HIVE Digital Technologies Ltd.

Hello, everyone, and welcome to today's webcast reviewing Hive Digital Technologies financial results for the first quarter ended June 30th, 2023. On slide number two, I would like to briefly note disclosures. Except for statements of historical fact, this presentation contains forward-looking information within the meaning of the applicable Canadian and U.S. securities regulations. These forward-looking statements are based on expectations, estimates, and assumptions as of the date of this presentation. On the next slide, I'm pleased to introduce today's presenters, Frank Holmes, Executive Chairman, Aidan Killick, President and CEO, and Darcy Dubaris, Chief Financial Officer. I would now like to hand the presentation over to Mr. Frank Holmes for a macro recap of the quarter. Frank?

Thank you, Holly. And thank you all of our loyal shareholders. I hope you find this informative, timely, and a little bit entertaining like this DNA of volatility, which I really like to always use as a case study opportunity. that every asset class has its own DNA of volatility. And what I thought was really surprised is looking at Hive, over 10 days is 26%. In over a month, rolling 20 days, it pushes 50% volatility. And I find that just incredible. But we've become a proxy for so many gold investors at the beginning. And then along comes crypto investors, especially those who are interested in the adoption, but they're reluctant after blowups like Celsius and FTX. Hive has become that proxy. So Please, when you look to invest in Hive, recognize that the DNA of volatility is six times that of gold, of the S&P 500, and substantially over any 10-day period. So the idea of buy the dip is wisely thought through here. now what's also interesting to us is that it appears a quant fund maybe a citadel or someone else that the the crypto stocks trade is a basket by the minute uh i do not believe that retail people jump in and out collectively like this but uh the correlations as you can see are very very high um hive is at 0.86 uh to riot is 0.96 to bit farms To HUT 8, it's 0.94. And I just think to Bitcoin, as you can see, Hive has this very strong correlation. The next one is to show you that we do offer like gold stocks, high-quality gold stocks due to the price of gold, that they give you two to three times the leverage on the upside, but they also give it to you on the chin on the downside on corrections. This is showing over year-to-date, and it's quite significant, gold surged to mid-July up to 81%. Gold this year was only up 7%. The S&P was up 17%, but Hive was up 315%. So it has done a positive 12-month rate of return, especially because of the rebound this year. Live Digital Technologies debuts a new name and strategic expansion to power the future of the artificial intelligence with its NVIDIA GPU cards. When we bought our NVIDIA cards, we bought them with a card that was not just mining Ethereum, but had the capacity to grow as we've Over three years ago, we started this journey of building out a B2B AI functionality using our chips. And so it's really come to fruition and quickly picked up the speed with the chat GBT. But I think that's what's interesting is that several of our peers bought NVIDIA chips, but they were single-purpose. Ours are not. Now, when it comes to this new space, it is very complex. And I've noticed several of our peers jumping in, making announcements, et cetera. But it is – Hive is the only one that has experience of mining with GPU chips. and branching out into offering AI services. And when you mine for Bitcoin, it's really simple. The quality of the chip, is it new? Is it more efficient on the energy consumption per hour? And you plug it in and you run it and you need a few people in managing a facility. When it comes to a data center where you're using for AI, it's much more complex. And I want to give you a simple metaphor I've learned on this journey. It's like going to McDonald's with three to four kids in the car, and you're trying to order, do you want a Happy Meal, do you want a Super Fries, what do you want, et cetera. And so when you go into this space, of artificial intelligence, those that want to use our services, they will look at, well, what type of short-term memory RAM do you have? We'll pay an extra 10 cents a minute. Long-term memory, we'll pay 8 cents a minute. Oh, this chip of 840, okay, we'll give you 70 cents a minute. 