Bit Digital, Inc. (NASDAQ:BTBT) Q2 2024 Earnings Conference Call August 20, 2024 10:00 AM ET
Company Participants
Cameron Schnier – Head, Investor Relations
Sam Tabar – Chief Executive Officer
Erke Huang – Chief Financial Officer
Conference Call Participants
Mike Grondahl – Northland
Kevin Dede – H.C. Wainwright & Co.
Joe Gomes – Noble Capital
Chris Sakai – Singular Research
Operator
Hello, and welcome to the Bit Digital Second Quarter 2024 Earnings Conference Call. Good morning, good afternoon and good evening depending on where you’re joining us from. Thank you for being here. [Operator Instructions] Also, as a reminder, today’s conference is being recorded.
I’ll now hand the call over to your host Cameron Schnier, Head of Investor Relations at Bit Digital. Cameron, the floor is yours.
Cameron Schnier
Thank you. Good morning and welcome to the Bit Digital second quarter 2024 earnings call.
Joining us on the call today are Sam Tabar, Chief Executive Officer; and Erke Huang, Chief Financial Officer.
Before we begin, I would like to remind all participants that some of the statements we will be making today are forward-looking. These matters involve risks and uncertainties that could cause our results to differ materially from those projected in these statements. And therefore refer you to our latest 20-F filing, yesterday’s 6-K filings and our other SEC filings.
Our comments today may also include non-GAAP financial measures, additional details and reconciliations to the most directly comparable GAAP financial measures can be found in our 20-F filings and yesterday’s 6-K filings, which are on our website.
After our prepared remarks, we’ll open the call for questions. If you like to ask a question, please hit star, one on your keypad.
With that [indiscernible] covered, I’ll turn the call over to Sam to discuss our performance. Sam?
Sam Tabar
Thank you, Cam. Ladies and gentlemen, thank you for joining us on the call today.
Today, I will dive into our second quarter results, discuss some of our ongoing strategic initiatives and provide some color on what we envision as the future for Bit Digital. Erke will then provide more detail on our financial results, and then we will open the line for your questions.
Our second quarter results were solid. We are encouraged by the progress we’ve made in diversifying our business. Our revenue more than doubled from the prior year. Our gross margins expanded by over 1,000 basis points. Adjusted EBITDA and EPS were impacted by an unrealized loss on our digital asset position. The first full quarter contribution from our HPC business is well timed. It helps offset the decline stemming from the post halving reduction and block rewards.
Our balance sheet remains pristine with zero debt. It affords us significant flexibility to prudently invest in the most accretive opportunities. The groundwork is firmly in place for us to realize our long-term vision of creating a diversified, durable platform that generates consistent cash flows and significant returns.
On the mining side of the business, we were mostly quiet during the quarter. That was by design. We were keen to wait and see how the environment shook out following the halving. Network hash rate proved to be resilient despite the reduction in block rewards and hash price subsequently fell to new all-time lows.
Our active hash rate ended the quarter at approximately 2.6 exahash. This is a slight decrease compared to the end of Q1. The decline is attributed to curtailments in Iceland and energy-saving measures at sites where electricity prices were unusually high. We produced 244 bitcoins during the quarter. This is a 23% decrease from the prior year, driven by increased network difficulty and a reduction in block rewards.
For the quarter, our average electricity price per bitcoin was approximately $0.047 per kilowatt hour. Our electricity price per bitcoin was $29,300. Our total cost of production, defined as electricity and other hosting fees divided by bitcoin production, amounted to approximately $43,200 for the quarter.
Profit-sharing fees amounted to around $11,100 per bitcoin for the first quarter. On profit-sharing fees, they vary based on the amount of gross profit per bitcoin we received. This fee accounts for the fact that we don’t own our own mining infrastructure. Other companies have charges under different line items that account for the costs of running a mining site, including labor. This is often overlooked by comp sheets comparing production costs.
Our average realized bitcoin price during the second quarter was around $65,800. This leaves around $22,700 in gross profit or a mining margin of around 34%. However, current economics present challenges, especially with the depreciation costs. This makes it difficult to recover the investment in new mining rigs within a reasonable timeframe.
We’re bullish on the future of bitcoin, but we are realists. The economics are difficult right now, and we are choosing not to aggressively grow into that backdrop without higher conviction that we can make a justifiable return on new equipment. Our principal bottleneck to exahash growth right now is our own menu of growth options. We see better investment options in HPC at the moment.
