Applied Digital (APLD) 2027财年第一季度业绩电话会议：营收增长322%，AI产能建设lg...

核心要点

• Applied Digital公布2027财年第一季度营收为3.419亿美元，同比增长322%（上年同期为8090万美元），主要由高性能计算（HPC）托管业务的2.626亿美元收入推动。
• 调整后EBITDA从上年同期的50万美元增至6440万美元。调整后净亏损为410万美元，即摊薄后每股亏损0.01美元。
• 归属于普通股股东的GAAP净亏损为2.210亿美元，即每股亏损0.76美元，其中包括巨额股权激励费用以及非现金按市值计价亏损。
• 该公司在三个州的五个园区拥有约360亿美元的已签约收入。管理层预计将在未来12个月内投入运营超过600兆瓦的容量，而过去12个月为250兆瓦。
• Applied Digital预计将在日历年年底前以大幅提高的价格签署约250兆瓦的扩建租赁协议。管理层表示，租赁费率增幅可能超过15%，且可能伴随更长的租期。
• 管理层表示，如果需求持续强劲且公司继续按计划交付项目，到2030日历年年底，运营组合容量可能达到3.5吉瓦至4吉瓦。

关键财务业绩

SG&A费用中包含与一次性业绩股票单位（PSU）相关的4130万美元费用。此外，由于Babcock & Wilcox认股权证及普通股投资的公允价值变动，Applied Digital还录得6750万美元的非现金亏损。

业务与运营表现

Applied Digital位于Polaris Forge 1园区的首批两栋建筑已完全具备投入服务条件。已投入运营的建筑包含10个数据机房和250兆瓦的关键IT负荷，而第三栋建筑仍在建设中。

公司预计到2026日历年年底，在北达科他州的总体关键IT容量将有300兆瓦上线。Polaris Forge 1完整的400兆瓦CoreWeave园区以及Polaris Forge 2处已签约的200兆瓦容量均已获得全额资金支持。

今年6月，Applied Digital按面值发行了15.9亿美元于2031年到期、票息率7%的高级担保债券。募集资金用于资助Polaris Forge 1的第三栋HPC建筑，并偿还了3亿美元的过桥融资。7%的票息率低于公司此前发行的首批Polaris Forge 1相关债券9.25%的票息率。

比特币数据中心托管业务在北达科他州的两个站点共运营286兆瓦容量。在有利的电力定价推动下，该板块的运营利润从上一季度的600万美元增至1330万美元。

Applied Digital持股约96%的ChronoScale宣布与微软达成一项50兆瓦的北美AI算力部署协议，使用英伟达GB-300-NVL72系统。管理层表示，近期签署的合同使ChronoScale有望在2027年实现10亿美元的年经常性收入（ARR）。

Applied Digital还签署了一项覆盖芬兰高达1吉瓦潜在电力容量的协议。首批100兆瓦预计将于2028年投入使用，随后容量将在2031年前逐步爬坡至完整的1吉瓦。

管理层展望

管理层预计将在未来12个月内投入运营超过600兆瓦的容量。此外还预计在日历年年底前以大幅高于早期协议的价格签署约250兆瓦的扩建租赁协议。

公司位于路易斯安那州和阿拉巴马州的开发项目预计将在2027日历年上半年开始贡献营收。

Applied Digital的目标是与第一梯队（Tier 1）投资级超大规模客户签署15年或更长时间的“照单全收”（take-or-pay）协议。管理层透露，下一批需要融资的园区——Polaris Forge 3、Delta Forge 1和Delta Forge 2——均已根据15年期“照单全收”合同出租给同一家投资级超大规模客户。

公司认为，到2030日历年年底，其运营组合容量可能达到3.5吉瓦至4吉瓦。要达到这一目标的上限，需要在2028年底达到接近2吉瓦的规模后，在2029年和2030年每年交付约1吉瓦的项目。

风险与关注领域

管理层特别提到，在某些市场，地方暂停审批禁令增多、选址要求收紧、许可审批周期延长以及社区阻力加大，影响了数据中心的建设开发。公司认为，这些限制可能会提升现有已建成园区的价值，但它们仍是对新项目开发的潜在障碍。

全行业的供应链依然紧张。Applied Digital表示，公司已锁定了关键零部件，在某些情况下甚至提前几年包下了工厂的全部产能，但目前仍在寻找额外的产能，以支持截至2029年的计划建设。

管理层预计，在经过汇率折算后，欧洲的项目成本与美国成本大致相当，不过欧洲的交付周期可能会更长。

实现3.5吉瓦至4吉瓦的组合目标取决于持续的客户需求、电力获取、融资可得性以及Applied Digital扩展建设和交付能力的能力。

分析师问答要点

• 租赁定价：管理层表示，拟议的扩建协议租赁费率增幅可能超过15%，且合同期限可能更长。
• 融资选择：对于由投资级超大规模客户支持的园区，除高收益债券市场外，公司正在评估项目融资和投资级债券结构。公司还优先考虑在建设风险降低后允许重新融资的条款。
• 北达科他州扩建：与Base Electron签订的购电协议覆盖紧邻Polaris Forge 3的约1200兆瓦天然气设施，预计将于2030年开始供电。
• 芬兰业务风险敞口：Applied Digital购买了该场地，但在需要支付额外款项之前保留了退出机制。管理层认为初始财务风险敞口有限。
• ChronoScale战略：管理层表示，在2027年和2028年，ChronoScale可能会运营几百兆瓦的容量，主要通过在Applied Digital核心大型园区之外建设50兆瓦至100兆瓦的站点来实现。
• 园区设计：Applied Digital表示，其设施的设计使用寿命为30年以上，并具备适应未来计算架构、冷却系统、网络配置和功率密度变化的灵活性。

业绩电话会议完整文字记录

---

完整财报电话会议逐字稿

管理层陈述

Operator

Good afternoon, and welcome to Applied Digital's Fiscal First Quarter 2027 Conference Call. My name is Pascal Berman, and I will be your operator today.