6,000, okay, we're going to give you $1.20 a minute. And so each client and each user is looking at a host of of services that you have to provide in this space. And to me, it was just such an interesting learning curve that it reminded me of this B2B business strategy is not easy and it is complex. And so investors have to recognize that very few people, except for pure data center providers, have this skill set. Hive does, and Hive has over six years of experience in GPU mining. So let me give you a macro recap of things that I think are important in the crypto ecosystem. Bitcoin is a decentralized portable wealth. And I always get concerned when I've seen this in the gold space that using J.P. Morgan's quote, which is here, J.P. Morgan said, gold is money and everything else is credit. Well, this is before the U.S. government basically changed the rules in 1933 and only allowed America's own numismatics. This is before the GLD ETF. Today, I am sure Jimmy Diamond would never say this because governments and the Bank of International Settlements get really upset that gold is not money, that gold is a wealth, an alternative asset class and wealth. And I think that that's as important. So is Bitcoin. Bitcoin is portable digital wealth. And that's what's really important to see how the adoption and Bitcoin has validated the ecosystem that's ushered in so many countries now trying to come up with their own central bank money. And I think that that's what's important for investors to recognize that transition. But let's talk about this. Let's talk about something that I've learned in my journey as a top-ranked gold fund manager and recognize that it's all about government policy. It doesn't matter if it's a Democrat or Republican at the helm. It doesn't matter who the prime minister of the country is. The responsibilities when you're that leader is basically fiscal policy. And what's interesting is the internet bifurcation to zeros and ones. And when you look at government policy, it also keeps this binomial model that is monetary or fiscal. And monetary are real interest rates and money supply. They have a big impact on the overall currency. And so do fiscal policies. Where is taxation going and regulations? And then where's the spending going? And historically, whenever there's a big imbalance between any country's monetary and fiscal policy, or you can get what's called monetary drag or fiscal policy drag due to regulations they're implementing, then gold becomes, on a historical basis, an important asset class to balance that. And Bitcoin has now taken over, this is important with millennials in particular, as their choice as the balancing factor between the imbalance between monetary and fiscal policy. When you understand that, you get a better appreciation. It's not just political parties, it's policies. so what i do believe in a macro trend that's happened this material this week is paypal's announcement and and this just really impacts the global adoption of bitcoin uh and the crypto ecosystem when paypal launches a stable coin it's to over 400 million global accounts uh That's not as big as when Facebook came out with its Libra coin, and there was such a big backlash by the Bank of International Settlements and the G20 central bankers and finance ministers that did a real pile-on to Facebook. It ended up being so damaged by it, they lost hundreds of millions of dollars, and they ended up changing their name. And I think it's sort of an interesting exercise to It's different today. I wish that the government had created policies and some basic framework for the crypto ecosystem, because that will still become a concern. But PayPal is already in digital payments. They already have a global footprint and a strong reputation. And so now after the Ripple lawsuit, the win by Ripple against the SEC gives them a new confidence to come out with this. And I think it just helps with the global adoption. I do believe... that the crackdown by the SEC was a huge speed bump to the adoption. And in particular, because it became so political with Sam Bankman-Fried and his FTX implosion and Celsius CEO getting arrested and put away in jail. I think these things are all part of a slowdown to the adoption, and now PayPal feels more comfortable, and this is important for the overall adoption. Coupled with, and very significant, they must have got a phone call with BlackRock, coming out with feeling confident enough that only six months ago had nothing but basically trash talk towards Bitcoin and the ecosystem and now is in full support of it and is trying to fast track their own spot Bitcoin ETF. So I think these are all positive parts in this global adoption process. But let's talk about Hive and how we've been the first. And we look upon ourselves as being a technology company. And we're the first to go public in Canada in September 2017. We were first to buy data centers. We were first to go and balance electrical grid. We were first to develop our own ASIC mining rig with Intel chips. And then we were first to have a green energy focus and first to have an AI strategy in how we were buying our GPU chips. And our last