NEXTDC Limited (NXT.AX) Earnings Call Transcript & Summary

Thank you for standing by. Welcome to NEXTDC Limited Financial Year 2026 Results Announcement Conference Call. [Operator Instructions] Please be advised that today's conference is being recorded. I'd now like to hand the conference over to Mr. Craig Scroggie, Chief Executive Officer and Managing Director. Please go ahead.

Thank you, Amber, and good morning, ladies and gentlemen. Welcome to the NEXTDC results presentation for FY '26. I'm joined today by our CFO, Oskar Tomaszewski. Our results announcement, presentation and annual report were lodged with the ASX overnight. I draw your attention to the forward-looking statements disclaimer in the presentation. Beginning on Slide 4. FY '26 was the largest contracting year in NEXTDC's history. Contracted utilization tripled to 740 megawatts on a pro forma basis, and we exceeded guidance on both net revenue and underlying EBITDA. Net revenue of $405 million was up 16% on FY '25. Underlying EBITDA of $248.8 million was up 15%. Contracted utilization increased 202% to 740 megawatts. Billing utilization increased 58% to 175 megawatts. Turning to Slide 5. NEXTDC's forward order book hit a new high, now 565 megawatts on the back of a record sales year where we contracted 495 megawatts. Every megawatt in that forward order book is a binding customer contract. Importantly, the forward order book will ramp into faster billing growth as [ David Dzienciol ] delivered clients quicker than at any time in company's history, with record revenue growth in FY '27 with 197 megawatts of new billing activation as multiple cloud and AI deployments ramp up. Turning to Slide 6. Our strong performance was highlighted by record key operating metrics. Net revenue, which is total revenue less direct costs grew 16%. Billing utilization grew 58% during the year to 175 megawatts, and our forward order book grew to a record 565 megawatts, which will more than quadruple billing utilization to 740 megawatts by FY '30, underwriting substantial organic revenue and EBITDA growth. We remain well capitalized to support our growth plans. Total assets of $10.2 billion, including $5.8 billion of property, plant and equipment as well as $3.2 billion of investment properties. We have available liquidity of $8.7 billion, giving us substantial balance sheet flexibility to fund our committed pipeline. Our first asset revaluations were included this year for M3 and S4, increasing reported assets and improving reported gearing. Qualifying assets will be revalued annually. Any movements may be gains or losses. Our funding and capital strategy continues to mature. We raised $9.75 billion in FY '26, which included senior debt upsizing of $5.8 billion, subordinated notes of $750 million, hybrid securities of $1.7 billion and $1.5 billion of new equity. Further capital initiatives are underway, including partnerships with third-party capital through joint venture structures for S4, S7 and M5 as those projects continue to secure new customer commitments, advance through key development milestones and are further derisked through a growing base of contracted revenues. We had another record development year. 42 megawatts were delivered at M3 and 12 delivered at M2. S4 is on track with its first slab on ground. Kuala Lumpur 1 opened to foundation customers, and our Edge network builds continue in Geelong, the Sunshine Coast and Darwin. I'll now hand over to Oskar to discuss the financial results in more detail.

