Good afternoon, and welcome to our earnings call to discuss Sharon AI's operating results for the quarter ended March 31, 2026. Joining me today is James Manning, Sharon AI's Chief Executive Officer, and Tim Broadfoot, Sharon AI's Chief Financial Officer.

I'll now take a moment to read the safe harbor statement. During the course of this conference call, we may make certain forward-looking statements within the meaning of the federal securities laws, including statements regarding our expectations, plans, prospects, strategies, future operating results, and financial performance.

Although they reflect our current expectations and are based on our current view of the industry and our business, they are not guarantees of future performance. These statements are subject to risks and uncertainties that could cause our actual results to be materially different from those expressed in these statements, and speak only as of the date of this call.

For more details on factors that could affect these expectations and cause these differences, please see our most recent Form 10-K and Form 10-Q and other SEC reports filed with the Securities and Exchange Commission, and available on the SEC's website and in the Investor Relations section of our website. Sharon AI undertakes no obligation to publicly update or revise any forward-looking statement, whether as a result of new information or future events.

In addition, during this call, we may discuss certain non-GAAP financial measures. Reconciliations to the most directly comparable GAAP measures and related disclosures are available in today's earnings release and/or on our Investor Relations website.

I'll now turn the call over to James.

Hello, everyone, and welcome to Sharon AI's First Quarter 2026 earnings call. My name's James Manning, and I'm the Chairman, CEO, and co-founder of Sharon AI.

This is our first earnings call since we listed on Nasdaq, and we plan to work with the following format: a brief summary of the quarterly highlights, a short overview of what Sharon AI does for those who don't know our business, some insights into specific events over the quarter, and a brief business outlook. Finally, we'll have some time for some Q&A at the end.

What an amazing first quarter it's been as we transition to life as a public company. Just looking through this chronologically, in January, I was appointed the CEO of Sharon, replacing Wolf Schuepert, who continues to work with the Sharon team on their North American projects.

In late January, we executed definitive agreements for the sale of Texas Critical Data Centers, and I'll talk to that a little bit more later on. In February, we listed on Nasdaq under the ticker code SHAZ, in a transaction which saw us raise $125 million, led by Lucid Capital.

We formalized our agreements with Canva, signing an MSA. In March, we signed GMI and then we executed the ESDAS contract.

Today, we are announcing our upgraded data center capacity to 100 megawatts by early 2027. Against all of this, we continue to build our team and governance structures, continually improving the way that we operate.

Importantly, we added Ben Adams and Drew Kelton to the board, both in a non-executive director capacity. Who is Sharon AI?

Sharon AI is one of Australia's leading neoclouds. We deliver AI-native HPC-grade compute infrastructure for our customers, with the goal of servicing three primary customer segments: hyperscalers, AI natives, and enterprise.

We do this by partnering with global leaders in industry to deploy the latest generation of GPU compute. While geographically deployed in Australia, our customers are truly global in nature.

Our recent customer wins reflect this story, with regional compute contracts for global compute needs. This reflects the desire to deliver mission-critical workloads in safe and secure compute environments.

Australia's geographic location in the world means it's well-suited to respond to GPU compute demands, being both close enough for latency, but far enough away from a security perspective. Latency remains an important consideration for GPU use, but arguably less so than most people think.

While Australia's geographic location in the world is somewhat isolated, the upsides really outweigh the downsides. We're only 81 milliseconds to Singapore, 130 to Hong Kong, 165 to the West Coast of the US, and 225 to the East Coast.

While latency is a factor in the way you think about your compute loads, the security benefits of being slightly remote in Australia and servicing that Asia Pacific region far outweigh the time differences. It's also worth calling out that some of these inference workloads are not as latency-sensitive as people believe.

I think the key message for us is that Sharon AI, while Australian-headquartered, we are Asia Pacific focused and really servicing compute loads for the entire globe. What is important for a neocloud to succeed?

In our mind, we're solving for one thing, and it's scarcity. We think about it in four broad buckets.

We think about the scarcity of compute, and being one of two NVIDIA Cloud Partners in Australia at scale, it gives us that sustainable advantage in accessing NVIDIA GPUs. It's not just the relationship with them, it's the relationship with our OEMs and our partners like WWT that help us procure and deploy that compute at scale and on time.