big purchase were not – which we originally had were AMD chips – This has been now pivoting over 18 months ago, buying NVIDIA chips to fast-track our ability to build out our data centers for AI. The leadership team, it's small. It's a small team. The CEO and president is Aydin Kilik, who's an electrical engineer. We have Darcy DeBarris, who's a CFO, has probably the longest standing experience as a CFO of a public company in the crypto data center business. Johanna Thorblad, who is the president of Sweden, and Gabriel Ibge, who's our general counsel. Hive is green energy focused in Canada, Iceland, and Sweden. five is a community commitment which we've done in at sponsoring the hive arena to because hockey is such a key element and the funding of 12 kids hockey teams uh is important so kids are not on the street kids are out being active and and this is a community that has three players that have gone on to go into the NHL and win Stanley Cup. So it's a very proud small community, and we're so happy and proud to be able to be involved in that community. Also, our vision in that community is to provide food security to the local communities, take away a global footprint from having to ship produce from Spain and Italy. And the vision is building up a 95,000 square foot greenhouse and taking the heat away. from our facility and recycling that to have a huge production of vegetables. We are successfully recycling heat in La Chute, Montreal, where we heat a building five times – sorry, four times the – see, five times the size, because one building is about 40,000 square feet where we're mining Bitcoin, and we've been able to heat during the winter a building – 200,000 square feet. So the capital structure of Hive, this visual of AI relates to the name change and the rebranding, which we focused in particular Canada as a rebranding. The ticker stays the same except for on the German exchange, it changed to VOO.F. We have a very tight float. We have been very, very conscientious of doing everything to have the highest cash flow return on investment capital. Revenue for the last four quarters. Darcy is going to give you more granularity in the presentation. But as you can see, it does follow Bitcoin price is a big impact. Now, we have been challenged, the whole industry, by more people mining Bitcoin, which is very rare that you would see Bitcoin prices fall over the past 12 months and finally rebound, but that more people would enter the business. And so the thought process of companies like Bitmain, who is one of the biggest manufacturers of ASIC chips, are growing. actually co-hosting and mining uh to get their capital back faster and and they've driven up uh the competition and the other parts of countries in desperate of us dollars uh have with low-cost energy have been mining bitcoin But all I do know is that America has become the dominant player in this ecosystem and leaders in the innovation. So I'm very happy to be involved in that sort of growth profile. But through this whole journey since 2017, Hive has always had a very low SG&A per Bitcoin mined. As you can see, that means minimum dilution, how we issue stock options, how many employees we have. I believe we have the highest revenue per employee against this whole system. We have a very low fixed debt on our balance sheet. We do not pledge any of our equipment. And so you've seen some of these other countries blow up and implode because of these debt structures from Core Scientific, who is the biggest. It ends up imploding. Argo implodes. So we can see that this idea of having a very conservative balance sheet, because I mentioned earlier my experience in the gold industry, when you have a high volatile asset class, then you have to be very conscientious of your debt structure and what you collateralize against. So for your shareholders, we give you the biggest bang for your buck. Hash rate utilization, Hive continues by Anthony Power to be able to give you a very high rate hash rate utilization. And that's because of the team of how they function every day from daily meetings early in the morning from the West Coast to make sure they touch base with everyone in Sweden before they slow down. at 4 o'clock in the afternoon and 7 o'clock in the morning in Vancouver. Every day is a scrum call. Every day is a know-what machines, any challenges or difficulties, and take things offline so that we can help and huddle and find solutions to problems. That's led to this incredible hash rate utilization. Our Bitcoin production per exahash, we're right there. It always seems to be one, two, and three. Interesting enough, Bitcoin is very strong in Canada and have expanded into Latin America, and they have a great team. We can see the share price growth in the last 30 days. We do, Aydin's going to go