Before this call, Applied Digital issued its financial results for the fiscal first quarter ended August 31, 2026, in a press release, a copy of which has been furnished in a report on Form 8-K filed with the Securities and Exchange Commission, or SEC, and will be available in the Investor Relations section of the company's website.

Joining us on today's call are Applied Digital's Chairman and CEO, Wes Cummins; and CFO, Saidal Mohmand. Following their remarks, we will open the call for discussions. Before we begin, Matt Glover from Gateway Group will make a brief introductory statement. Mr. Glover, you may begin.

Matt Glover

Thank you, Pascal. Hello, everyone, and welcome to Applied Digital's Fiscal First Quarter 2027 Conference Call. Before management begins remarks, we would like to remind everyone that some statements we're making today may be considered forward-looking statements under securities laws and involve a number of risks and uncertainties. As a result, we caution you that there are a number of factors, many of which are beyond our control, which could cause actual results and events to differ materially from those described in the forward-looking statements. For more detailed risks and uncertainties and assumptions relating to our forward-looking statements, please see the disclosures in earnings release and public filings made with the SEC. We disclaim any obligation or undertaking to update forward-looking statements to reflect circumstances or events that occur after the date the forward-looking statements are made, except as required by law.

We also discuss non-GAAP financial metrics and encourage you to read our disclosures and the reconciliation tables to the applicable GAAP measures in earnings release carefully as you consider these metrics. We refer you to our filings with the SEC for detailed disclosures and descriptions of our business as well as uncertainties and other variable circumstances, including, but not limited to, risks and uncertainties identified under the caption Risk Factors in our annual report on Form 10-K and our quarterly reports on Form 10-Q. You may access to Applied Digital's SEC filings for free by visiting the SEC website at www.sec.gov. I would also like to remind everyone that the call is being recorded and made available for replay via in the Investor Relations section of Applied Digital's website.

I'll turn the call over to Applied Digital's Chairman and CEO, Wes Cummins, Wes?

Wesley Cummins

Thanks, Matt, and good afternoon, everyone. Thank you for joining our Fiscal First Quarter 2027 Earnings Conference Call. As we begin fiscal 2027, I'd like to provide an update on our vision, our strategy and most importantly, our execution. Today, we have approximately $36 billion in contracted revenue across 5 campuses, 3 states and 2 very different geographic regions of the country. Our focus is on converting our contracted portfolio in profitable growth while continuing to selectively pursue new opportunities across our expansive land and power portfolios.

During the past year, success was primarily measured by aggressively growing our portfolio of campuses and our total contracted value associated with those campuses. Over the near to medium term, we expect to continue expanding our existing campuses and selectively adding new locations with a strong focus on execution and bringing contracted capacity online for our customers. Over the next 12 months, we expect to place over 600 megawatts into service versus the 250 megawatts over the past 12 months.

Let me turn to our vision and strategy, which remains unchanged. Our goal is to become the category leader in design, build, deployment and operation of our purpose-built AI factories. Achieving that requires us to balance several things well. Selecting the right customers, matching those customers with a funding strategy that optimizes our financial returns and executing consistently on our bills. We believe we are well positioned to do all 3. We have a proprietary and proven data center design that has gained acceptance by the leading top-tier hyperscalers. We have a well-honed supply chain, deep energy partnerships and a demonstrated ability to deliver on our lease commitments.

Importantly, our data center design is different from many of our competitors. Some competitors' facilities are designed primarily around the typical 5- to 6-year life cycle of the GPU. We've taken a different approach and are building for the long term with 30-plus year horizons. We believe our AI factories are designed to be substantially more adaptable as compute architectures, network configurations and electrical and mechanical requirements evolve over time. Our design and our investment strategy are based on a set of core principles intended to extend the useful life of our AI factories and campuses well beyond any single generation of compute architecture. That matters because we believe the customer signing long-term leases with us expect these facilities to continue serving their needs well beyond the initial lease term. As a result, we believe our design and build strategy, together with our franchise model are highly differentiated and will generate outsized returns for our stockholders over the long term.

Our primary focus remains straightforward, build large, sustainable AI factory campuses and signed durable, high-quality long-term contracts. We target take-or-pay contracts of at least 15 years with proven Tier 1 investment-grade hyperscalers that are leaders in the AI industry and are well positioned to convert their investment into sustainable and profitable growth. We view our customers' ability to convert their AI investments into durable and sustainable earnings is an important metric as we make decisions on who we will build for. This is a relatively narrow group of customers who represent enormous demand and significant long-term spending potential. These customers have told us that they value our quality of design, the quality of our bills and the strength of our world-class team.