Thank you, Craig. Let's now turn to Slide 8, which provides a summary of our profit and loss for the year. The statutory results reflect total revenue of $496.5 million, up 16% on FY '25 and net profit after tax of $82.1 million, which includes $129 million fair value gain on investment properties, as well as the recognition of previously unrecognized tax losses. Our non-statutory highlights include net revenue, which was up 16% to $405 million. Direct costs and power pass-through revenues rose in line with customer consumption. Our facility costs grew $11.3 million or 17%, with targeted investments ahead of record capacity expansion, including land acquisitions, headcount across new and expanded facilities as well as maintenance costs across capacity additions and expansions. Our corporate costs increased by $11.1 million or 17%, as we invest in systems and people to support new site expansion and accelerating customer deployments, with over 400 megawatts of build capacity growth expected across FY '27 and '28. Underlying EBITDA was up 15% to $248.8 million, almost 4% above the top end of the guidance range. This year, resulting from 2 of our material customer contract wins, we have an accounting reclassification and a measurement policy change, which we'll talk to later in the presentation. On to Slide 9. Net revenue growth accelerated in the second half, with strong growth expected to continue to FY '27 and beyond. Underlying EBITDA growth also accelerated, with the company's core operating platform now ready to deliver operating leverage and record earnings growth. Slide 10 sets out our revenue per unit metrics. The blended rate per megawatt reflects a deliberate shift to more efficient hyperscale and AI deployments. These reflect a lower annual per megawatt rate on a 10- to 15-year initial contract term, with extension options up to 35 years. Rapid growth in billing from high-density hyperscale is rapidly changing the revenue mix, with more than 70% of billing capacity now coming from hyperscale workloads. Our new and existing facility expansions are increasingly using highly efficient cooling technologies, including liquid to chip, which is driving down the cost per megawatt to deliver capacity. and higher power volumes with more efficient cost structures are driving returns, with new hyperscale and AI deployments targeting over 10% yield on cost. Slide 11 summarizes our balance sheet and liquidity position. At 30 June, NEXTDC had property, plant and equipment with a carrying value of $5.8 billion, as well as investment properties with a carrying value of $3.2 billion. The investment property values come from revaluing our M3 and S4 data center assets. We will now have periodic property valuations for qualifying assets going forward, with those assets revalued at least annually. The change to investment property accounting also reduces depreciation and introduces straight-line revenue recognition for the relevant sites going forward. We have included significant detail on the accounting policy changes and the valuation process and governance in today's results materials. Our net assets stand at approximately $6.1 billion, and NEXTDC has approximately $8.7 billion of liquidity as we completed several capital initiatives during FY '26, including those previously mentioned by Craig. Slide 12 details our debt funding profile. This slide illustrates our evolving capital structure, detailing where the $8.7 billion of available liquidity comes from and confirming all drawn senior debt at 30 June and subordinated capital is fully hedged. Importantly, due to the company's credit standing, it is no longer subject to any leverage ratio covenants. Our debt facilities are currently subject to a gearing ratio and an interest cover ratio, with contracted revenues not yet billing are included in EBITDA for covenant testing purposes. Slide 13 details our debt funding maturity profile. Across our diversified debt capital stack, our facilities have a weighted average maturity of approximately 5 years, with no debt maturities until FY '30 and a staggered maturity profile beyond that year. Slide 14 details our disciplined approach to capital. In the second half, NEXTDC has completed a $2.3 billion senior debt upsizing, issuance of hybrid securities with $1.7 billion of face value, providing flexible long-term capital to support growth, subordinated notes of $750 million, broadening the capital stack to include a new pool of investors and completed a pro rata entitlement offer of $1.5 billion, which was extremely well supported. Additional capital initiatives are underway, including, but not limited to, project financing where S4 and S7 are expected to transition to a ring-fenced asset level debt structure over time ahead of formation of JVCo and JVCo itself, where we continue to evaluate capital partnership options across our development portfolio. We believe the combination of existing liquidity, additional debt and hybrid capacity, other financing options in flight as well as selective asset capital recycling initiatives provides us with multiple pathways to fund the contracted pipeline while protecting shareholder value. I'll now hand you back across to Craig to go through our business performance and outlook for the 2027 financial year.