into greater detail of showing you. The amount of dilution, though, that's taken place with the financing that's been done or the stock issuance in the SG&A is much higher for many of these other companies. So I think I'm proud to see that we've seen as Bitcoin went to 31,000, we with our AI strategy and the rebranding to high digital and our strategy there has seen good shareholder support. I have always believed that green and clean Bitcoins would over time become more valuable as a digital asset because the supply is capped at 21 million coins. Like any raw art, I have said when supply is capped and adoption expands over time, the value of its prints have gone up substantially. Now with the explosion in ordinals, we are experiencing new growth in special number Satoshis. And Aydin is going to go through in greater detail, but I think it's really important to explain to the shareholders that the utility of Bitcoin was long limited to storing and transferring value with minimal developer activity on the network. The developers were all focused on Ethereum when it was proof of work, and now it's gone on to proof of stake. So it's sort of interesting to see that these developers are going to the Bitcoin infrastructure. Now, the emergence of Bitcoin DeFi launched protocols like Wrapped Bitcoin, WBTC, Stacks, STX, and Rootstock, RBTC, brought crypto borrowing and lending to the Bitcoin network. opening up new cases for the cryptocurrency. Now the arrival this year of Ordinals, the protocol, has expanded the utility of the Bitcoin network even further. The Ordinal protocol enables developers and artists to create and store digital artifacts directly on the Bitcoin blockchain. The protocol has been a massive hit with over 100,000 ordinals already minted. The introduction of the technology has caused Bitcoin metrics to skyrocket with a number of non-zero Bitcoin addresses hitting a record 44 million following the launch of ordinals. Already, Bitcoin's block sizes have also increased, touching an all-time high of 2.5 megawatts on February 12th. And I think that so what the Bitcoin Ordinals and why they're causing such a stir earlier this year is the Ordinals protocol allows users to send and receive sats that carry optional extra data such as text, JPEGs, audio, and videos. Adding such data to a Satoshi is known as inscription, and the result has been considered a Bitcoin NFT or called an ordinal. Well, nearly 100,000 inscribed Satoshis have been minted since the launch of the Ordinal Protocol, including the collection of Ordinal Punks and Ordinal Penguins and Bitcoin Offshoots and popular NFT collections. It's important to note that Ordinals are only possible thanks to the witness section of the Bitcoin transactions that was introduced after the Taproot soft fork in 2021. This is where JPEG's audio files and videos are inscribed into the individual sats. Bitcoin ordinals can be considered NFTs, but they have some significant differences. Most NFTs are created and tracked through a smart contract with the assets they represented hosted elsewhere. In contrast, Ordinals are inscribed into the Satoshi, and therefore Bitcoin NFTs are stored on-chain. Traditional NFTs usually have metadata that allow creators to change the appearance and characteristics of an NFT, but since ordinals are stored on-chain, their data is immutable and cannot be changed. Also, unlike NFTs, ordinals do not provide traders with the option to receive royalties on subsequent sales. In short, ordinals are immutable and complete digital artifacts that cannot be tampered with and reside completely on the chain. So it's an important innovation. You're seeing money being created for two things to basically fast track the adoption into what in addition to what PayPal has been doing, it is the lightning network so they can get down the cost of transactions. and the idea of funding more culture, more artists, more bootstrap funding as now getting VC funding to attract developers into this ecosystem and really embrace art. Art is such a key component for global adoption in the space. So innovation, as we can see, the daily inscriptions I mentioned earlier exploded earlier this year. They've papered off, and I think we're going to see other waves coming. Bitcoin ordinals innovation. One Bitcoin is 100 million satoshis. And what I'm just going to talk about is that what we found is that something that's worth next to less than a penny, all of a sudden, if they're at an inflection point with a limited supply, that they're uncommon, they can become extremely valuable. And so I think it's important that you listen to what Iden as he goes through, gives you more granularity on this macro trend. But now I want to turn it over to a snapshot of financial growth by Darcy, our CFO. Thank you, everyone.

