GAAP operating income guidance to $960 million to $1.03 billion. At the midpoint of the guidance range, operating margin is 22%, compared with 14% at the midpoint of our initial February guidance. Currently, our non-GAAP diluted earnings per share guidance is $3.10 to $3.40. To summarize, we are delivering record results, raising our full-year outlook, and making necessary investments to capture the enormous market opportunity before us. At the same time, we are expanding margins and maintaining a strong balance sheet. II. Q&A Session 1. JPMorgan Analyst Mark Strouse KR, I'd like you to talk about the company's recent decision to expand its real estate footprint in Fremont. You spent quite a bit of time on the call discussing various ways to expand the addressable market. However, I'd like to know why you chose to make this decision at this time? Also, you previously mentioned that the current manufacturing footprint can support more than 5 gigawatts of output per year. So should we now think that once the new facility is operational, annual capacity will exceed 10 gigawatts? K. Sridhar Mark, thank you for your question. For us, one thing is very clear. In this uncertain world, some things are absolutely certain. Electricity use, electricity demand, and the need for electricity will all grow sharply. This is not only because of artificial intelligence, but also because many market segments in the digital age need electricity. This trend will not only happen within the United States. We are already beginning to see the same trend emerging in Asia, Europe, and other parts of the world. The factors driving demand for on-site power rather than grid power include not only data centers requiring self-generated power but also various other sources of demand. These include resilience and security. In various scenarios around the world, critical facilities increasingly need reliable, secure, and on-site power. Taking all these factors together, the conclusion is obvious. In fact, we have always been clear on this. In terms of timing, 5 gigawatts is just a milestone in a long journey. We can clearly see ourselves approaching this milestone, so we need to prepare for what comes after. That is why you are seeing us take action now. We will never increase capacity in a single large step. We will adjust capacity gradually based on demand, like turning an analog knob, keeping capacity in sync with demand. We will neither expand ahead of demand nor move so slowly that we become a bottleneck for customers seeking power. This is the process through which we manage capacity. We will not provide a specific timeline, but we have firm confidence in this. The market feedback we have received is positive enough to justify beginning preparations for demand beyond the 5-gigawatt milestone. That is why we are taking this action. I hope this answers your question. Mark W. Strouse Next, I'd like to quickly follow up with Simon on one question. Regarding the full-year gross margin guidance, you maintained the original guidance despite raising revenue expectations. I think you mentioned that at least part of the reason is revenue mix. Could you elaborate on what other factors are involved? Also, I know you haven't provided 2027 guidance yet, but directionally, as revenue continues to grow, should we continue to expect gross margins to improve further? Simon Edwards Okay, Mark. First, I'd like to step back and look at how far we have come. Total gross margin was 28.7% in 2024, 30.3% in 2025, and has already reached 34.5% in the third quarter of this year. So I think we have demonstrated that margins can continue to improve as the business scales. Along the way, I am very pleased with how product gross margin has performed. As I mentioned in my prepared remarks, our pricing remains strong, and the cost reduction measures we continue to drive are already reflected in the financial data. At the same time, we need to balance margin expansion with capturing the market opportunity before us. As KR said, we are ramping capacity on new production lines and accelerating delivery to meet customer timelines. I think this will bring some near-term costs. In addition, we are advancing multiple engineering, sourcing, and manufacturing cost reduction programs. These measures will take time to gradually flow through cost of sales. As for next year, I would suggest viewing this from a long-term development perspective. We know that scaling brings startup costs, and we have factored this into our relatively prudent gross margin guidance. As these projects enter production, production lines mature, and capacity utilization improves, we expect to gain more benefits from scale effects. I would also like to add that in my first six months with the company, I have been very impressed by both the operational progress the team has made and how that progress is reflected in the financial data. 2. Evercore ISI Analyst Nick Amicucci Simon, I'd like to dig a little deeper into free cash flow performance. Operating activities used $27 million in cash this quarter while your margins expanded significantly. Could you elaborate on the factors involved, offsetting items, or one-time factors? Simon Edwards Okay, Nick. First, I want to emphasize that we do view this business as self-funding. In other words, the operating performance of the business can fund the investments needed to drive growth. I particularly want to emphasize this because it is very important to us. It reduces our dependence on external capital, significantly lowers execution risk, and gives us the flexibility KR mentioned earlier to invest according to customer demand. So I would suggest looking at this from the perspective of full-year cash generation rather than simply extrapolating from a single quarter's performance. We do believe that as the business continues to develop, we will be able to become a company that is free cash flow positive and able to fund its own growth. Returning to the first part of your question, we used cash in operating activities this quarter, but that does not mean a trend has begun. Year-to-date, even as we increased operating investments in people and capability building, cash flow from operating activities remained positive. In addition, as I mentioned in my prepared remarks, we also provided the working capital needed for deployment on certain customer projects. Our ability to fund year-to-date capital expenditures while continuing to increase operating investments speaks volumes about the capital efficiency of our business. If you also consider the returns that capital expenditure investments can generate, we are very satisfied with our current trajectory. Nicholas Amicucci