Thank you, Oskar. Turning now to Slide 16. Customers have now contracted 740 megawatts. We have built 288. Demand is running 2.5x ahead of everything we have ever delivered and less than 1/4 of what is contracted in billing today. That gap is not options or reservations. It is signed customer contracts and it converts to revenue as we deliver. Billing utilization grew 58% to 175 megawatts this year, and the contracts already signed, take it to 740 by FY '30. Billing more than quadruples from here. Turning to Slide 17, our non-financial metrics. The breakdown of contracted capacity by customer category shows 95% of our capacity is now contracted across multiple key cloud and AI customers. These customers are the key driver of density and scale, delivering operational and cost efficiency and improving returns. Interconnection by customer category shows the ecosystem is built on key network and provider partners, as well as enterprise and government customers, with ICT providers leveraging AI and cloud to deliver critical services to the enterprise. Slide 18 sets out our capacity and utilization. We now have a total planned capacity of over 3 gigawatts across our land bank portfolio of facilities that are either open, in development or development ready, subject to development approval. S4 had another capacity upgrade this half to 365 megawatts of IT load. As customers are contracting higher density deployments often for artificial intelligence, we now expect to deliver more billable IT power in the same size footprint. AI is also accelerating the speed of [Technical Difficulty] customer deployments, significantly shortening the time to payback and cash generation, supporting a faster pace of growth in the business. Our construction activity is matching the faster pace of customer deployments. 537 megawatts of built capacity is currently in progress and a further 130 megawatts is in plan. In New South Wales, we added 16 megawatts at S3, the last 12 megawatts for S3, 10.8 megawatts for S6 and the first 250 megawatts for S4 is now under construction. In Victoria, we added 12 megawatts at M2 and 42 megawatts at M3. Both facilities have their remaining capacity to complete now in progress. M4's early works have commenced with 10 megawatts in progress. Expansion works are in progress across key metro and Edge locations backed by key customer wins, including material enterprise and government customers as well as strategic network cable landing station infrastructure and satellite partners. KL1 opened to a foundation customer this year with 10 megawatts. It has a further 15 megawatts in progress and another 15 in plan. In Japan, Tokyo1's construction works have commenced with early excavation and retaining construction development underway. Disciplined site selection activities continue for additional sites across Asia. On Slide 20, we provide a summary of our ESG highlights. As the data center landscape continues to grow as a major feature in our landscape, our focus on sustainability is crucial. With cloud and AI demand exploding, powering digital infrastructure growth sustainably is critical. This year, we published our first Sustainability Report under AASB S2. S2 Sydney received the Uptime Institute Sustainability Assessment Award. M3 is running 27% below its embodied carbon baseline, and our construction partners achieved waste diversion of 94.7% at M3. On Slide 21, we provide a summary on our safety highlights. As our national fleet of mission-critical infrastructure assets continues to grow in size, so too does the importance of keeping our workforce safe. Across a record construction program, we delivered a construction LTIFR of 0.3 and an operational LTIFR of 2.2, both well below industry benchmarks. On Slide 23, before I turn to guidance, a comment on the proposed energy reforms in New South Wales and at the federal government level. These proposals apply at the point of new grid connection and planning approval. Our operating portfolio and the entire 565-megawatt forward order book are unaffected. And we have written confirmation from Transgrid that its new capacity allocation policy does not apply to S4. None of these proposals are law yet, and we are engaged in every consultation, including chairing the industry's Energy Policy and Technical Committee. We expect these reforms to play to our strengths and our FY '27 capital expenditure guidance allows for these matters. Turning to Slide 23 and our FY '27 guidance. We are pleased to provide our revenue and earnings guidance for FY '27, with net revenue of $615 million to $640 million, reflecting annual growth of more than 50%. The forward order book of 565 megawatts is now accelerating its billing ramp, with 197 megawatts of billing conversion in FY '27 and a further 221 megawatts in FY '28. NEXTDC's national metro footprint positions us for AI inferencing capability where sovereignty and data gravity require workloads to co-locate with enterprise and government. Underlying EBITDA of $385 million to $410 million as we make key investments to deliver a record increase in data center capacity in line with contracted customer commitments. Operating leverage expected in FY '27 is accelerating in line with the conversion of the forward order book, driving rapid near-term earnings growth. Total capital expenditure for the year is expected to be between $5.25 billion and $5.75 billion as we accelerate investment following 495 megawatts of new contract wins in FY '26. It's important to note inside this capital expenditure forecast, we estimate $500 million of reimbursable customer fit-out. 537 megawatts of built capacity is under development in line with contracted capacity. Accelerated expansion works for M2, M3, KL1 and S4 are all on schedule. Early works for M4 are in progress, and S5 is expected to commence construction in FY '27. Strategic metro and Edge development continues across all our sites, supporting enterprise, government and critical national network infrastructure, with colocation, inferencing, satellite and cable landing station capability featuring strongly. In FY '27, we signed record contracts, setting us up for a record revenue and earnings growth. The operating platform is now in place, and it positions the company to scale materially through '28 and beyond. Demand from both established hyperscale and emerging AI customers continues at a scale and pace that creates an enormous opportunity for the business. As AI adoption moves from experimentation into production, demand is no longer the question. Demand is shifting towards inference. That inference needs to sit close to the enterprise and government data it draws on. And NEXTDC's metro presence positions us for that shift. The infrastructure being built today underpins the next decade of productivity growth. And NEXTDC is proud to be building that platform at scale, at speed, safely, securely and sustainably. Amber, we can now open the line for questions.