How you get access to power. The scarcity of power is a real component to getting GPU compute workloads on time and online.

Late 2025, we secured 50 megawatts with NEXTDC. Today, we've announced that we've expanded that total energy capacity in Australia to north of 100 megawatts for early 2027.

The regulatory environment makes scarcity even more complex, because not all countries have access to the latest generation of GPUs, and Australia's very fortunate to be a close ally of the US and have that ability to access the latest-generation GPUs. Finally, scarcity in talent.

People are so important to our business, and we think having a really experienced team like we do with high-performance compute expertise gives us a differentiator in the market all together. It's fairly simple, guys: the fastest-trained models equals the fastest inference, the best insights, and the most innovation.

All of that, though, requires accelerated compute-as-a-service, and that's what a neocloud is. We sit right in the middle between the hyperscalers and the AI natives or the enterprises and the consumers of that data.

We're the fabric that pulls together multiple vendors and solutions to drive the fastest accelerated compute. We achieve that through our partnership-driven model, leveraging the best-in-class expertise across the value chain, both global and domestic, across the hardware, power, security, connectivity fabrics.

By doing this, we reach the market significantly faster and more capital-efficiently than the traditional vertically integrated providers to deliver that GPU-as-a-service. With all that infrastructure in place, commercial success requires two additional components: a really strong go-to-market and sales strategy, and high-performance connectivity into those customer environments.

At Sharon, we don't try to own everything, and we have these great partnerships who add that value and ensure our service offering gives us those two huge advantages, which is the best-in-class expertise at every single layer, and the ability to move much faster and more capital-efficiently than vertically integrated players. With the scarcity solved and partners in place, it's on to how we contract.

Contract signing isn't where it starts, however. Before we contract, we spend a considerable amount of time behind the scenes working with our DC partners, OEMs, and our internal engineering team to find the best technical solution for the space we are operating in.

Once we have that design, we typically run a customer engagement process where we talk to both the available capacity and the compute format we think we should deploy. Once we find the right customer for that space, we agree principal terms and we sign these non-binding term sheets and move to an MSA agreement.

From contract signing, it can take us between 3 and 7 months to deliver that compute infrastructure to the customer. We take delivery, we test, deploy the compute, we verify it's all operational, and we hand it over to that customer.

The majority of our contracts are 5 years, and all of them are generally take-or-pay. While an average hourly rate might be inferred by these contracts, the take-or-pay nature ensures that we're paid for 100% utilization, which gives us clarity on our revenue forecast.

At the end of the term, we can then enter into new multi-year contracts, or use the compute for our own compute platform. What's actually converted this quarter?

I think we've announced three meaningful contracts. Firstly, it was Canva.

While it wasn't financially material contract win in its current form, it did highlight the power and the value of the Sharon AI platform, and how we could provide Canva with both speed and output improvements. We are really confident with that relationship and that we'll be able to expand those services over time.

We then won GMI Cloud. GMI Cloud is a AI-native inference cloud provider in the US.

Then our largest customer contract to date, which was ESDAS. ESDAS is an Indian-based IT service provider, where that contract was a $1.25 billion TCV over 5 years, with those contracted revenues expected to begin in September this year.

While not in this quarter, this week we announced a $950 million, 5-year take-or-pay contract with a global technology company with a major Asia Pacific presence, with revenue expected to commence in the third or fourth quarter of 2026. Cumulatively, this takes our TCV to more than $2.2 billion contracted, and we expect our exit 2026 revenue run rate of at least $470 million.

While we made great progress on customer wins, we also crystallized the gain from our project in Texas Critical Data Centers, or TCDC. In January last year, we established a JV with our partner, Rivera Energy Institutional.

Over the course of 2025, we purchased over 438 contiguous acres of land on two in two separate transactions. We signed a non-binding LOI with a hyperscaler.

We did a bunch of engineering design work. We then went in exclusivity with a hyperscaler.

Late 2025, we agreed to exit that. In January, we consummated that in a transaction that saw us realize $74 million, generating a material return for the business.

While these development opportunities are non-core to our business, they can be highly profitable. This won't be a material part of our business going forward, but we do reserve the right to continue to explore further opportunities in this space.

We can recycle this non-dilutive capital, and it can be an accelerator for our core GPU business. Now I might just turn to our outlook.