In addition to signing new contracts, one of our priorities is to continue to expanding our existing campuses. We also see opportunity to capture premium pricing as demand grows and we realize the economies of scale inherent in large AI campus developments. We have continued to make progress on expansion opportunities with our existing customers while also advancing discussions around additional campuses. We currently expect approximately 250 megawatts of expansion leases to be executed by calendar year-end of materially higher pricing compared to prior leases. As we continue to expand our existing campuses and seek to lock in premium pricing, we believe these campuses contain significant and underappreciated embedded value. We expect that value to grow over time as we execute against our contract portfolio and focus on delivering outsized long-term returns to our stockholders.

I'd like now to turn to our execution in North Dakota where we are currently building multiple AI factories across 3 campuses. North Dakota is our largest operating region and remains core to our strategy. We have repeatedly said its low cost of power and abundant energy resources, low population density and favorable climate make it one of the most compelling regions in North America for large-scale AI factory development. We expect additional Tier 1 hyperscalers to recognize those advantages over time. This is also the place where we have made the majority of our investments to date [ bolt ] in the campuses and the campuses we are building and in the communities with which we have partnered.

Power remains the single largest gating factor in our industry, and North Dakota is one of the few regions where we believe new generation can be added at meaningful scale. Speaking to that scale, we recently took a step a significant step in expanding our power position in North Dakota by entering into a long-term power purchase agreement with Base Electron, an independent power producer in which we hold an approximate 10% equity stake. The agreement covers an approximately 1,200 megawatt natural gas facility that Base Electron is developing in Center North Dakota adjacent to our Polaris Forge 3 campus with deliveries expected to begin in 2030. This is exactly the kind of outcome our model is well produce. While many operators are waiting in interconnection [ cues ] for capacity, they do not control, we are contracting for dedicated generation at the scale of our campuses with an agreement that can follow our tenants and a time line aligned with our future development needs.

We believe this dedicated supply substantially strengthens our ability to expand our North Dakota campus on our own time line rather than the grids and gives us greater visibility into the power resources that will support our future growth in the region.

To put the pace of our execution North Dakota in perspective, we signed our first lease with CoreWeave in May of 2025 for our first 2 buildings at Polaris Forge 1 the first of which was then under construction. By the end of calendar year 2025, the first building was fully operational, capable of sorting more than 40,000 liquid cooled NVIDIA GV-300. The second building, which broke ground in summer 2025 became operational approximately 1 year later. As of today, we have achieved full ready-for-service or RFS across our first 2 buildings at our Polaris Forge 1 campus. With our third building progressing on schedule. The operational buildings encompass 10 data halls and are supported by 250 megawatts of critical IT load. This represents a 150% increase in delivered capacity.

At [ land ] alone, we believe we are substantially ahead of our closest competitors in delivered capacity. And with the expected initial operation of our Harwood North Dakota campus, we anticipate having 300 megawatts of critical -- total critical IT online across North Dakota by the end of this calendar year. We believe this level of delivered capacity places us among the leading companies in our category regardless of geography.

In the United States, beyond North Dakota, we are developing campuses with substantial expansion potential in Louisiana and Alabama. These markets offer compelling characteristics, including access to power and strong state and local support for the build-out of national AI capacity. We view the presence of other top-tier hyperscalers developing large-scale AI campuses in these regions is further reinforcing their strategic importance. Our development activity in Louisiana and Alabama continue to progress, and we remain encouraged by the scale of customer demand across the region. We expect revenue contribution from this region of the country to begin in the first half of calendar year 2027.

We recently took our first steps outside the United States. While our focus remains firmly on executing our domestic projects, we believe select European markets may provide attractive opportunities to leverage our expertise and extend our platform over time as we continue to build Applied Digital into a category leader. As part of that early look, we signed an agreement for up to 1 gigawatt of potential power capacity in Finland. We view this as a measured first step. The agreement provides meaningful long-term potential while limiting our initial exposure and protecting downside risk while our near-term execution priorities remain centered on the United States.

Turning to our data center hosting business. Our 286 megawatts of capacity for Bitcoin mining across our 2 North Dakota sites continues to operate efficiently and provide steady high-margin revenue with minimal ongoing capital investment. And now to ChronoScale in which we have an approximately 96% ownership stake. [ Neo Clouds ] continue to represent one of the fastest-growing segments in this market. NVIDIA expects [ Neo clouds ] to exit this year with 8 gigawatts of installed capacity up from 3 gigawatts at the end of 2025. During the quarter, ChronoScale announced plans with Microsoft for a 50-megawatt AI compute deployment in North America featuring NVIDIA [ GB-300-NVL72 ] systems. This deployment is expected to significantly expand ChronoScale's existing business with a Tier 1 investment grade customer and represents another important step in scaling the platform. With recent new contracts, ChronoScale is on track to reach $1 billion of ARR in 2027 and momentum for additional contracts continues to build.

With that, I'll turn the call over to our CFO, Saidal Mohmand for a detailed review of the financials. Saidal?