So, based on what we see today, we think it’s unlikely that we will hit 6 exahash by the year-end. Material exahash growth for us into year-end would require either a significantly better view on mining economics or a significant deterioration in the opportunity set we currently see on the HPC side. Near term, our focus on the mining side is on improving the efficiency of our active fleet and lowering production costs.
Switching gears, the HPC segment generated $12.5 million of revenue at 63% gross margins during the second quarter. That was the first full quarter of operations for this business. We recently announced a binding term sheet with a new customer, Boosteroid, the world’s third-largest cloud gaming provider, following Microsoft and NVIDIA. We believe that this represents a key end market expansion for our HPC business beyond AI applications or LLM training specifically.
This term sheet carries a locked five-year term. It also provides for up to 50,000 GPUs over a five-year period following the signing of the MSA. The entire contract, if Boosteroid elects to deploy the entire 50,000 GPU allotment, would be worth more than $700 million of revenue for Bit Digital over the five-year contract subject to market conditions. The initial deployment will be a test [indiscernible]. This will likely represent a $2 million to $3 million annualized revenue opportunity in its starting form. We expect that initial deployment to take place over the next few months and expect to scale that contract throughout 2025. We’ll be able to provide further details on the timing and build-out of the contract after we execute the MSA, which we are working hard on completing as soon as possible.
We look forward to growing alongside Boosteroid and helping them achieve their goal of becoming the global leader in cloud gaming. This deal has great growth potential and aligns with our business model of growing alongside our customers. Our pipeline in HPC remains strong and continues to grow.
The major bottleneck to date in terms of completing incremental contracts has been bandwidth; we simply haven’t had the adequate manpower to move highly customized customer deals through the finish line. But, we have begun to solve for that need. We made our first full-time hires in the HPC segment to help lead the revenue process. We will formally announce those hires very soon, and we’re excited for them to ramp up and lead the sales charge. These people have had a very impressive track record in growing an HPC platform and customer base. We are also working on additional hires to complement our existing strengths and fill in strategic gaps. Beefing up our tech stack is a key priority on the hiring front near term.
For our anchor customer, we previously announced that we were expecting to install an incremental 2,000 H100 and begin revenue generation on those units in the late third quarter. In late July, our customer brought in some new technical talent that ultimately led to the review of their future hardware portfolio. As a result, our customer asked us to pause the H100 order while they assess the possibility of upgrading the servers to newer generation models. We reached an agreement with the OEM to delay our server order and have paid only for certain longer lead time networking equipment. It’s important to note that there is no change in this customer contract. The contract remains in full force and effect.
The biggest change from growing from a bitcoin mining to HP services is the customer-facing aspect within HPC. You don’t have to worry about keeping your customers satisfied in bitcoin mining. There are no customers in bitcoin mining. We’re very keen on keeping our customers in HPC happy, and we want to help position them for long-term success. So, when our customer asks us to pause an order, while they assess their technological needs, we oblige them with that courtesy, despite the potential negative impact that the delaying the onset of the revenue may have. We will have greater clarity on the customers’ plans over the next several weeks, and we’ll provide an update as soon as feasible.
Recent headlines suggest that NVIDIA’s Blackwell B200 chips may not be available until early 2025. So, there is a chance that if our customer opts to upgrade to the contract, to upgrade the contract to those chips, the revenue contribution to Bit Digital would be delayed until such time until we can procure those chips. However, we are confident that we can secure an early allotment of those chips based on the conversations we’ve had with some of our key relationships. There is also a chance that our customer just decides they’d rather continue forward with the H100 or possibly the H200s. I would note that the lead times for the H100 have come down quite significantly, and we’re confident that we can procure those units today in about three to four weeks’ time.
Lastly, if our customer does decide to press forward with the Blackwell B200 chips, revenue recognition will likely be pushed into 2025, but we would expect the scope of revenue to be greater than the $42 million annual figure stipulated under the H100 contract. Notably, and in case there are any doubts of how serious our customer is, we received a non-refundable $30 million prepayment from our customer earlier this month.
Our growth pipeline remains quite strong. We continue to believe that we’ll be able to reach our $100 million annualized revenue target by the end of 2024, even if the 2,000 GPU expansion deployment with our current existing anchor customer is close to 2025. We expect the remainder of the year to be a very busy one for Bit Digital.