We remain incredibly confident about the ongoing growth of our core operations. We continue to see both Australian and rest-of-world client demand materially outweigh the GPU supply coming to market at scale and in a timely manner.

Our position of one of Australia's leading NVIDIA Cloud Partners, or NCPs, together with our strong partner network, positions Sharon AI to continue to capitalize on these strong GPU market dynamics. Today, we've announced this expansion of our data center capacity from 70 megawatts in 2026 to over 100 megawatts in early 2027.

This reflects our ongoing ability to secure additional power allocation and data center capacity from our network of data center partners. Lastly, but certainly not least, we've made significant additions to our technical and management team to support our expected growth.

Our people are integral to our ongoing success. Before closing, I'd just like to thank all of the people at the Sharon AI team, our partners, investors, advisers, and all the stakeholders who have supported us over the last 3 months in particular, but really over the last couple of years.

We're really excited by what we've achieved in the short timeframe that we've achieved it, but we're even more excited about the opportunities that still lie ahead. Operator, please open the line for some Q&A.

Hi. Thanks, guys, for taking my question, and congratulations on the successful IPO and the recent announcements that you've had so far.

Was wondering if you can talk about kind of your strategy for selling the remaining capacity that you have as well as the new capacity that was just announced here. Looks like the first couple big deals you've announced have been 5-year terms.

Should we expect to see more deals of that nature? Then I have a couple of follow-ups there.

Sure, Alex, and thanks for the question. Look, I would start with the contract duration because that's the easy one.

We are generally trying to target 5-year-plus contracts, and I think when you look at the ESDAS contract, for example, that's a 5-year contract with two 1-year options, it has the option to get out to 7 years in total. Critically within that contract, and this is what we're trying to replicate in other contracts, we're seeing the option period renewals to occur in year four, year four you've got to make the decision about year six.

Part of our view strategically around that is, obviously, we think the market price will be higher in earlier years, the earlier we get that contract renewal, we think the safer those extension periods will be for us long-term. Generally contracts 5 years is the target.

If we can get those extensions and options in them, that makes a lot of sense. I think more broadly and from the thematic that we're seeing is storage is becoming a really important factor in all these contracts.

We actually see storage as being part of the stickiness of these customer contracts as you get towards the end of term. We really expect to see those customer contracts renew on the back of this stickiness fundamentally because of the data we're ultimately storing on our equipment and devices.

Now that's sort of how I think about the renewals. I think renewals will occur naturally.

I think storage is a stickiness and part of the egress piece that makes it really logical for customers to stay with us, apart from us just having that compute capacity online. When you get to that fifth-year period, they've already been with you for a very long period, and they know you.

Hopefully, as we're seeing with some of our early customer contracts, they're already talking about that expansion. This is definitely something we're thinking about from the contract renewal piece.

Then the other question was talking through our capacity and how we're looking to contract that additional capacity. I think if you think about how we came to market in February, we said, look, 55 megawatts.

We announced the ESDAS contract that took us to 70 megawatts. Today we're telling the market we've contracted an additional 30 megawatts.

I think we've been very transparent that we do have a bit of a pipeline of energy assets, and we're sort of matching those to customer contracts as we're getting through customer contracts as well. I have at previous times described how we do that contracting process.

Broadly, when we get a space like, for instance, M2, which is now our next big project that we're focused on, which is 40 megawatts in the public domain, it's a hugely valuable asset to get 40 megawatts coming on in Q4 in Australia, which is really a Tier 1 country, is very, very easy asset ultimately to bring to market and sell to end customers. It's extremely valuable in the market right now.

We're actively having conversations around those megawatts and how we place them. We're through the design process for that compute cluster down in Melbourne, for instance, and we're working through how we just match the right customer with the right customer offtakes and get that contracted up.

I think you'll see us contract that in the near term, and then we'll announce that and we'll talk to the other bits of capacity. With the additional 30 megawatts that we've announced today, some of that is in Australia, some of that is in New Zealand.

We are thinking about, with specific customers for specific end-use cases, how those megawatts are going to come online in that space.

That's really helpful. Then you mentioned M2.

Just kind of curious what your opportunities are with that, right, having a contiguous 40 megawatts to be able to mark. Obviously, you could be selling that in 5 or 10-megawatt chunks like some of the deals we've seen so far.