[Operator Instructions] And our first question comes from the line of Eric Choi from Barrenjoey.

I had a few questions. The first one, I'm trying to use the new info you've given us to work out what the true value of the $9.2 billion that you've historically spent on land, building and fit-out is. And specifically, you said your hyperscale yield on cost is 10% plus DC cap rates are usually 6% to 7%. So if I put my REIT hat on, it sort of suggests you're generating a 1x or 1.5x development spread on every dollar that you deploy. So again, just through a brief investor lens, the value of what you have already spent, that could be $14 billion before I add future build and the value of your platform. So, maybe for Oskar or Craig, is that broadly correct? Because I'm also conscious I might be undercooking that development spread because you've got some other historic projects like S1 and M1 in that base as well. So, that's the first one.

Thanks, Eric. As you pointed out, those numbers, I think, are directionally right. I think that math is reasonable. I would just note that the yield on cost is a forward-looking estimate. So, won't include, obviously, the higher-yielding historical colocation business, but is focused on our expectations as it relates to the entire development for these hyperscale and AI deployments in the future.

Awesome. Can I take that one step further, Craig? Obviously, that excludes the value of your pipeline. And if I just hone in on the S4 and S7 portion of that pipeline, you're going to spend $15 billion of CapEx on that. Same rough math suggests that project could be worth $23 billion once complete. So, I think you previously said you're going to have $10 billion of debt for that. But if you take $10 billion of debt off, assume you guys have a 20% piece of the residual equity, it suggests those projects could be worth $2 billion to $3 billion for you guys, like pre any management fees, obviously, a future value. Sorry, can I just check that's broadly in the zone?

I think your comments, Eric, on S4 and S7 capital requirements are reasonable. The one thing I would make clear is that we have made no final decisions in relation to the structure of JVCo. So the commentary regarding 20% is not decided based on the conversations with the partners that we continue to undertake. There are many and varied structures that range from a number of different percentages. So, I would just caution you to note that whilst 20% may be one end of the spectrum, there are plenty of other options and alternatives as it relates to the considerations on JVCo. So, I think that your S4 and S7 commentary is reasonable. But I would just note that no decisions on the final equity or management structure in relation to JVCo have been made, and they remain under negotiation as part of our JVCo review at this point in time.

Helpful, Craig. While I'm annoying you with so many numbers, can I just fire you one last one, which is on Slide 5. You're explicitly guiding to about $400 million of EBITDA in FY '27. If I just take Slide 5 and assume you're adding megawatts at, I don't know, $1.5 million to $2 million, is that kind of implying a $700 million to $800 million outcome in '28 and then maybe close to $1 billion in FY '29, just ballpark?

Again, I won't endorse your number. But I will say that I think it's directionally right. The key thing to watch will be the shape of the slope and how quickly revenue is activated in '27 and '28. So, whilst we have an extremely high degree of certainty over the '27 forecast and revenue and the CapEx that will flow to support it, the '28 number, we deliberately don't give out years because as the program develops, if we see the opportunity to accelerate the delivery program, that may mean that we can activate capacity earlier depending on the progress of the site development. So again, I think your numbers are directionally reasonable. Difficult for us to give a firm number on '28 because it is subject to how fast the construction program is able to progress in FY '27. One year is a very, very long time in our business year. And now every year that moves forward, the overall speed at which we develop, deploy, activate and convert into cash is significantly accelerating.