As previously noted, we believe that having our own infrastructure would help mature our HPC business, and we are actively working on ways to solve for that. We are evaluating opportunities that we think could complement our existing business well. It’s hard for us to provide any incremental detail, but I’ll say that this is an area that we view as a very attractive potential use of our capital. The bottom-line is that we’ve been deliberately strengthening our balance sheet and have considerable dry powder to deploy into opportunities that we view as highly accretive and strategic. We expect to share these exciting developments on the capital deployment front on our next earnings call.
Lastly, we materially increased our ETH position late in second quarter by bitcoin conversions. As of the end of June, we had over $90 million worth of ETH compared to approximately $37 million worth of bitcoin. We remain bullish on ETH long term. We believe the market will soon appreciate its versatility and sound monetary policy. As it relates to the recent approval of spot [ETTFs] (ph), we are happy about the institutional inflows that they have brought and will bring into ETH. We also think they will be helpful in terms of bringing both awareness and education around ETH to the broader market. However, one key aspect that the ETFs lack is the ability to stack ETH — to stake ETH. Bit Digital continues to have that advantage over these new vehicles.
I’ll now hand it over — I’ll now hand over the line to Erke, who will discuss our financial results.
Erke Huang
Thank you, Sam. I will now discuss certain financial results for the second quarter of 2024.
Total revenue was $29 million, a 220% increase compared to the prior year. The revenue increase was driven by the first full quarter of our HPC services business and by higher realized bitcoin prices. Our bitcoin production decreased 23% year-over-year to 244.2 bitcoins. The decrease was driven by an increase in bitcoin network difficulty and by the reduction in block rewards following the April halving. Bitcoin mining revenue increased 80% from the prior year to $16.1 million due to higher bitcoin prices. Our HPC services business recognized $12.5 million of revenue during the quarter. We earned 109.4 ETH from native staking rewards during the quarter, representing approximately $374,000 in revenue.
Our total cost of revenue was $15.2 million compared to $5.6 million the prior year. The increase was driven by an increase in our active mining fleets, higher bitcoin network difficulty, the start of our HPC business.
Gross profit increased more than 300% from the prior year to $13.8 million. Mining contributed $5.5 million towards gross profit and HPC added $7.9 million. The total gross margin expanded 1,100 bps from the prior year to 48% with the induction of — with the addition of the HPC offsetting lower bitcoin mining margins related to increased bitcoin network difficulty.
General and administrative costs were approximately $5.5 million compared to $5.4 million during the prior-year quarter.
Depreciation and amortization expense was $8.4 million for the second quarter compared to $3.7 million last year, with the increase driven by larger mining fleet and our GPU fleet that was deployed in early 2024.
Adjusted EBITDA was negative $3.8 million for the quarter compared to $1.9 million last year, while gross profit increased year-over-year and G&A was roughly flat. We recognized $11.5 million unrealized loss on digital assets, which reduced the adjusted EBITDA number.
GAAP earnings per share was a loss of $0.09 for the quarter compared to a loss of $0.03 the prior year, with the change in digital asset prices being the primary driver of the decrease.
Turning to our balance sheet. We held approximately $61 million of cash and restricted cash as of June 30 this year, and our digital asset position was worth approximately $130 million. Total assets amounted to $315 million at the end of the quarter, while shareholders’ equity was $295 million.
Our balance sheet remains debt-free, but we continue to evaluate potential debt financing options as a means to accelerate growth of our HPC business. Given the strength of our balance sheet, we can afford to be patient on that front.
CapEx was approximately $5 million during the quarter. The CapEx was used for the purchase of approximately 1,150 new miners and for deposits on certain HPC equipment.
I will now turn the call back to Sam for closing remarks.
Sam Tabar
Thank you, Erke.
Over the past several months, we’ve seen a large increase in investors that are now interested in bitcoin mining stocks because of the HPC angle. Many are curious about AI’s future. This is like asking where bitcoin’s price is going to be. I believe that AI will transform the global economy and bitcoin will be worth significantly more in the future. I just can’t predict exactly when or to what magnitude this will play out. What I do know is that both AI and bitcoin will require significantly higher computing power in the years ahead. We want to be positioned to supply that computational power on the HPC side, while maintaining the flexibility to opportunistically increase our mining output.