Thank you, Frank. As usual around this time of the presentation, I will be taking you through a snapshot of the part year, looking at the most recently completed quarter and some indicators. Next slide, please. First of all, I'd like to remind our listeners that our earnings are comprised of our operational earnings plus our investment earnings, which includes realized and unrealized earnings, a lot of which include non-cash charges. Mark-to-market and accounting is an accounting practice that involves – actually, we can take it to the next slide. There we go. Mark-to-market is an accounting practice that involves adjusting the value of an asset to reflect its value as determined by current market conditions. Market value is determined based on what a company would get for the asset if it was sold at that point in time. Mark-to-market losses are paper losses generated through an accounting entry rather than the actual sale of the security. Swings in digital assets impact paper profits and losses each quarter. So our Bitcoin digital assets do generate unrealized gains and losses each quarter, depending on the price of Bitcoin compared to the prior reported period. And it's important that investors understand the differences in operating earnings or losses in addition to mark-to-market paper gains and losses each quarter, as it can swing the results drastically. Non-cash charge. is a write-down or accounting expense that does not involve a cash payment. Depreciation, amortization, depletion, stock-based compensation, and asset impairments are common non-cash charges that reduce earnings but not cash flows. Next slide, please. During this most recently completed quarter of June 30th, 2023, we recorded $23.6 million of revenue and 5.3 million in adjusted EBITDA. This was driven by production of 834 Bitcoin equivalent mined at an average cost of 18,687 US dollars per Bitcoin. Turning to the next slide, our cash position stood at 4.5 million at June 30th, 2023, along with an additional 59.5 million in Bitcoin digital currencies. We also had $9 million in amounts receivable and prepaid. Amounts receivable mainly consisted of sales tax receivables this quarter. The market value of our strategic investments decreased during the quarter by approximately $630,000 as a result of the mark-to-market accounting spoken about, bringing it to $2.2 million. We did not acquire or dispose of any investments during this period. We maintain a strong net cash position and healthy working capital to fund our operations and growth. Going on to the next slide. Our Bitcoin holdings at June 30th, 2023 was 1,957 Bitcoin down from the June 2022 month end period of 3,231. We have strategically sold some of our Bitcoin over this quarter to invest in higher efficiency ASIC machines to get prepared for the halvening in the first half of 2024. And also using these funds to invest in high performance computing equipment. Next slide, please. Our gross operating margin, which equates to our total revenues minus direct operating and maintenance costs, decreased in absolute dollars to $8 million or 34% in the most recent quarter compared to $27 million or 61% in the prior year comparative quarter. Gross mining margin is also partially dependent on various external network factors, including the high mining difficulty we continue to experience. the amount of digital currency rewards miners receive, and the market price of the digital currencies at the time of mining, which are on average lower than the prior comparative period. In addition, the company is no longer mining Ethereum since the merge took place on September 14, 2022, which has contributed to the decrease in gross revenue from digital currency mining by approximately 47%, but operating costs decreased by only 9%. In this most recent quarter, we are reporting a loss of $0.19 per share compared to a net loss of $1.41 per share reported in Q1 last year. Taking a look at our year-over-year revenue on the next slide, we generated total revenue in the first quarter of fiscal 2024 of $23.6 million versus $44.2 million in the prior year fourth quarter. The decrease in revenues versus the same quarter in fiscal 2023 can be attributable to three main headlines, and these might sound familiar. The ever-increasing Bitcoin difficulty hash rates over the past year, the significant drop in the price of Bitcoin, and to an extent, the Ethereum merge in September 15, 2022, as this quarter and operations moving forward does not include any Ethereum revenues. This triple punch contributed strongly to the significant drop in revenues that we experienced. As mentioned previously, our gross mining margin, which equates to our revenues minus direct operating and maintenance costs, decreased in absolute dollars to $8 million in the most recent quarter compared to $27 million in the prior year comparative. Moving on to the next slide, comparing our current fiscal Q1 quarter to the previous Q4 quarter, We generated revenue in this first quarter of fiscal 2024 of $23.6 million versus $18.2 million in the previous quarter. The increase in revenues versus the prior quarter was impacted by a stronger price of Bitcoin and 42 more Bitcoin mined this quarter. As the Ethereum merge took place on September 14th, 2022, which was in Q2 2023 of our reporting cycle, neither of these quarters had any Ethereum mining revenues. Our gross mining margin increased in absolute dollars to $8 million in the most recent quarter compared to $4 million in the prior quarter comparative. Increase in gross mining margin versus the prior quarter was impacted for the same reasons as stated above for revenues. Operating and maintenance costs only marginally increased by 9%. Moving on to the next slide. Our adjusted EBITDA increased in this first quarter of fiscal 2024 to positive 5.3 million versus a negative adjusted EBITDA of 1.4 million in the prior quarter. I will highlight again that adjusted EBITDA is a non-IFRS figure. In the first quarter of fiscal 2024, we experienced a loss of $16.3 million compared to a profit of $6.6 million in the prior quarter. The main reason for the profit in the prior quarter and loss in the current quarter, other than the reasons mentioned earlier, is the increase in Bitcoin price in Q4 2023, leading to a significant drop revaluation of digital currencies gain, whereas the BTC price did not fluctuate too much in the current period. I'd like to thank our loyal stakeholders, and at this time, I would like to turn the presentation over to our CEO and President, Aydin Kilik. Aydin?