And our next question comes from the line of Paul Mason from E&P.

Couple from me. The first one, just -- you've given pretty explicit guidance on greater than 10% target on cost for these hyperscale deals. In private, that gets talked about in the industry, but sort of we haven't heard you talk about these things in public before. You've probably been a little bit hesitant to discuss things explicitly on calls before. So, can you tell us, like, what changed that means that you are able to be as open as this now?

Thanks, Paul. Your line was a little hard to hear, but I think I got the gist of your question. What's changed is that when you considered the historical colocation business, whilst the numbers were small and the returns were pretty significant, the time at which we would sell and over what period of time and what period the facilities would fill was really the uncertain or unanswered question. And we were designing facilities for an unknown future. The key point of difference when you are doing build-to-suit at scale and you go from doing 10 or 20 or 50 megawatts to doing 250 or 500 or even potentially 1,000, the economics essentially are locked at the contract point. The costs on the estimate for the development, the long lead time items, the bulk procurement, the negotiation with the general contractor, all of those variables are certain on day 1 at the outset. And so as we start to think about a more mature financial position for the company, those conversations with joint venture partners show us that whilst the returns in the scale business in deploying tens of billions of capital may be lower than the traditional or historical colocation business, they are absolutely certain. And that gives us a high degree of confidence to be able to take a more mature approach and have a higher degree of certainty over what the returns will look like on these scale projects. So, that's the key change for us in communicating what our return expectations are is that once we have negotiated the contract with the customer and have the build cost in place with a high degree of certainty of the return.

Okay. Great. And just the second one. Just on your comments on Eric's question about potentially being able to move things forward depending on the pace of development on your sites, maybe just, like, are the customer contracts actually framed in a way where there's not like a fixed billing date, but you're actually able to start billing as soon as you hand things over? Or is there like a change to sort of normal contracting that's gone on there to enable that? Yes, sort of -- what's sort of the difference there versus previously when you have like a fixed date?

A couple of comments, Paul. First one, obviously, for Dave Dzienciol and the commercial organization, this year was a record year, selling 0.5 gigawatt was a really great result. I think it was probably 4 or 5 years in the making land bank, preparing for those scale sites to be putting us in a position to take advantage of that. So it wasn't 1 year's worth of work. I mean, the team did an extraordinary job on the execution side, but there's probably 4 or 5 years of historical work gone into putting us in a position to be successful. As it relates to then delivering those, almost every customer would take the capacity as early as we can deliver it. Now, there are a number of factors that come into play when we are doing BTS that are different to doing colo and preparing shells because largely, the delivery program is just fit-out. The reason you can't build 250 megawatts of inventory is that every hyperscaler or foundational model player or neocloud has a different design requirement. And given the significant variability of those designs and the sheer speed at which the GPU architecture itself is continuing to change, just this year was the first time that we had locked a gigawatt scale reference architecture. So, designing single sites for more than 1,000 megawatts of capacity, the reference architecture for those designs didn't exist before that. And so now as we enter this new age and we are required to plan for the development of these multiple 100-megawatt single buildings and over a campus, there may be many of those, it does mean that we can speed up the development if we are able to largely produce a manufacturing style of methodology for deploying the site. So off-site containerization, manufacturing, the preparation of the transmission infrastructure for scale power and then delivery of those, which would speed up our time to activation. So, there is a very, very good chance that as we are improving the manufacturing methodology of containerizing the development of the sites at up to 1,000 megawatts or greater, we will be able to very significantly increase the time to cash by simplifying the construction methodology and playing a larger assembly on the site style of role. So Paul, it does mean that my expectation is over time, whilst we haven't started a construction program given the variability of design requirements, historically, it may have taken us 18 months to build capacity and deliver it. My expectation now is that we can go from ground to fully operating facility within approximately 9 months.