We might not be the largest bitcoin producer, but size isn’t everything. Exahash and production are hollow metrics without the underlying returns profile. We’re okay with not leading in production because the real value lies in making the right moves at the right time. For now, we’re seeing better opportunities in HPC. The barriers to entry in mining aren’t that high, and we can scale up when it makes sense for us to do so.
When it comes to AI, I can’t tell you what the next killer AI app is going to be. AI is evolving very fast, and new applications will continue to pop up. People will never have imagined Uber or Twitter or Instagram when the Internet was invented. I’m nearly certain that new and unexpected use cases for AI will emerge, and that in five years, the world will require far more high-performance computing power than we need today.
We’re not just focused on AI tailwinds. The binding term sheet we announced with Boosteroid today, a cloud gaming company, is a good example of how we’re diversifying. That deal also provides for significant growth, which is ultimately the fabric of our business model to align ourselves with customers and grow in a mutually beneficial way. So, I might be disappointed about the temporary delay of our anchor customer GPU expansion, but these things happen in a customer-focused business. Once the revenue comes in, the margins will likely be even better if they decide to upgrade the new hardware models. We remain long-term greedy.
We’re not taking a myopic view on the business. We want to grow alongside our customers for the next decade and renew contract after contract. Long term, I’m confident that this approach will maximize shareholder value. We’re working on a lot of very exciting things right now, and I wish I could share more details, but at the best of my counsel, all I’ll say for now is stay tuned.
With that, I would like to open the line for some questions.
Question-and-Answer Session
Operator
Thank you. [Operator Instructions]
Mike Grondahl
Hey, Sam, this is Mike Grondahl. First question is really on the $700 million potential Boosteroid contract. It looks like it starts at about $13 million in revenue over five years. And you talked about it scaling in 2025. Is there any minimum levels? Or how should we think about that scaling?
Sam Tabar
So, we expect to be able to provide greater detail on the growth cadence over time. The initial allotment is a starting quantity. And upon successful deployment, we expect to start fulfillment incremental deployments. However, this contract is designed for us to grow in tandem with Boosteroid. We’re not looking to deploy the full run rate out of the gates. Our baseline for 2025 is probably to get through around 30% of total deployment. Boosteroid’s growth trajectory and our own capital allocation plans are the key variables to the cadence of that deployment?
Mike Grondahl
Cool. Okay. Well, hey, 30% is at least something that go on and it’s a lot. Secondly, you kind of talked customer one in the pause. I think that’s the right approach, and they already have 2,000. But you said too that you can still reach your $100 million run rate by year-end ’24 without customer one. Could you talk a little bit about your HPC/GPU pipeline? Is it a couple large ones like Boosteroid, or is it numerous small ones? Just trying to get a flavor for that pipeline.
Sam Tabar
Yeah. So, we announced Boosteroid yesterday. Other clients are very close to the finish line. Our new hires will help accelerate that closing process as mentioned, that was a pretty large gating item, just not having the manpower to deal with all these reverse inquiries, and we continue to have that. But we do think we can hit that $100 million mark by the year-end even with that expansion being paused potentially to 2025. And by the way, the decision hasn’t been made yet on that.
Mike Grondahl
Sure.
Sam Tabar
But beyond that contribution from Boosteroid, we do have several clients close to closing, and we’ll get to pretty close to that number. So, we’re pretty confident. We’re not revising that target. We still sticking to about $100 million.
Mike Grondahl
And did your new Head of Sales actually start, or is that person coming on soon?
Sam Tabar
No, he started.
Mike Grondahl
Okay. And then, one more. You talked a little bit towards the end of your prepared remarks about acquisition and whatnot. And in the past, you’ve mentioned possibly you’re looking at a data center. Is that what you’re referring to there kind of that’s been where you’ve been spending some time and you couldn’t say more? I’m just trying to connect the dots there a little bit.
Sam Tabar
Yeah. No, we understand. This is ongoing. We don’t want to weigh too much into the details about what we’re seeing out there and what we’re targeting. We can’t provide too much detail on our process, but it’s something we’re looking at. Yeah, so I can’t say more than that, but I can say that owning our own site certainly on the HPC side of the equation will allow us to offer different services like colocation and on-demand compute.
Mike Grondahl
Got it. Yeah, that would be great. And then maybe lastly for Erke. In Q2, Erke, you guys hit the ATM roughly 16 million shares. Since June 30, about another 9 million shares. How are you attempting to finance this going forward? There’s a bunch of CapEx. I know you’ve talked about a credit facility in the past. But how are you just thinking about using the ATM versus getting a credit facility in place?