Executive update on our business. Let's jump into it. So it was a great quarter in terms of Bitcoin mining profit margins. We saw an uptick. This quarter, we had 34%. You could see that difficulty actually increased. If you average a difficulty each quarter, it increased 24% quarter over quarter. But we managed to keep our cost to produce a Bitcoin quite consistent, only a 4% increase on our cost to produce a Bitcoin, which, of course, we saw a lot of strength in the price of Bitcoin as it rallied to $28,000. And in addition to that, although difficulty increased 24%, we produced actually 5% more coins. So we outpaced the growth of the network difficulty by expanding as we march towards our year-end target of 6x a hash, and we're currently sitting at 3.7, which I'll talk about more shortly. Next slide. So July 2023, we just issued a press release for this, and we produced 263 Bitcoin, an average hash rate during the month of 3.46x a hash, but we've been steadily increasing plugging in machines. We ended the month at 3.64x a hash. and did an average of 8.5 Bitcoin a day with a very, very strong 75.9 Bitcoin per exahash. Again, I've consistently ranked amongst the top crypto miners when it comes to Bitcoin per exahash. Again, a credit to our technical team, Resilience, High Fidelity Software. Running 24-7, you know, in seven time zones, we always make sure that our facilities are run top-notch, and I'm very proud of my team for continuing to have great performance. Let's jump to the next slide. So ASIC acquisitions. Now, to date, we have actually purchased over 11,000 ASICs just in 2023 alone. This does not include the purchases made in November and December of last year. which, by the way, have now ROI'd almost 80%. That's right. The S19J Pros that we purchased for about $11 a terahash on average in December have now ROI'd almost 80%. So that's a stellar ROI profile, given that it's only been eight months. And so we're targeting a one-year ROI on those machines. And so this is how we distinguish ourselves. This is how we analyze the Bitcoin mining hash rate economics. And we only search for the best deals. What is the best dollar per terahash price in consideration to the machine efficiency? We don't always buy the best... We certainly don't buy the best, most expensive machine. Some companies just rush out and do big, massive purchases when a new model comes out. No, we are all about analytics. We're all about best cash flow return on invested capital. And so throughout the course of 2023, February onwards, we made a series of purchases, some pro pluses, as you can see here, J-Pros. and even some XPs as we want to overall upgrade our global average fleet efficiency. And one exciting announcement that we just announced today along with our financials is that we purchased another 2,000 S19 XPs. which will be immediate delivery. So, you know, we also like to buy machines that are readily available to ship. So as soon as that cache is deployed, they can start auto-lying. So that's another exciting update. So let's hop to the next slide. So update on our hash rate growth outlook. We're still targeting 6x a hash for the end of the year. We have about 2,000 machines from our 11,000 machine order ready to be plugged in the next week or two. We're just adding some PDUs to New Brunswick. So we should be at 4x a hash in a week or two. And I've updated the target for September to be 4.2x a hash to include those additional 2,000 XPs that we've just purchased.

Next slide.

Overall infrastructure, 150 megawatts. So we've expanded a little bit our footprint in Iceland. And so we may not always utilize all of this power capacity because, for example, we might downclock some machines in order to have better unit economics, to have better efficiency. So they'll consume less power, but they mine more profitably. What we want to communicate to the market and let them know is that we have completed and available to us 150 megawatts of operating infrastructure.

And so this is what this slide summarizes for you. Next slide.

Our crown jewel, the flagship facility in New Brunswick, four buildings, two dedicated substations, which we own, the land and all the buildings and infrastructure upon it. A beautiful shot of one of the great data halls. You could see that's the hot aisle. You've got the dual path cathedral architecture. It's always a pleasure when I visit New Brunswick. And you could see it gets quite cool in the winter, covered in snow.

Let's hop to the next slide.