What kind of opportunities are there for you to use a 40-megawatt block that you might not have had with some of the smaller allotments?

I think you've hit it spot on. I think 40 megawatts is a size that's likely to go to a large single customer.

I think, when we talk about our customer profile, we've always said we've had hyperscalers in the pipeline, we've got AI natives, and we've got enterprise customers. I don't think 40 megawatts is going to be broken up into a lot of small enterprise customers.

It's going to fit into the first two categories. Either a hyperscaler or a large LLM or an AI-native enterprise will be the logical offtakers around that sort of contract.

Slightly different profile of custs, slightly different use case. I think we're looking at a lot of this very much with the lens around storage, and we want customers that are going to grow not just in the GPU compute, but also grow from a storage compute perspective.

We're definitely having those conversations with customers where we think we can add value over time into these contracts, because we're seeing in those early customer contracts that the storage demands are exploding, and we're trying to find customers where we find not just from the face value contract, but that expansion opportunity through the customer lifecycle.

That's helpful. Then lastly, great to see that you secured another 30 megawatts of power.

Interesting that some of that's going to be outside of Australia in New Zealand. Curious what kind of a line of sight you have to additional power availability down the road beyond the 100 megawatts, and are you actively looking at potential data center opportunities in other countries outside of Australia and New Zealand, just trying to understand the roadmap here?

Look, I I won't talk to, we I think what we've done very, very well is trying to not get over our skis the entire time we've been out, since we've been listed, and and when we've come to market and spoke about this. As I said, back to back to the 55 in February, customer announced 70, 100 today.

I think, we we recently did a a non-deal roadshow North America, caught up with a lot of investors that participated in the IPO, and the and the message was give us give us a little bit more guidance about megawatts, and that's what we've tried to do. We tried to, meet some requests for for some forward pipeline.

We're confident we have a lot more megawatts in the pipeline, ultimately. We're trying to match those part that those megawatts for when they can come on, who the end customer is, and and really go through a bit of a matching exercise around those megawatts to current customer demand.

Guiding towards, the megawatts we're announcing are are, near-term in nature, and and we're really trying to find, near-term customer contracts attached to those near-term megawatts. We're going to stay in our region geographically.

What we are seeing is that demand for for GPU compute in the region is absolutely outweighing the supply. We're just focused on trying to find those premium global customers that are operating in Asia Pacific, and specifically customers not with just raw GPU compute, but with storage requirements as well.

That's really helpful. Thank you very much.

Hi, guys. Thank you for taking my questions.

Nice to see the additional contracts. I want to spend a minute talking about kind of your visibility into GPU deliveries.

It feels like you guys are bringing on compute quite quickly given kind of when these deals are being announced, and what we're hearing is anything in 2026 is effectively going to get sold. I guess what advantage do you guys have in terms of announcing a deal, call it yesterday, and saying in 4 months, you're going to be able to start generating compute revenue?

I think there's no surprise that capacity is constrained in 2026. What we're seeing for OEMs, depending on what form factor you're looking at, and look, we're generally building out in GB300s, we're sort of seeing that 12 to 14 weeks on GPU.

Network's a little bit longer, but we've got really good supply chain partners, and our relationship with WWT really helps that. Depending on ultimately who the end customer is, there's obviously some incentives to try and get these things online as fast as possible.

For instance, we do try and get ahead of some of that supply chain aspects. We have pre-ordered network for the 8k B300 cluster, which enables us to bring some of that timeframe in.

The longer lead items, we are trying to get ahead of from an engineering perspective and procurement perspective. We're feeling pretty good about our delivery at this point in time, and just trying to make sure we manage that to customer expectations.

Awesome. Then maybe just on the CapEx and funding side, I think between the two recent deals, we're looking at maybe $700 million or so kind of money to be spent on GPUs.

When should we expect the timing of that to kind of hit the financials?

I think from a timing of the CapEx, well, we've already incurred some of the CapEx. As I've sort of indicated, we've gone out and we've spent and pre-ordered storage, for instance.

We've pre-ordered network. Some of those longer lead components we've tried to get ahead of the curve.

We're very fortunate with some of the storage deals that we did. We got in prior to some price rises.

We got ahead of some of that CapEx. We've bought about 70,000 GPUs worth of storage to get ahead of the curve on those things.