KR, I'd also like to follow up on the utility question. You mentioned growing demand and strengthening partnerships with utilities. Historically, you seem to have provided more behind-the-meter solutions; but now it appears your solutions are becoming more complementary. Given what has been announced so far, we see fuel cells being included in integrated resource plans (IRPs) in different jurisdictions. I'd like you to clarify how large you believe the addressable market for this opportunity is, especially in the regulated utility segment? K. Sridhar Nick, that's a very good question. First, let's look at it from the customer's perspective. Here I want to distinguish between two types of customers: end customers, who pay electricity bills, and utilities, which serve those customers. If we optimize for both types of customers at the same time, the more they view us as a distribution asset believing that deploying our systems can lower total costs while continuing to serve their customers the better it is for both parties. This is better for electricity users and also helps utilities apply to policymakers to include spending on Bloom Energy Server in the regulated asset base and incorporate these devices into their transmission and distribution systems as transmission and distribution assets. What is the result? Utilities can deploy our equipment in front of the meter and allocate power based on load growth. They do not have to wait for a specific customer to place an order. They can procure based on planning and expected demand, just as they currently develop integrated resource plans (IRPs), without having to lock in a specific customer in advance. Why would they do this rather than purchase centralized generation assets? Because transmission and distribution infrastructure and its related costs, bottlenecks, and required construction timelines no longer suit their needs. They are gradually recognizing this. If customers turn to behind-the-meter power solutions, utilities could lose customers and also lose the corresponding regulated asset base. If they lose customers and regulated asset base, they would have to raise costs for other electricity users. This is neither what state governments want to see nor what end users want to see. Therefore, at this time, we can provide truly valuable assets to the entire industry. As regions gradually recognize this, you will see this model spread across multiple regions. Our team has done a great job explaining to utilities the benefits of on-site power generation as a complement to the grid. We have been pointing out the grid's shortcomings. Now, utilities finally have greater room from regulators to include such assets in planning, so they are actively seizing this opportunity. We believe this will become a major business driving the company's growth in the future. 3. Morgan Stanley Analyst David Arcaro KR, you have multiple potential opportunities ahead, including carbon capture, combined heat and power, and opportunities arising as the entire industry shifts to 800-volt DC power supply. I'd like you to describe the current progress of these future businesses, including how commercial opportunities are taking shape and the market adoption you are observing in the near term. K. Sridhar David, thank you for your question. I cannot fully express how excited I am about the 700-megawatt carbon capture project announced by AEP. The project will provide near-net-zero emissions power to a hyperscaler, and that customer is Google. This solution will be able to provide on-site power in a timely manner while achieving reliable power supply, no air pollution, no water use, and near-net-zero emissions. What solution could be better than that? I think investors on this call need to understand a key point: currently, combustion technologies cannot provide an economical and scalable path to carbon capture. In commercial applications, Bloom's technology has a unique advantage because it can provide high-concentration carbon dioxide, facilitating carbon capture. This is the brilliance of Bloom's technology and the key to our ability to achieve multiple advantages simultaneously. Our solution benefits both customers' economics and the environment. Therefore, I believe that when the world is ready to accelerate decarbonization, one of the fastest ways to decarbonize will be to use our technology to generate large amounts of reliable and environmentally friendly power fueled by natural gas. When the world is ready for this, I believe this market will exceed $1 trillion. That is my judgment on its market potential. As for 800-volt DC power supply, this is almost self-evident. We noted in a published report that a gigawatt-scale data center using an 800-volt DC power architecture can save more than $5 billion in its first five years of operation compared with using the current outdated power architecture inherited from the industrial age. 800-volt DC power supply will become the trend regardless. We do not need to push the market for this shift to occur. But given that, why go through all the intermediate power conversion steps and spend an extra $5 billion? I am very confident that once server racks using 800-volt DC power supply are launched, we will see adoption of this architecture. We are in discussions with all hyperscalers. They are showing strong interest in this area. This trend will occur as data centers using 800-volt DC power supply as a standard are built. The question is not whether it will happen, but how soon. As for the specific timing, customers are in a better position than I am to judge. Regarding heat recovery, or combined heat and power (CHP), we fully believe that heat generated by on-site power generation can be used for cooling. If this heat is simply discarded, it wastes a resource that can meet cooling needs and reduce electricity demand by 20%. Therefore, CHP is a natural complement to all our solutions. Whether you provide AC power or 800-volt DC power to a data center, and whether or not carbon capture is integrated, this does not change. CHP is the obvious choice. There is another interesting aspect of CHP: most industrial applications need heat in the form of steam or cooling capacity, and we believe the commercial and industrial (C&I) segment will become a huge growth direction for the company. These customers will adopt CHP. Therefore, I believe CHP will become an important growth driver for the company and will also create value for commercial and industrial customers. The advantage of this technology is that it uses the same platform as our existing business. That is why we have always told everyone: Bloom provides a platform, not a single product. We can provide any one or more solutions on the same platform according to customer needs. 