Mohammad Saidal Mohmand

Thank you, Wes, and good afternoon, everybody. This quarter, we completed the financing of our first campus. Earlier this year, we had only one remaining tranche of that to place for the final 150-megawatt building at Polaris Forge 1. In June, as we discussed on our last call, we closed that tranche, a $1.59 billion offering of 7% senior secured notes due 2031 issued at par. With the proceeds of the offering, we have funded construction on the third HPC building at Polaris Forge 1 and have repaid the $300 million bridge facility we put in place in May. With that, the full 400 megawatts of the CoreWeave campus of Polaris Forge 1 and the 200 megawatts of contracted capacity at Polaris Forge 2 are now fully funded.

The pricing of that deal is worth pointing now. Our first Polaris Forge 1 related notes placed last year carried a [ 9.25% ] coupon. This tranche priced at 7%. We believe the pricing reflects the credit enhancements we secured with CoreWeave earlier this year and growing investor confidence in our platform. We are earning that confidence by executing and delivering our buildings on track and on budget, and we expect the first [indiscernible] Polaris Forge 2 similarly be on track and on budget. This success supports market confidence in our ability to execute. It drives us to reach our previously stated goal to keep bringing our cost of capital down, first at the project level and over time by refinancing operating buildings into lower cost markets such as ABS.

We also added flexibility at the corporate level. In June, we upsized our revolving credit facility to $430 million of commitments with an additional $120 million accordion remaining available. Looking ahead, the next campuses we need to finance are Polaris Forge 3, Delta Forge 1 and Delta Forge 2. All of these are leased to the same Tier 1 investment-grade hyperscaler under 15-year take-or-pay leases. We believe this fact changes the financing conversations of potential creditors. Recall, Tier 1 investment-grade tenants open access to a deeper pool of capital, which we believe supports more attractive terms. We are following the same framework preferred equity from Macquarie Asset Management, followed by project level debt, which allows Applied Digital shareholders to retain a majority ownership stake in each site. We are already in active discussions with [indiscernible] lending institutions and given the credit quality of this customer, we have a high level of confidence in our ability to fund these campuses at favorable rates and terms.

Now let's turn to the quarter. As a reminder, ChronoScale's is included in our GAAP consolidated results, but is excluded from our non-GAAP financial measures. We reported total revenue of $341.9 million, up 322% from $80.9 million in the prior year quarter. Our HPC hosting business generated $262.6 million of revenue, consisting of [ $65.8 ] million of base rent, $183.5 million of tenant fit-out services and $13.3 million of tenant recoveries. Our data center hosting segment, which operates our 2 Bitcoin hosting sites generated $37.8 million in revenue, in line with the prior year quarter. We continue to be very pleased with this business which generated $13.3 million of operating profit in a single quarter on approximately $111.9 million of segment assets. Segment operating profit this quarter more than doubled from $6 million in the prior quarter, driven by a favorable power pricing.

ChronoScale's revenue was $41.5 million this quarter, including approximately $23 million of GPU hardware sales. Services and other cost of revenues was $245.7 million, primarily consisting of $176.1 million and tenant fit-out services performed within our HPC [ Ocean ] business, $22.4 million associated with ChronoScale's GPU hardware sales, $19.8 million in energy costs and $17.6 million in depreciation and amortization. Data center rental and other cost of revenues was $43.9 million, primarily due to $22.4 million in depreciation and amortization expense on our operational AI factories at our Polaris Forge 1 campus. [indiscernible] in expenses, which are reimbursable as tenant recoveries and $7.8 million in personnel and other operating costs supporting our facilities.

SG&A was $114.7 million this quarter. This was primarily due to $65.4 million in stock-based comp, $16 million of professional services mainly related to transactions, $15.6 million of personnel expenses and $4.8 million of lease expenses. One item worth calling out in the quarter. Our stock-based compensation expense included $41.3 million tied to onetime performance stock units. We also recorded $67.5 million of a noncash loss from the change in fair value of our Babcock and Wilcox warrants and the common stock investment. Both are mark-to-market every quarter as B&W stock price changes. Net loss from continuing operations attributable to common stockholders was $221 million or $0.76 per share. Adjusted net loss was $4.1 million or $0.01 per diluted share. Adjusted EBITDA was $64.4 million, up from $500,000 in the prior year quarter. Net operating income, or NOI, which is exclusive to our HPC hosting business was [ $58.8 ] million, representing an 89% margin.

From a balance sheet perspective, we believe we remain well positioned. We ended the quarter with approximately $2.9 billion in cash and cash equivalents plus $0.7 billion of restricted cash and approximately -- against approximately $6.4 billion of debt. Note, more than 80% of that principal does not come due until fiscal 2031 or later. Stockholders' equity was approximately $1.6 billion. Our goal remains to maintain one of the strongest balance sheets in the industry through the majority of the construction phase and will execute as such.

With that, I'll turn the call back over to Wes for closing remarks. Thank you.

Wesley Cummins

Thank you, Saidal. Before we open the call for questions, I want to close on something that comes up in nearly every investor conversation we have today, the growing scrutiny around data center development across the country. We are seeing more local moratoriums, tighter selling requirements, longer permitting time lines and greater community resistance in certain markets. Some may view this as a risk to our industry. We see it differently. We believe it actually increases the strategic value of our campuses that are already powered operating and supported by their communities.