And the question comes from the line of Jonathan Atkin from RBC Capital Markets.

So, a question about Slide 17. And is there any way to unpack the composition of the order book? You give kind of cloud and AI, but any way to kind of think about notionally the mix of traditional cloud operators versus neoclouds versus LLMs and other types of AI start-ups? And then I have a follow-up.

Thanks, Jon. We haven't split it out. Probably just give you one comment. As it stands today, all of that business is largely -- the majority of it is AA credit-rated counterparties or better. So whilst we do have some neocloud customers in addition to the traditional hyperscale customers, those neocloud customers are backstopped in contract by NVIDIA for a decade or more. So where we have minimum decade to 15-year long contract agreements with up to 35-year options for our hyperscale customers, we are very pleased with the outcome of NVIDIA's role in financially backstopping our AI cloud customers in that category. So today, I wouldn't necessarily need to split out the number because in Australia, all of the AI deployments are for inferencing, not for training. The regulation hasn't changed. The regulatory environment in relation to training of models has not changed. That may change at some point in the future, but certainly no indication on whether that will or what time frame that could happen. So, all deployments today are hyperscale or backstopped by NVIDIA of scale on 10 or 15 minimum year agreements with out to 35-year options. So the credit quality question for us is the primary one that we need to consider. And when that credit quality for the large majority of all of the 0.5 gigawatt we signed in FY '26 is AA or better, that would be how I would answer your question.

That's very clear. And you answered most of my second question. But when you get to, say, the non-AA or better category that is backstops, what do you think about kind of yields? You've talked about 10%. But is -- to you, is there an opportunity for yield enhancements by dealing a little bit more with that latter category? How do you think about target development yields for the latter category versus the former category, in other words?

Jon, the forecast we've given is obviously a conservative one. Any time you make these commitments, you need to carefully consider that we are 100% certain that we can deliver on those promises. And whilst they are forward-looking statements, we have a very high degree of confidence that, that will be the minimum expectation. Where that starts to lean into upside is based on the scale of the deployments. And so what I can tell you based on the S4 site today, we've made those estimates on a 365-megawatt site at the minimum yield on cost. But as you start to consider sites like S7 at 650 megawatts or even the recently acquired Melbourne M5 site at 1,200 megawatts of IT load, there are substantial opportunities to improve yield on cost given the sheer volume of containerization and the manufacturing style of methodology and approach that we're taking to the construction of the sites [Technical Difficulty]. The other key engineering point of difference for us outside of the actual manufacturing-led approach to the construction is that those build-to-suit hyperscale campuses for a 1 gigawatt DSX reference architecture will be single-story buildings. So, once you have slab on ground and in-ground services activated, the time line is quite certain because we're not building up, we're not exposed to weather and other things as much as you would be in a multistory development. So, single-story developments of 1,000 megawatts or greater certainly have an opportunity to improve on our minimum yield on cost return expectations for a site like S4 at 365 megawatts. So yes, I do believe that, that opportunity exists, Jon.

We will now take our next question from the line of Tim Plumbe from UBS.

Two questions from me if possible, please and apologies. But going back to that activation profile, I guess the 73-megawatt announcement that you guys did at the end of July is kind of the newest bit of news that we've had the least of bit amount of information on. Craig, I've just conservatively assumed that it's pretty back-end weighted, and you mentioned there might be scope to accelerate that. But is that just fair to assume that like all the other AI contracts, once it does deploy, it basically accelerates and is fully deployed relatively rapidly? And then when you mentioned site development, et cetera, are there any other constraints that we have to think about in terms of availability of gear, ability to be able to unlock and draw down on the power? Or is it purely just how quickly you can build the site?