Erke Huang
Thanks for the question, Mike. As we always expressed the use of ATM, we are very cautious. But as we all know, this business requires a large procurement down payment, so it’s always good to have the best balance sheet to support our growth. And at the same time, we are talking to various credit facility providers to determine what’s the best term for us to deploy. So, we’re working on that as well.
Mike Grondahl
Okay.
Sam Tabar
Let me — if I could just add to that, Mike, we are — we continue to evaluate term sheets on debt financing. And in terms of the indicative terms that we’re seeing, we see some reasonable terms, but nothing for us to be super excited about to accept. And we haven’t been forced in a position to accept less than optimal debt financing terms, and that’s important.
Mike Grondahl
Got it. Hey, thanks guys, and congratulations.
Sam Tabar
Thank you.
Erke Huang
Thanks.
Operator
We will take our next question.
Kevin Dede
Hi, Sam, Erke, Kevin Dede. Can you hear me okay?
Sam Tabar
Hi, Kevin. Loud and clear.
Kevin Dede
Good. Okay. Thanks. Just first on the bitcoin side. Obviously, you spoke to decline in your self-mining hash rate. I’m just wondering what you’re just thinking about is doing with those machines. And then I think if I understood or heard correctly, new machine purchase, maybe you could just sort of fill in the blanks on where you think your hash rate target will be at year-end given you’re kind of pulling back on the reins there on economic conditions…
Sam Tabar
Sure. Yeah. Kevin, that’s a good question, understandable. Here, the long-term plan for bitcoin mining for us is to remain opportunistic. It’s not a business that we believe that should be consistently invested into regardless of prevailing economics. So, we have different levers. So, we’re in a very fortunate position on that. But you could expect potentially some modest growth, but we can’t give any guidance on that because, again, we wanted to just remain opportunistic. We want to see how the market conditions have evolved. But at current economics at current economics, we would likely not spend more than $5 million to $10 million on ETH upgrades, on the current economics. And that’s been going sideways for a while, the economics there. And also on one hand, it would be — we would have to see a deep acceleration in the HPC demand that we’re seeing. And we’re currently see a good amount of capital to fund opportunities for the HPC side.
So, I think — I mean that’s the answer. I don’t know if Erke wants to add to that, but that’s my thought about it.
Erke Huang
Yeah, I agree with you, Sam.
Kevin Dede
Yeah. So, you did make a machine purchase. And then, what’s the thinking regarding the machines that you’ve taken offline, the ones that perhaps are much higher levels of, I guess, inefficiency?
Sam Tabar
Well, if they’re not economically viable for certain models, they go into warehouse or we sell them, which is something we’ve done.
Kevin Dede
Very good. Okay. Then sort of switching gears and looking at HPC hosting sites, my thinking would be that Boosteroid facilities would need to be close to metropolitan areas versus, say, customer one, which I think is mostly inference. Can you talk about how you’re lining up facilities to meet the obligations of the Boosteroid deal?
Sam Tabar
Yeah. I’d like to pass that to Erke or Cam. I’ll add to it after.
Erke Huang
Yeah, we have different locations, including locations in the US as well as Europe to deploy those equipment.
Cameron Schnier
Yeah, we’ve already sourced the data centers, Kevin. And you are right, they are close to metropolitan areas. It’s like a lot more akin to the inference side than training. So, they’re not spread out in very remote areas. They’re close to more densely populated regions, which follows the cloud gaming industry where latency is obviously a paramount concern.
Sam Tabar
Yeah, I would add that as the size — sorry, Kevin, go ahead.
Kevin Dede
No. Go ahead, Sam. You go, please. Thanks.
Sam Tabar
Well, I was going to say, at the size we’re looking at data center capacity hasn’t been a major issue. We haven’t lost any deals because we couldn’t find capacity. Although we do think owning our own site will make us more flexible, and we’re working pretty hard right now on ways to solve for exactly that need.
Kevin Dede
How would you characterize pricing on the sites that you’ve located for Boosteroid versus what you did for customer one? And what can you say to…
Sam Tabar
And I’ll pass that to Erke.
Kevin Dede
…in terms of trends?
Sam Tabar
In terms of what, sorry? Repeat that, Kevin?
Kevin Dede
Yeah. Just general cost trends on — because we hear there’s huge demand for HPC and that would obviously raise a flag of cost trends on the power side.