We think the reason is straightforward. At our campuses, once they're established we believe much of the hardest work has already been done. The land is controlled, power and interconnection are in place, permits have been secured, fiber, workforce and local supply chains have been established and perhaps most importantly to us, the community knows who we are and sees how we operate. This is why we think expanding at existing campuses is materially different than starting from scratch in a new market. We are already seeing this dynamic across our portfolio. Customers tell us they want additional capacity in places where they have confidence it can actually be delivered. As new data center development becomes more difficult in certain markets, we believe the scarcity value of established power and community supported campuses increases. We think that dynamic creates meaningful barriers to entry and can support stronger economics on expansions and renewals. And over the long term, higher terminal values for our assets.

Put simply, our view is that every new restriction elsewhere meets what we already own, harder to replicate and more valuable. This is also important when we think about the long-term value of these assets. Our leases have 15-year based terms, but the land, power, interconnection and community support behind these campuses have the capacity to last far longer than any single lease. When a base term ends when we see that a tenant with tens of thousands of GPUs deployed and a powered liquid cool campus faces a simple choice renew with us or try finding comparable power and approval somewhere else in a market where new sites are becoming increasingly difficult to approve.

We believe that makes renewals more likely support stronger renewal pricing and allows our tenants to continue refreshing their technology within the same facility for decades. This is why we believe in the durability of our leases and that the terminal value of these campuses is meaningfully higher than a traditional real estate model would have so. We also believe that the value should continue to grow as new supply becomes harder to build and [ locales ] have told us the reason our communities want us to expand is because we have worked hard to earn their trust.

Our approach has always been simple, do it the right way. In North Dakota, MDU's customers have saved about $45 million since our Ellendale project began because we use excess grid capacity and help spread fixed costs across more load. In Harwood, we paid for over $100 million of substation and grid upgrades ourselves, which benefit the entire community. And our model already meets the rate payer protection pledge, the Governor [ Armstrong ] joined alongside 21 other governors. Local leaders have also shared with us that the positive economic impact on these communities has been significant, the strong national attention, including coverage on Fox Business News and the all-in podcast.

Ellendale is a great example. Since we began our partnership with the community, local tax revenues expanded roughly tenfold from approximately $400,000 a year to nearly $4 million. These tax dollars are helping fund schools, infrastructure and emergency services. Construction workers on our sites are in 6-figure incomes on average than many of those dollars stay in the local economy. We were honored that [ Mid-Arica ] Economic Development Counsel recently named Polaris Forge 1 its Project of the Year, recognizing the campus for its economic impact, community investment, partnerships and technological innovation. That award belongs as much to the Ellendale community as it does to us. and we are incredibly grateful to everyone who continues to make Polaris Forge 1 possible.

We also invest directly in our local communities. Through Applied Digital Cares, we recently committed $350,000 across 5 organizations in Oliver County home to Polaris Forge 3. That includes funding the new Sheriff Deputy, a new life pack for the county ambulance repairs to 2 fire trucks, computers so every student in grade 7 to 12 at [ Center Statin ] has a device and repairs to the community center and pool in [indiscernible]. As Governor Armstrong put it, North Dakota wins when companies like Applied Digital choose to be more than an employer. We don't just build in communities we build with them and that is why they are welcoming us back to grow and we believe each of our campuses have significant embedded value that will continue to build over time.

In closing, we believe we're in the right place at the right time and doing this the right way. More than ever, we are in control of our own destiny across the full stack. We own the land, we design, we build and operate the AI factories. Through ChronoScale, we reach all the way to GPU compute and through partners like Base Electron, we have visibility into the power of the feeds at all. If demand continues on its current path, and we continue to execute and deliver for our customers and our communities as we've been doing. We believe we have the ability to grow our operating portfolio to 3.5 to 4 gigawatts by the end of calendar year 2030.

With that, operator, we're happy to open the call for questions.

Operator

[Operator Instructions] Your first question is from the line of Mike Grondahl with Northland Securities.

分析师问答

Mike Grondahl

Applied clearly as a first mover advantage in North Dakota. Can you talk a little bit about the perception of North Dakota as a market and the receptivity you're seeing from hyperscalers?

Wesley Cummins

Sure, Mike. Thanks for the question. So what we're seeing in North Dakota, obviously, we have 3 customers there ourselves already. We're aware of a self-build project at roughly 1 gigawatt self-build project from the Tier 1 hyperscalers in the state as well. And we've been speaking -- that is outside of the ones that we already have in the state and we're having multiple conversations with other the Tier 1 hyperscalers for additional capacity in the state of North Dakota.

The answer is I think all of them will eventually be there. And that's a state that has one of the most attractive things about the state beyond the power and the climate and the population density is they have a framework for how data centers are treated in the state. And so everyone knows exactly what to expect as you go through a process in that state. And as long as you do the development the right way, it's an extraordinarily attractive state to be in, and I think it's just going to become more so over the remainder of this year in '27 and '28.

Mike Grondahl

Got it. And then maybe for Saidal, the bond market has been challenging to say the least. Can you talk about current options you're seeing for investment-grade hyperscaler debt funding?

Mohammad Saidal Mohmand

Thanks for the question, Mike. And it's a great question. As you can recall, we completed the Polaris Forge 2 and the Polaris Forge 1 financings, which you could argue are not the same credit quality with great customers. They execute great, great tenants to have.