Thanks, Tim. As it relates to the 73 megawatts that we announced most recently, obviously, that NVIDIA deal and having NVIDIA, obviously, as a partner on that was a really important strategic piece of business because that will continue to grow. That 73 megawatts is under construction, will be delivered in record time. Obviously, we're already forecasting a relatively short activation period. So, I think that the FY '27 number is clear that it is back ended. So, your observation is correct. And that is that, yes, it will activate quickly, but it will be back-ended for the time frame for that to turn on. But it will all turn on largely at one time. So the activation schedule itself doesn't have the type of ramp that you would have traditionally associated with a cloud style of deployment that would have been ramped or grown into. As it relates to your question on constraints, there are many in the long lead time category. One of the benefits of scale that play to our advantage as a company versus many that are entering the industry don't have the history or the balance sheet or the existing supplier relationships is when you're securing hundreds of generators, you can deploy them depending on where you need them at any given time. I think our size, scale, financial capability puts us in a very strong position to manage those constraints effectively. And as we move to more manufacturing style of construction methodology, it does allow the LEGO pieces to be moved around the construction program depending on the time requirements. The third point that you asked in relation to power, of course, as we have made very clear as it relates to the existing development program, certainly, the 0.5 gigawatt that we have contracted in FY '26, we have all of the certainty required in order to deliver those programs and the power that's required. We did put a reasonably succinct summary of the current legislative issues into the pack in order to help people understand what is certain and what is uncertain as it relates to federal and state government changes on the regulatory requirements and capital contributions and augmentation costs further upstream in the energy system that would be potentially required to be paid. So, I would just add to that, Tim, that none of those things are certain for the future developments for us as a company and for the industry broadly. There are no clear outcomes as a result of those determinations. But they certainly don't change the degree of confidence that we have over the existing program. They will be considered as it relates to all of the future program outside of things that we have currently contracted.

Great. I had another question, but recognize there are a few subpoints there. So I'll leave it for other people to go.

And the next question comes from Andrew Gillies from Macquarie.

Just a quick one on debt. Like high-level unlisted assets could comfortably run LVRs sort of above 60%. You obviously made some changes to the accounting for these hyperscale only or build-to-suit assets. Like it would sort of suggest that financiers are quite comfortable with higher gearing levels, especially relative to how the equity market might think about it. Can you just give us a bit of a steer on how you think about like reported group level metrics and maybe why the way that we would look at things might not reflect how your financiers are thinking about it? Just really interested in the cadence given you're talking about project financing and eventually JVCo, which sounds like it's closer as well. That would be great.

I'm happy to take that one. Thanks for the question, Andrew. And there's a few elements to it. To keep it brief, the way the financiers look at the capital stack is a little bit different to the way that, I guess, you may look at it traditionally. If you have a look at the subordinated capital, $1.7 billion of hybrids and the $750 million of subordinated notes, as far as senior lenders are concerned, that is subordinated to them. There is a reasonable amount of flexibility in that capital structure. So, they don't look at it like equity, but they do understand that it is subordinated to them, and therefore, that capital does not go directly to covenants. If you then sort of extend the question out to project finance and some of the other flexibility that we have in further financing developments, we have a certain percentage of our assets that are granted as security for the senior debt facilities. Within that, we have flexibility to move certain projects around. So for example, if we wanted to, we could exclude the S4 development from the guarantor group, which would allow us flexibility to fund the development of S4 through project finance. And then we're approaching essentially project financiers, with those project financiers looking at the S4 development underpinned by very highly creditworthy counterparty under long duration contracts and have a conversation -- have a sensible conversation with them as to how best to finance that project. Project finance is quite a healthy market. There's quite a lot of volume in that space. And the nature of our counterparties, the fact that we've got a very strong track record of developing our facilities on time, on budget, gives that market a lot of confidence and appetite to finance those sorts of projects. And as far as the senior lenders are concerned, again, that project is excluded from the guarantor group, and therefore, it doesn't go towards the covenants that we report and that they see. So hopefully, that gives you a bit more insight into how the lenders think about funding and how they get comfortable with our funding structure.

