Welcome back, everyone. We have an update from BitFuFu Inc. It trades on the NASDAQ under the symbol FUFU. It's a world-leading Bitcoin miner and mining services innovator who's committed to empowering the global Bitcoin network through its industry-leading cloud mining platform, rapidly scaling infrastructure, innovative mining services. Happy to welcome the VP of Investor Relations, Charley Brady. Welcome back, Charley. We're looking forward to hearing your update.
Thanks so much. Happy to be back here and tell the BitFuFu story to folks. Let's jump to our first disclaimer slide here. I'm not going to read this, obviously, but this is our safe harbor statement. So please feel free to review at your own leisure. So I'm just going to briefly cover who BitFuFu is, what we are, what we do, what our strategy is for those who may not have seen a prior presentation. And then I will go into our most recent third quarter results, which were released last month. And then we'll leave some time at the end for any of your questions. So as you just heard, we're a world-leading Bitcoin miner and mining services innovator. On our cloud mining platform, we have over 648,000 users as of the end of November of 2025.
From a computing standpoint, as of end of November, we manage 26.4 exahash of computing power for Bitcoin mining. Total capacity there is 478 MW. And as of November 30th, we held over 1,700 Bitcoin. Just to kind of dig a little bit more about what we do and how we do it, really on a vertical and horizontal integration basis, we've got a self-mining business, which is a typical self-mining. We have rigs deployed in the field that are mining Bitcoin for our own account that go to our balance sheet directly. That's a smaller piece of our business. It's probably around 30% or so. The larger piece of our business is really the mining services side, and that's dominated by the cloud mining business.
When we say cloud mining, what we're doing here is we're leasing out that hash rate, that exahash that I just mentioned, that computing power. We're leasing that out to customers, both institutional and retail customers. It is heavily majority institutional revenue at the moment, but we do have retail customers that are in there as well. So the way it works is a customer, rather than wanting to go out and buy their own mining equipment, secure cheap electricity, which is the largest input cost of mining Bitcoin by a wide margin, and then finding a hosting center and a mining pool and doing all the logistics that goes involved with Bitcoin mining these days, we simply provide a one-stop shop to do that.
And the customer can go on and select how much mining computing power capacity, how much terahash they want to purchase or lease, I should say, and for what period of time. And we generally lease it in 30-day increments, anywhere from 30 days up to 360 days. Our average duration among our customer base is somewhere in the 60 to 120-day range. It kind of moves where we are in the cycle. So it's in that range roughly. Again, mostly institutional money. And as the Bitcoin gets mined, it doesn't go from the pool to BitFuFu and the customer. It goes directly to the customer. So the customer will then choose a mining pool. They will take their own personal wallet and they will attach that into the pool.
And then when that Bitcoin gets mined, that Bitcoin is going directly from the mining pool directly to the customer's wallet. We're not in between that. So it's full transparency. You can see what's being mined in real time, either on the website or on the mobile app. But we really want to be very transparent about what's being mined, how much you're going to be mining. And obviously, the upfront cost is very transparent. When you go to purchase that contract, you know exactly what you're paying per terahash per day and what the service fee per day is going to be on that as well. Our network really is on three continents here. So we are just a little over half our computing capacity. Our hosting capacity is in the United States at 239 MW.
Again, this is as of end of November of this year, just under half in Ethiopia, 235 MW there. And then we got a little bit of leased capacity, about four MW in Brazil. Now, of this 470 MW that we control, that we have computing on a daily basis, we own 164 MW of that. And the rest of that capacity is leased capacity. And so part of our strategy right now really is to go out and acquire more owned capacity. The company originally started as an asset-light company, but we've since broadened out to where we want to own our own data centers and kind of shift that. Because when we do that, we get more control of the operations, obviously. We do lower our overall costs when we run the sites ourselves with our own machines. Our cost to mine Bitcoin drops dramatically.
And it also gives us optionality down the road if we want to pivot into HPC AI, which we're not doing currently, as a lot of miners are, but it's something that we are very seriously evaluating and looking at. But really, to do that in a big way, we are going to have to probably acquire, certainly acquire data centers, additional data centers, owned data centers. This just kind of covers a little bit what I have, but just on the mining services side, in addition to the cloud mining piece, we do offer, again, full service. We have our own mining pool, a BitFuFu Pool. We have our own mining software, our OS system, which allows customers, as well as ourselves, to overclock and underclock machines and really optimize the operation of those ASIC mining rigs in the field.
And then we have a monitoring software as well that monitors every mining rig in the entire fleet in real time. So we're constantly optimizing these machines. They don't just get put on a rack in a field and turn a switch. We overclock them, we underclock them, we manage the software so that those machines are fully optimized depending on the electricity cost, depending on the temperature that they're running at. The outside temperature obviously makes a big difference. And we really want to optimize that. And that software helps us to do that and make sure that those machines are up and running at a very high rate. We guarantee 95% uptime to our customers. Realistically, we actually run probably 98%, 99% uptime across the fleet. Let me just jump real quick into our third quarter results. They were really, really strong.