So when you think about the go-forward with Polaris Forge 3 and Delta Forge 1 and 2, given that it is in high investment-grade hyperscaler, there is multiple options beyond the high-yield bond market, which we have liked and we are -- we've been an issuer in with the past projects. So for instance, the project finance market is well open for high investment-grade offtakes, as well as there's some new innovations within the investment-grade bond market. So what we have seen, there's flexibility amongst various products that we can tap. Obviously, we weigh those and find a mix that works for our profile. But what we are focused on in a rate environment that can move around 50 to 100 basis points. One, we don't underwrite leases where a 100 basis point move would impact our economics really.

But then two, we focus then on terms where we can refinance in a year or 2 years with minimal penalties. So once that construction risk, which generally is anywhere from 25 to 100 basis points of construction risk premium on a debt instrument. Once we go past the construction space, we have the flexibility to lower our costs relatively quick into a more permanent financing. That is our focus. We're nimble and it's a great thing having a high investment-grade hyperscaler with multiple options to run at.

Operator

Your next question comes from the line of Nick Giles with B. Riley Securities.

Nick Giles

Yes. Thanks, operator. Congrats on the recent execution. I just wanted to ask about this Finland announcement. I think in the release, you highlighted limited initial exposure. And I was just hoping you could expand on that point, what kind of options are embedded in the current agreement and how you kind of see the commercial side of that site progressing?

Wesley Cummins

Sure, Nick. Thanks. So the for the site, we purchased the site and then we have the option. So the initial power of that site comes in 2028 and then additional power, which is typical of almost any site that's going to scale [ 2 ] gigawatt, there's breakpoints where get approvals for additional power ads. There's pretty good line of sight to that. But then there would be additional payments at that point, but there's an off-ramp for us prior to those additional payments. So it's a pretty minimal amount for us to have this kind of exposure in a market that we think has become pretty attractive as far as the European market goes. And we've spent the last, I don't know, 12 months looking into that market. And we've looked at probably 30 to 40 different sites, and we landed on this one. We think it's a good site and it's kind of coincided nicely to a point where our existing and the other potential customers that we talk to sort of asking more in the last 4 months about European capacity.

So those things kind of go pretty well together, but it gives us a lot of optionality to the upside with pretty minimal financial risk.

Nick Giles

Got it. And maybe just switching gears. In your prepared remarks, I think you really highlighted the longevity of these data center campuses today and I heard it as kind of a change in tone or maybe highlighting that more so than you have in the past. So I was just wondering if there's -- what were -- the genesis of those comments. I mean, obviously, GPU refresh cycles continue to be part of the discussion here. But yes, anything that you'd highlight from that perspective, I'd appreciate.

Wesley Cummins

Sure, Nick. So a couple of things. We've talked since we started the design that we use today, right, we why call it our fourth generation data center design. With this design, we made it extraordinarily flexible with all of the learnings we had gathered through the processes that we've gone through with all of the hyperscalers. So 1 one flexible enough to work for any of the hyperscalers. But two, we make it flexible to future proof of as much as possible. And some of that comes into how we do the electrical, some of it in the cooling. A big part of it is the amount of space that we put inside of the data hall so that the trend is for higher power densities, and that's the foreseeable future. But if it goes to much lower power densities, we have the space inside of the facility to accommodate that as well. So we just built maximum flexibility. We've talked about this for a few years now.

But what I'm seeing in the market is a lot of things happening that are just hyper-focused on how quickly something can come online and doing close to unnatural act, I would say, to just put capacity online as quickly as possible. And while I'm not saying that's a good or a bad business strategy, I just want to differentiate from what we're doing, which is a much more longer-term view of how we build the data centers so that they're very long-term assets. You can never fully know what the future brings, but that's the view that we have as a company. That's how we design them. That's how we design them from a redundancy from both a power and cooling perspective, because we could build a slightly cheaper product. We do have to make sure that we can operate this product for at least 15 years on our current leases within the SLAs, but we want to operate them longer than that. But it's just really more insight into kind of the thought process and the longer-term thinking behind our data center design and building the operations versus some of the other things that we're seeing just overall going on in the market.

Operator

Your next question is from the line of Derrick Whitfield with Texas Capital.

Derrick Whitfield

Good afternoon, and congrats on your continued commercial and execution progress. Wes, I wanted to start with the 1.3 gigawatts of U.S. capacity that you're currently marketing. During the last call, you noted you're experiencing a healthy market at present, and would expect base rates to improve on a risk-adjusted basis. How would you characterize what you're seeing in the market today since really the last call? Evenly everything that's happening from a regulatory moratorium perspective is playing your benefit, but would just let your perspective and color on that.

Wesley Cummins

Sure. So from a demand perspective in the market we're seeing no -- I wouldn't say big change either way. I mean it feels like we're at [ MAX ]. I've thought of that kind of the last 2 years and it continues to notch up but demand remains extremely robust for right now. People are still pushing for things they can get in 2027 really almost everything in '27 should be fairly contracted at this point. We'll start moving into '28 and '29. We have stuff is contracted into '28 but we're seeing a lot of demand for that as well. And really, these locations, like we keep pointing to North Dakota, but Louisiana is a great market for us as well. And so is Alabama we've got a great experience there with the community.