Yes. Certainly. And then maybe just moving to another quick one. NVIDIA overnight, obviously, started talking about the removal of revenue share agreements on antitrust concerns. I'm sure you guys have probably had some conversations there. It doesn't sound like it's affecting the capacity backstop. But if you could maybe talk to NVIDIA's support for the industry remaining on that capacity backstop side and maybe the impacts the lack of revenue share might have on these new neocloud customers, although I do appreciate they are a relatively small part of the contract book at the moment.

Thanks, Andrew. I'll take that one as well. The credit quality of NVIDIA is obviously very strong. I think they're AA rated and have provided significant support to the industry. We take a lot of comfort in the backstop that NVIDIA has put in place. The exact contractual nature of their arrangements with NCPs or neocloud operators is a matter for them. We're less affected by any revenue-sharing arrangements. We just take a lot of comfort in the fact that NVIDIA is a backstop. And obviously, that significantly helps get us comfortable with those counterparties, and it also obviously helps those counterparties finance their deployments as well.

And our next question comes from Siraj Ahmed from Citi.

Maybe 3 questions. Just first one, Craig, '26 was a record year, right, close to 0.5 gigawatts of contracting. Just in terms of '27, how are you thinking about it? Do you need to get the DAs for S7, especially the power allocation? And how are you just thinking of contracting this year? That would be great.

Thanks, Siraj. As it relates to '27, we don't need to do anything that we haven't already done. So, 100% of everything that we require to deliver our FY '27 numbers is already locked in. As it relates to '28, I would just restate my earlier comments that depending on progress during '27, that will help us understand the activation schedule and our ability to deliver on time or potentially deliver early in some of the '28 components. So as it relates to development approvals or power, we do not require anything in the '27 forecast that we don't already have. But obviously, as it relates to new developments in future years, '29 and '30, we would like to see the development approvals for S5, S7, M5. S5 and S7 have been in the state's significant development approval process for quite some time. And I won't forecast how long it will take the government to make a decision because that is their responsibility. But they have been in the approval process for considerable time, and we have addressed all issues that we are required to clarify in order for those development approvals to be granted. So, it's my expectation that those will move forward, and we will begin construction of them as soon as we possibly can. On our side of the program, early works, ground readiness, designs, long lead time, item procurement are all in place. So the moment that we receive those final development approvals, the programs to deliver S7 and S5 will immediately be undertaken.

Got it. Craig, can I just clarify? That's super helpful. So on S7 specifically, this new wind power requirements and the Transgrid allocation policy, you don't really expect that to hold S7 back from a time frame perspective?

Well, I think we don't know what we don't know at the moment. The government haven't made a clear policy position. Certainly in a regulatory sense, that's not law. There's a lot of policy discussion. So, we will see what the final position is on cost allocation. But I guess the point I would make, I think, is important, Siraj, regardless of what happens with the cost arrangements on network augmentation, doesn't really change anything for us insofar as how those costs would have historically been dealt with. We have been required to pay transmission infrastructure costs for as long as we've been developing data centers. Some of the further steps that the governments are looking to legislate on renewable components are already undertaken by our customers. So if we compartmentalize the responsibilities on the power regulatory side, the first category is the things that we have always controlled and been responsible for, the building and development of the substation, the funding of the transmission infrastructure in order to get the electrons to site. But our customers have always been responsible for the second category, which is the procurement of the electrons themselves. And on larger sites, that will only continue to be true. And so some of the regulatory considerations as it relates to the further upstream augmentation costs that may flow into the net generating cost that will be an output of the cost per megawatt to deliver power at the site will still be the responsibility of our customer. I think we have a high degree of certainty over things that we have always controlled. There is a little bit more uncertainty for our customers as far as their final net energy cost is concerned as it relates to that energy procurement determinations. But again, I'll just state that no determinations have been made and there's no current legislative proposal and those are the things that are being reviewed at this point in time. So, we will certainly continue to stay highly engaged. We're on all the relevant energy and advisory committees with government and through the industry association. I think we're well positioned at the table, providing good advice and feedback to government on what is practical implementation and obviously, practical to support the continued growth of the industry in Australia.