We had a good solid quarter there. You can see our cloud mining revenue was up 78% year over year. It was up, also strong sequentially from the second quarter. Self-mining revenue down just 2%. Some of that is due to the halving impact from last year, as well as an allocation of mining computing power from self-mining into the cloud side because the demand for cloud mining has been very, very strong for the past couple of quarters, and so we've actually taken some capacity we would use for self-mining and allocated that into the cloud mining side to meet customer demand. We did have about $36 million of mining rig sales in a quarter. That was an unusually high quarter. We've generally been running about $5 or $6 million a quarter for the past several quarters.
So that $36 million was kind of just a bit of a one-time bump. I wouldn't expect that to be a normalized run rate going forward. From an EBITDA standpoint, our adjusted EBITDA was just over $22 million. It was up from $6 million in the prior year. And just a reconciliation on this. And one thing to note on the EBITDA number is that we, on the income statement, we have to mark to market our unrealized gains and losses on digital assets, which for us is Bitcoin, as do all the miners who are following GAAP accounting. And so you see the footnotes here. That's one thing to note when you're looking at EBITDA numbers from Bitcoin mining companies. It's really just for that non-cash recognition that mark to market every quarter because it does make a pretty big difference.
It wasn't just a one-quarter phenomenon. We've had pretty strong results and strong growth for a number of years. We've been profitable every year in operation, which is not something every miner can say. We expect to continue to be profitable. One of the reasons we can do that is because this cloud mining business allows us to manage the industry cycles, whether it's an upcycle or downcycle. We move that allocation between cloud mining and self-mining to make sure that we are staying at the highest profitability level regardless of where we are in the industry. I'll just stop there and I'll take any questions that might have come through.
All right, thanks, Charley. Yes. How do you balance growth between self-mining operations and cloud mining services, both in terms of CapEx and resource allocation?
Yeah, I mean, that's a great question. The majority of our mining rigs are actually still leased. So we do own probably maybe 20%-25% of our rigs, but the 70%, 75%, 80% are leased rigs. And we do that because it allows us to not have to be that capital intensive. So we don't have $200 million or $300 million of assets that are going to be obsolete in a year or two sitting on the balance sheet. When those machines come off lease, we generally lease the newest generation machines, which if you look at our efficiency ratio over the past 10, 12 quarters, you see a steady improvement in that efficiency rate because we're always moving up the chain to the more efficient, newer machines.
In terms of what determines cloud and self-mining, it's been heavily cloud-driven for the past couple of years as that business has really grown. But on a rule of thumb basis, when Bitcoin is very, very high, we generally lean more towards the cloud mining business. When Bitcoin is low, we lean more into the self-mining business because our cost structure has some elasticity to it and it makes more sense to do that. But quite frankly, the cloud business has been so strong from a demand standpoint, we can't fill demand. We're trying to acquire more capacity, quite frankly, but we have more demand for cloud mining than we actually have capacity today.
How is sustainable the recent growth trajectory?
Yeah, I mean, I think the demand for cloud mining, again, is not waning. We've seen that if you look at our results coming in November, we do release monthly statistics that cloud business remains strong. I'd say our biggest issue we have to deal with is really finding more capacity and more hash rate. We do lease a good portion of that hash rate, and so we're always looking for additional capacity. We do have a pipeline of potential M&A transactions that we're looking at as far as owning our own data centers. Most of those are in operational centers today, so we're not going to build those from scratch, although that is an option as well for some things we're looking at.
But we could get up. Once we acquire that data center that's already operating, we can get up and running and integrate that very, very quickly and upgrade some of those machines as well. So that's going to be the key to growth. Really, it's going to be growing that computing capacity and growing those MW.
Last question, what are current and targeted hash rate levels and power capacity for year-end of 2025?
Yeah, we've already, I think we said several months ago, we wanted to be year-end 2025 somewhere around 28-32 exahash exiting the year. A lot of that's going to depend on some lease contracts. We did have a lease contract roll off in November, so you did see a dip there, as you have seen in the past. And we also took a bunch of older miners, S19s, offline, and we're replacing those with the newer generation S21s, but there's a little bit of a lag time getting those machines offline and putting those new S21s into position to get those rigs running. So that's why you saw the dip in November. But that's generally been our guidance. In terms of MW, our guidance has been somewhere in the 550-600 MW.
That's going to, again, some of that came offline in November, but we expect that to come back online as we renew some of these lease contracts and secure some third-party hash rate contracts.
Perfect. All right, Charley, well, thank you so much for this update. We look forward to following along with more of this progress in 2026.
Great. Thanks very much.
All right, everyone, we'll be right back.