And the certainty of being able to continue to get the permits of existing power that we have in campus. But the supply chain for us to continue to execute. I would just reiterate those comments about we fully expect to do some lease expansion and new campuses at materially higher rates versus what we signed our last contracts for in the first half of 2026.

Derrick Whitfield

Very good. And then for my follow-up, I wanted to come back to Finland. While I realize you're early in the process with the emerging European AI market, I wanted to ask to give us [ Bill ] on how build costs might compare versus the U.S. as you think about the labor, building and equipment supplies?

Wesley Cummins

Yes. So we've actually went through that exercise and the build cost is we expect it to be very similar on a dollars basis versus euro. And so -- and by that, I mean, if we converted the euros back, it would be roughly the same dollars. What I would expect there is a more elongated delivery time line in Europe, just kind of how it works versus what we do here in the U.S. the goal is to replicate the same thing from the supply chain and design and build perspective in the markets when we enter Europe versus year, but we've done that work and expect a very similar cost.

Operator

Your next question comes from the line of Rob Brown with Lake Street Capital Markets.

Robert Brown

Good afternoon. Congratulations on all the progress. Just follow a little bit more on the North Dakota and the ability to expand there and how Base Electron fits into that given your PPA? How does that sort of start to enable the expansion there and what years are most affected by that?

Wesley Cummins

Sure. So we start to think about 23, Rob, primarily on Base Electrons. So Base Electrons building the 1.2 gigawatts in center expect Base Electron to build a significant amount of capacity on the eastern side of the state as well. We'll help campus expansion over in the Harwood campus and then potentially in the middle part of the state around James town. So there's a new WBI pipeline that WBI is owned by MDU that they approved and are moving forward with the pipeline. It's been in the works for years. And so we'll be an offtaker of gas at Center and at Harwood for certain and possibly in Jamestown.

The way we have worked that with the gas pipeline is to have almost unlimited capacity for expansion in Center. It's going to be the closest location to the beginning of the pipeline, and then these pipelines typically taper the further they go out. So to have a really big expansion capability at Center. This will all be front of the meter grid connected, but also expanded different locations across the state.

From an equipment perspective, we're looking at procured equipment. It's roughly 2.4, 2.5 gigawatts of generation capacity there already but we expect it to go significantly larger given the demand that we're seeing in the future there in North Dakota.

Robert Brown

Okay. And then just a quick follow-up on the supply chain work you've done. You've executed very well on plan. How have you mitigated any supply chain constraints where do you think in terms of supply chain?

Wesley Cummins

Sure. I mean supply chain, I think, is tight across the industry and some components more so than others. The biggest we've talked about this before from a supply chain perspective. The biggest thing we did was almost 2 years ago, locked in key components in the supply chain with certain of our suppliers. And in some of those instances, we actually purchased full factory capacity output from certain factories for, I believe, a 4-year time frame to make sure that we had those key components locked in. So that's been a big help to us. We're pushing to upsize upside that amount that we can take because we think as we go out into '27 and '28 and '29, we're going to need more supply chain to continue to build to the demand that we're seeing in the marketplace.

But that was the key piece for us is putting those relationships, those key relationships in place quite some time ago to make sure that we are locked into our places in the supply chain for key components.

Operator

Your next question is from the line of George Sutton with Craig Hallum.

George Sutton

Wes, you used the term selective when you were talking about the future build. I want to make sure I understand the term selective you're referring to. I think you're referring to the quality of folks that you'll work with versus the amount of capacity that you'll ultimately be able to build? Is that a fair conclusion?

Wesley Cummins

Yes. That's right, George. The company today is in a different place versus what it was a year ago. And we have a limited amount of capacity to build. And so we're being more selective with the customers that we'll build for and you're right on the credit quality aspect of that.

George Sutton

So relative to the 250 megawatts of expansion, you basically define them as likely coming at premium pricing. Can you just give us a sense of how you're defining premium pricing? How broad is that statement versus the base rent, what might else be involved?

Wesley Cummins

Well, we're pushing for on that, George, we're not to get into too much detail, but we've talked about this before, which is potentially longer duration and higher -- significantly higher lease rates don't to pin that, but as you can think of definitely north of up 15%-plus increases on the lease rate, but also could include longer duration on the leases. That's how we're defining that.

Operator

Your next question comes from the line of Darren Aftahi with Lucid Capital Markets.

Darren Paul Aftahi

Congrats on the progress. Just on Finland, I had a couple of points of clarification I wanted to get answered. So is there any way, Wes, you can elaborate on how much power initially is going to be at the site? Second point, what kind of community or governmental risk is there in that region either kind of near term or longer term? And then lastly, was the site sort of procured on an absolute basis? Or is this sort of procured with a specific customer or customer [indiscernible] in mind?

Wesley Cummins

Sure. So we view the location that we found in Finland. It felt a lot like the communities that we build in, in North Dakota. So there's a really positive receptivity from a local level on the community level. The country as a whole has been fairly receptive to data center. So we really focused on that for our to make this decision on the site plus the land availability, the fiber, this site happens to be about 30 kilometers from a self-build of one of the large hyperscalers. And then as far as the staging of the power, the first 100 megawatts is available in 2028 and it continues to ramp up through 2031 to the full gigawatt.