Sam Tabar
Yeah. Just before I pass that to Erke, as mentioned, data center capacity has not been a major issue, and we have not lost any deals because we couldn’t find capacity, just to mention. But I’ll pass the rest to my colleagues.
Erke Huang
Yeah. I mean in terms of the details of the deal, we will certainly provide more clarity after we sign the MSA. So, we will provide shortly.
Sam Tabar
Yeah, we’re working pretty hard on completing and executing that MSA very soon, and we’ll be mentioning that to the market.
Cameron Schnier
Kevin, I think, directionally, you could assume that data center requirements for training a large language model are a lot more extensive than for cloud gaming purposes. Directionally, you wouldn’t expect it to be as expensive.
Kevin Dede
Okay. So — I’m sorry, Cam, rewind that a bit. You’re saying that training is less [indiscernible] the requirements to meet train demands are less expensive, which would be my assumption?
Cameron Schnier
The type of data center and ancillary equipment you need for a two or three data center to forge for training in LLM are far greater than the data center requirements that you would necessitate for cloud gaming deployment, that’s sort of the essence of that. So like directionally, this deployment would be less expensive than the training side.
Kevin Dede
Interesting. Why is that because the compute isn’t as high? I mean, it just seems to me that you’d want that same low latency and redundancy.
Cameron Schnier
Yeah. I mean, certainly, just like the CapEx going into the data center isn’t as high necessarily. We can have a long-form discussion of this offline and we do bring in some heavy headers for you.
Kevin Dede
Okay. Thanks. Last question for me. Just help me understand the difference between the binding term sheet and the LOI and the MSA and what sort of financial obligations are associated with a fracture of any of them?
Sam Tabar
Yeah, it’s worth mentioning, this term sheet is binding. It’s not just like some sort of intentional aspirational LOI in the binding term sheet. And the MSA we’re currently working on, so that’s the binding term sheet has the high-level stuff, the economics and so on and the MSA has the more legal nitty-gritty, and that will be announced pretty shortly. Does that answer your question? I feel like I missed something, Kevin. Can you…
Kevin Dede
Yeah. There’s just one nuance in it, Sam, is what would happen if either party decide to not consummate the deal?
Sam Tabar
I mean, there’s legal recourse, but we’re not in the business of — there’s legal recourse, but we don’t believe that’s going to happen. If we thought that was going to happen, we would not have announced the binding term sheet.
Kevin Dede
Very good. Okay. Appreciate the color, Sam. Thanks very much…
Sam Tabar
But you’re right though in theory, you’d expect, but we’re very confident and we’re very close to the clients.
Kevin Dede
Well, good. Congratulation. We’re looking forward to see how that progresses.
Sam Tabar
Thank you.
Operator
And we’ll take the next question.
Joe Gomes
This is Joe Gomes from Noble Capital. Can you hear me?
Sam Tabar
Loud and clear.
Joe Gomes
Thanks, Sam. So, on the Boosteroid contract, maybe just conceptually, can you kind of relay what kind of the margin on that versus the anchor HPC contract margins? How do they compare?
Sam Tabar
The margins are still nice in that, but they’re not as good as the margins on the H100s. Having said that, if our anchor client — not that you asked, but I do want to emphasize that if our anchor client decides to go with the Blackwell B200s, I mean, sure, there is a delay, but we’re long-term greedy. We’ll have even better economics and even better margins on that particular hardware. But with respect to the hardware related to Boosteroid, the margins are still quite good. We will — the MSA is yet to come out, but it will not be as high as the H100s.
Joe Gomes
Okay. Thanks for that. And on the mining business, can you talk a little bit about how the gross margin for that segment flowed over the quarter? Was it pretty flat over the quarter or were the — was the gross margin at the end of the quarter significantly different than what it was at the beginning of the quarter?
Sam Tabar
Cam, do you want to take that?
Cameron Schnier
Yeah. I don’t know if there’s necessarily much variation beyond that. I mean, I guess, there could be certain instances where it fluctuates, but really it’s pretty steady state.
Joe Gomes
Okay. That’s good. And then, I was going through the filings and fleet efficiency decline sequentially. And I was just wondering if you could talk to what was behind that.
Cameron Schnier
Yeah. I mean it declined numerically, but I mean it actually improved. I mean, our fleet efficiency improved from, I think, 28.3% to 27.9%. So I mean that just means it’s incrementally more efficiently, and that was purely by replacing certain less efficient units with newer models that have a greater energy efficiency.