4. Royal Bank of Canada Capital Markets Analyst Chris Dendrinos Congratulations on the strong quarterly results. I'd like to ask about the time required to supply power. Looking back, you mentioned in your prepared remarks that a year ago it took less than two months to deliver power to Oracle. As demand grows and the company scales, is the time from project start to power supply still less than 90 days? Or has delivery time been extended because backlog is very strong? Also, could you elaborate on the speed of power supply to the Nebius site? I recall you mentioned the first power modules are about to be put into use, but what scale of power supply in megawatts does that correspond to? K. Sridhar Thank you, Chris. That's a great question. We are ready to supply power for the Nebius data center. We stand by our commitments and have in fact delivered on the commitments we made to them. We will not be a bottleneck for them as they launch their entire 300-megawatt data center. We are ready and can supply power at any time. However, you should understand how data centers are built. A data center cannot go from zero to 300 megawatts in a day; it is built in stages and modules. Their first module is nearing readiness, and we are ready to supply power for the first module. No matter how many modules they build next, we will be ready in advance before each module needs power, ensuring we do not become their bottleneck. As for the construction and commissioning timeline for the rest of the data center, the customer should speak to that, and we will not comment on it. However, returning to the broader question you raised. Earlier today I think it may have been today Berlin time, or possibly yesterday the NVIDIA Corporation CEO said at the GTC conference in Berlin that every quarter of delay for a gigawatt-scale data center means $10 billion in lost revenue. In other words, a 100-megawatt data center delayed by one quarter would lose $1 billion in revenue. This loss far exceeds any other value that power modules can provide. Currently, there is no on-site power technology cleaner than ours that can also obtain air emissions permits faster. Nor is there any other technology that can provide power to customers faster. In addition, we can provide full future adaptability and all the additional value mentioned in the previous question. This is why we are optimistic about this segment and what it means for Bloom. Christopher Dendrinos I'd like to follow up on Mark's earlier question about expanding the capacity footprint. Do you have opportunities to further increase automation? How do you view opportunities in manufacturing and the cost savings that automation could bring? K. Sridhar I think anyone who has visited our factory has heard a saying from our General Manager Brad Fields. He is excellent, and one of his sayings is memorable. He reminds everyone that years ago, when we could only produce 100 megawatts of stacks per year, it took about 600 employees working around the clock, 24 hours a day, 7 days a week. Today, when our capacity reaches 1 gigawatt per year, this production line still requires only 600 employees per year. That means a 10x increase in capacity. By the time capacity reaches 5 gigawatts, this production line will still require only 600 employees. This shows that cost reduction and efficiency improvement are embedded in Bloom's DNA, and we continue to innovate. So the answer is of course yes. We benefit from AI not only because AI companies are our customers, but also in many other areas, including automation. 5. UBS Group AG Analyst Manav Gupta KR, I'd like to go back to the previous earnings call. You made a very insightful point at that time: using levelized cost of energy (LCOE) for on-site power generation is not a reasonable measure; what data centers really care about is total cost. Could you help us better understand the difference between what data centers see as LCOE and the actual cost of power delivered to them? How should we understand that the pricing of actual power supply includes higher costs while your solution is actually more efficient? Also, I'd like to quickly ask about the report you mentioned released on September 15 or 16. It was a good report because it described how much cost could be saved by switching to your solution. I recall you said a 1-gigawatt facility using your solution could save up to 9% in costs. One last question. I promise this is the last one. We know your solution has lower nitrogen oxide (NOx) and sulfur oxide (SOx) emissions, but we believe you also have an advantage in water savings. Your solution consumes less water than competitors, which could be a major competitive advantage. Could you elaborate on some of these issues? K. Sridhar Okay, Manav, thank you. You asked many questions, and I will try to remember and answer as many as possible. Let me start with water use. We have done relevant calculations before. If you compare the water consumption of a gigawatt-scale plant using combustion technology versus using our technology, the difference is so large that it cannot even be reflected within the rounding error of the second decimal place. If a data center reaches 1 gigawatt, then over its entire lifecycle, the amount of water consumed by combustion technology would be enormous. For example, if wet-cooled combined cycle gas turbines (CCGT) are used, the water consumption associated with a 1-gigawatt data center over its entire lifecycle could reach 25 billion gallons. That is a very large number. By comparison, our solution uses 1.4 million gallons of water. So from a water use perspective, compared with 25 billion gallons, our water use is not even enough to be called a drop in the ocean. That is how we view the water issue. Next, let me discuss the cost advantage you raised compared with grid power. From the grid to the server rack, power goes through multiple stages. First, power provided by the grid is not entirely clean here I do not mean clean in the environmental sense, but in terms of voltage and power quality. UPS can improve power quality and enhance supply reliability. After that, power also needs to be converted to the required voltage and DC form. These stages all add to the cost from grid power supply to rack power supply. Moreover, within the server rack, because grid power technology cannot quickly ramp load up and down, and turbines also cannot quickly ramp output up and down, additional efficiency losses occur. To avoid damaging turbines, engines, or battery systems at the other end, data centers can only waste a portion of electricity. But with fuel