Broadly, CapEx, if I could guide you, about $39 million a megawatt at present, all in. That's with all the non-recurring costs, data center inclusive for that.

We're seeing a bit of movement there, but well within the numbers that we planned for and we've sort of disclosed to. We're seeing very good order timeframes around that.

Then I think to the capital requirement, we're funded fundamentally from all the equity components for all the currently announced contracts. Between the Oaktree note, which is due to close early next week, and the IPO proceeds plus customer deposits, we're in a fairly good position from that.

We have appointed Jarden domestically here for the GPU financing route. We have several term sheets, and it's just a matter of us matching the right term sheet to our ultimate financing desires, I think.

We're very focused on that. We're very, very well-advanced on it.

As I've sort of said, we're materially advanced enough that we're negotiating term sheets around the debt piece.

Awesome. Then maybe just to hammer home on one point, the additional 30 megawatts you guys are talking about by early 2027, should we be expecting, call it by early 2027, 100 megawatts, give or take, of kind of contracted deals to be up and running?

I think you could guarantee that will be signed, the 100 megawatts.

Love to hear it.

Very confident of signing the 100 megawatts.

Perfect. Really appreciate it, guys.

Thank you.

Hi, guys. Thanks for taking my question, and congrats on progress.

I thought we could talk about pricing trends that you're seeing in general for GPU hour, and just customer demand in general for inference. You mentioned storage, how should we think about customers that do require storage versus just, that would be the bare metal compute?

Appreciate it.

Look, customers, each customer's got their own piece and what their own needs, but if I talk to you on a megawatt basis, we're seeing around that $15 million a megawatt from a revenue perspective. That's with an allowance for some storage, and so if you think about the broader NVIDIA reference architecture, 3 PiB of storage to a 1k cluster is a broad rule of thumb.

What we're now seeing from those customer contracts are requests for 6, 9, and 12 PiB. So 3, 4x the storage component compared to the standard GPU component on the compute clusters.

For us, the storage is quite a high-margin component, and it's a real value-add proposition within the business. If we think about that, it's really an opportunity for us to get additional revenue, but it's super additional revenue on these contracts not just in year one, but what we are seeing is customers get 6 months into a contract and go, we need some additional storage, what are we going to do?

By then they're already hooked on your GPU platform, it's quite a easy upsell at that point in time. I'd say that about broadly the technology piece.

As you've highlighted, I think everyone knows, market pricing's very strong at the moment given the scarcity of compute, if you can bring on capacity this year, you're in a very strong position. That materially outweighing demand of supply dynamics means that we've seen really strong customer pricing, and the customers are really trying to lock in those longer-term contracts because I think what they're seeing is the spot pieces are a lot scarier.

Appreciate that. Then also if we could talk a little bit about the tech side and your relationship with VAST Data.

How important is that relationship there, and what exactly does it provide you?

VAST is the storage piece that sits alongside the GPU piece, and we've got a phenomenal relationship with VAST. They've got an incredible technology stack, and we are looking to how we further integrate that relationship with VAST.

They have some amazing tools that allows us to really allow that AI compute access into the data layer to work really, really efficiently. The storage becomes that story of 2026 in our mind.

You need more and more of that specialized storage and storage tools that integrate very well into your inference engine to get those workloads working really efficiently for the end customer. We're making great margins on storage generally, and we want to continue to see these customers want more storage for every deployment, because it's a great value add for us as far as the business.

Great. Appreciate that.

Hi, and thanks, everyone. I just want to take a moment to say thank you for joining our first Q1 2026 earnings call.

That's the first one since we've listed on Nasdaq, and it's one of many more to come. I really want to thank our staff, our team, all our support from internal, external stakeholders and our shareholders.

If I look back at our business, it's expanded significantly, and we continue to see both Australian and rest-of-world demand materially outweighing supply. Having upgraded our expected data center capacity twice this year from the initial 55 to 70 and now to 100 megawatts, and with the funding in place with the previously announced 350 convertible note, we're well-placed to continue to grow our customer base over the coming quarters.

We just want to reiterate that we continue to see that strong demand across enterprise, hyperscale, research, government, and AI-native sectors throughout Australia and the APAC. We're really looking forward to future announcements and getting together with everyone on the next earnings call.

Thanks a lot.