Joe Gomes
Okay. And then, one last one for me. On the delay with the client, I understand it could be resolved in a day, it could be resolved in beginning of 2025, but are there any costs associated with that delay for you guys where you won’t be seeing the revenue be coming in, but you will have the incurring expenses?
Sam Tabar
So on that front, so we ordered certain long lead time network equipment, but we didn’t place the order for the H100 yet as the lead time for that is now three to four weeks. So that’s good.
Joe Gomes
Okay. Great. Nice quarter. Thanks for taking my questions.
Sam Tabar
Thank you.
Operator
And we’ll take our next question.
Chris Sakai
Hi, Sam. Good morning. Just a question on HPC…
Sam Tabar
May I ask who is speaking? Sorry. May I ask who is speaking?
Chris Sakai
This is Chris Sakai from Singular Research.
Sam Tabar
Okay. Hi, Chris. How are you?
Chris Sakai
Sorry, they didn’t announce me. But on the HPC business, is this more of a strategic shift now to HPC than from bitcoin mining?
Sam Tabar
Well, here’s the thing. For a lot of pure-play bitcoin miners, there is this old saying, if all you have is a hammer, everything you see is a nail. So, all they can do is when you only have one lever is regardless of the economics, you got to grow. We are in a really incredible position to have multiple levers. We have three levers: bitcoin mining, Ethereum staking and HPC. And so, we’re able to work creatively since we have different levers in our capital allocation plans. And we have been always ahead of the trends. And with respect to HPC, we were one of the first, if not the first, to announce material contracts in this sector, and we continue to grow exponentially with a pretty pregnant pipeline.
So, we were right in identifying and using and leveraging the skills of a bitcoin miner into this space. We have very deep know-how in procuring specialized equipment. We have great relationships with the right counterparties on that. And we have — we’ve been able to execute really, really well for our current clients, which makes it even easier to get other clients because we have incredible track record. And we have global reach. We have offices around the world, including Singapore, and we’ve been able to serve underserved markets around different regions, not just the US in terms of client sourcing.
So, we’ve been in a very fortunate position, and we’re leaning pretty hard on hiring ahead of revenue and sales staff, which we’ve done, which we haven’t announced yet, but that has happened, they have started to help just from an organizational design perspective, lead the sales charge, because so far, we’ve done this without a sales team. So, with the sales team, we expect to bring a lot of these clients through the finishing line.
So, going back to your question, whether this is a strategic move, it certainly is, and we believe we were right. And you could see that the sector is following suit. And frankly, every investor and shareholder and institutional meeting I have, we have, are asking about the HPC side, which makes a lot of sense. You can model out cash flows much better that way to the penny. Whereas with bitcoin, it’s a little bit difficult to predict where it’s going to be and yet you have to invest all this money into the equipment and you have no visibility on the price of bitcoin. So, we don’t want to run a business based on hope. We want to run a business based on alpha and strategy and this move speak that in spades.
Chris Sakai
Right. Understood. And then, can you talk about — do you see any potential bottlenecks to onboarding new HPC clients? And for 2025, what do you see — can you give color as to what’s your pipeline there? How many new customers do you think will come on for HPC?
Sam Tabar
Yeah. As mentioned, our pipeline has been bandwidth and manpower. We’ve received plenty of reverse inquiries. And we’ve already built this business with zero salespeople. So that has been the bottlenecks, just manpower, and frankly, cognitive overload. And so, we’ve been able to solve for that, and that’s happening. And as mentioned, we have geographical reach, including global presence and offices in various regions that provide us key relationships and the ability to compete for business in underserved parts of the world. So, we do see — we already have a pregnant pipeline and frankly, the future is pretty bright in the HPC side of the business for the rest of the year and 2025.
Chris Sakai
Okay. Great. Thanks for the answers.
Sam Tabar
Thank you for the questions.
Operator
And that does conclude the question-and-answer session. I’ll now turn the conference back over to you for any additional or closing remarks.
Sam Tabar
Well, thank you again for joining us today. We appreciate your continued interest and support, and we do look forward to speaking with you again in the next quarter, pretty exciting times for us. And this officially concludes our call. Have a great day, everybody.
Operator
Thank you. That does conclude today’s conference. We do thank you for your participation. Have an excellent day.