Operator

Your next question comes from the line of John Todaro with Needham & Company.

John Todaro

As we look after the 4 gigawatts or so by year-end 2030, I guess just what steps do you need to take to ensure you kind of get there. How much is reliant on Base Electron? And then also in Europe, too, it seems like some of that's going to come a little bit later in 2031, as you said. So is there other sites in your effect you need to get to as well?

Wesley Cummins

Yes. So thanks, John. And I'll just clarify, I saw how you went right to the 4 gigawatts, it was 3.5 to 4 gigawatts that we talked about by the end of 2030. So that's really from a power perspective, we have power in the power pipeline. Some of that includes the additional capacity from Base Electron that comes on in 2030. But from a power perspective, we believe we have that in the pipeline. This was really an exercise around how much more can we ramp up our build and delivery. And if you kind of work through the math, for us internally. If you want to get to the high end of that number, to get to the high end of that number, that really has us delivering about 1 gigawatt per year in '29 and '30, right? So that kind of if you take a bridge to the end of calendar year '28, getting to around 2 gigawatts. We, as a company, think that, that kind of the level we can ramp our construction and delivery up to.

So it's really those 2 things combined, but much more so on the ability and capacity to do construction and deliver for customers.

John Todaro

Okay. Understood. That's helpful. And then a follow-up just on ChronoScale, signed the 50 megawatts. Obviously, it seems like you could start ramping quite a bit from here. Just wondering kind of as we look out a few years, could we almost expect ChronoScale to be at 400 megawatts or so in Applied sites, almost like what CoreWeave did or less so just curious if ChronoScale [indiscernible] downline as they were?

Wesley Cummins

Yes. So thanks for that question. That's a great question. So when we look at that right now and you've seen over the coming month or so, the strategy we're using data center capacity of ChronoScale is outside of the large-scale Applied Digital sites. Those sites we contract to what we have said, which is Tier 1 investment-grade hyperscalers. The ability to secure other capacity, we have found a nice strategy for securing more near-term grid capacity that isn't it that scale at a single location, but at multiple locations, but I think you could see ChronoScale moving up to multiple hundreds of megawatts in capacity online outside the core Applied Digital data center builds with over '27 and '28.

But I want to bifurcate this where it is 2-step strategies where Applied Digital strategy is it remains the same, build for high-quality tenants at large scale. And then we're using our knowledge base to help ChronoScale secure capacity at sites that are more -- like you see the first Microsoft side of 50 megawatts that are kind of 50 to 100-megawatt sites that have been passed over because they don't scale to the gigawatt side or the multi-hundred megawatt side. But there's a lot of those around the country that are grid-tied that we think we can help them bring online over the next 2 or 3 years, a significant amount actually in 2027. So that's really -- I appreciate you letting me make that clarification around the strategy.

Operator

Your next question comes from the line of Michael Donovan with Compass Point.

Michael Donovan

Is it fair to see your existing Macquarie relationship would extend to Finland?

Wesley Cummins

So CoreWeave has been -- I couldn't have asked for better.

Michael Donovan

Macquarie, Macquarie.

Wesley Cummins

Macquarie, I thought I said CoreWeave I was going to talk about the partnership there. Macquarie is North America that partnership is in North America. I'm not saying we couldn't have something similar, but the current agreement we have is just North America.

Michael Donovan

Appreciate that. And if you had a big interest in CoreWeave, I will not stop you there. But...

Wesley Cummins

[indiscernible] CoreWeave that's what I have for that.

Michael Donovan

Excellent. And then to follow up on Nick's question. As [ Infra ] becomes a larger share of customer workloads, are you seeing changes in the mix of GPUs, CPU, storage requirements? And if so, are any of those changes contributing to the premium lease rates that you mentioned or expecting?

Wesley Cummins

So I'll tell you where we're seeing a change. So we're seeing a change over the past 18 months on the tenant fit out or the technical fit out TFO. And what we're seeing in -- because we execute that in almost actually every lease that we have, we execute that. The TFO is becoming much more flexible than what I would say. So if you roll back a year ago or especially 18 months ago, the TFO, we would know exactly what was going into our data center based on the TFO. And now we can't really tell. And the become a larger dollar amount for our customers, but I think they're building a significant amount of flexibility inside the facilities to be able to accommodate whatever workloads they want to deploy and that between -- we'd always talk it like GPU TPU, but I think it's also in the amount of CPU mix, the storage, all of those pieces, network topology. But that's really where we're seeing that.

Michael Donovan

Appreciate that. And then one more, if I may. You mentioned economies of scale for your current builds. Are there any circumstances where you would consider smaller modular build-outs?

Wesley Cummins

I think that's a great place for ChronoScale. Now just to clarify on that, are you talking about like the containerized solutions that people are talking about for edge compute or just like a 20, 50, 100-megawatt containerized solution.

Michael Donovan

More in line with 20 to 50, not so much containerized one, but or like small [indiscernible].

Wesley Cummins

We're looking at those. That's one of the ways we're securing data center capacity at ChronoScale is those types of modular builds on a smaller scale.

Operator

This concludes our Q&A. I will now turn the call back to Wes Cummins for closing remarks.

Wesley Cummins

Thanks, everyone, for joining our call, and we'll look forward to speaking to you again in January.

Operator

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

原文链接