So just a little bit about our corporate strategy. You know, we've talked about this before, but it's worth revisiting. We actually expanded this analysis. And so it's going to show up on the next slide, but here I want to emphasize as a precursor to the next slide, we at Hive, right, we have a lean and mean mindset, right? High revenue per employee. We found out through third-party partnerships from Luxor's Hashrate Index and as well as Anthony Powers, who does great writing for Compass Mining, that we have amongst the lowest G&A in the sector as a percentage of revenue, the lowest share-based comp. And if you look at share dilution last year, we had the lowest share dilution while maintaining a hodl position of 2,000 Bitcoin. So, Let's hop to the next slide and I'm going to show you how that impacts your earnings. So we all talk about revenue and revenues measure the same across all crypto miners. Now, what I have to emphasize is there are non-cash charges that all crypto miners have to take on. And those can vary company to company because those are decided by the auditor. The auditor will determine what they're comfortable with. And therefore, we, for example, have a aggressive two-year depreciation of ASICs. So even if the ASICs are paid off quicker or slower than that, it doesn't matter. And by the way, we try to target a one-year ROI whenever we make our investments as evidence from... December purchase that's already at 80% paid off, which, again, you know, will become to free cash flow very soon after. And that's how we really want to grow the businesses was with free cash flow. Right. So we'd always have to raise money and most importantly, deliver value to shareholders. We're all about cash flow return on invested capital. So getting back to the non-cash charges, those are all recorded differently for all companies. Some have a four-year depreciation for ASICs, some three, we have two. So that's not shown here. But what everybody does record the same is their revenue, their cost of goods sold. I mean, that's your direct operating costs, your electricity, your data center costs, the staff, et cetera, insurance, whatever it takes for you to run your Bitcoin mine properly as a public company. And in addition to that, if you look towards the right of this chart, you have the SG&A. Well, that's the corporate operating costs, right? You know, executives, salaries for the executives, your lawyers, your auditors, D&O insurance, traveling, going to conferences, any sponsorships you do, and any of the above. And so what is the corporate entity spending to run? And so these are all sort of apples-to-apples comparisons that are on a cash basis. So the gross mining margin is reported. These are all from public filings. By the way, you can go to Yahoo Finance, Edgar, or CDAR. and you could see in the center the gross mining margin column right this is calendar year 2022 by the way and i will note hive has a march 31 year end so what we did was we added up our individual quarters um just because i mean you could do that manually and we do that to allow a consistent comparison across all companies and in fact Some of these other companies have a year end other than December 31st. So we took the liberty of adding up their quarter so we could do a January to December 2022 comparison. And this is what it looks like. And you see gross mining margin, which is the core operating business. is actually negative for some of the crypto miners we've got the color coding so it's easier to read that now hive along with a couple of our peers you know we're sort of uh north of 70 million i think you can see the top four here had uh gross mining margins 60 70 and about 80 million okay but what about the gna well the gna i mean you could see varies drastically it goes as high as 200 million and you know for some smaller companies as little as two million now hive is is a big company i mean we're one for one percent of the bitcoin mining network we've got our gpu business we're in five jurisdictions sweden iceland bermuda um even switzerland so This SG&A for us is only $14 million when a lot of our peers have $50, $60 million and even more to run their corporation. So what we do is if you take off the G&A from the gross mining margin, what's left is what we call a corporate margin. And this is where Hive has led the pack, $55 million corporate margin last year. And from there, this time we didn't look at six companies. We looked at almost 20 companies. And you can see here, There was only about four companies that had a positive corporate margin, and the rest were in the red. Something to think about when you're looking at investing in crypto mining stocks. I'm not here to really do anything other than just illustrate publicly available information through the financial filings. Very interesting, and I'm very proud of my team for keeping it lean and mean. Next slide. So, you know, we have seen hash price. It's sort of back at the 7 cent range. But what we noticed in May was this big rally to well over 10 cents caused by ordinals. And so what we did was in working with one of our strategic partners, Luxor, we did a scan of our wallets and we found that we had some uncommon Satoshis. We did a press release about this. This is very exciting because we can get a massive multiple. Let's look at the next slide. And so just to understand why there's such a demand and multiple payable that people are willing to pay for these uncommon Satoshis is because of their rarity. I mean, you look at the total number of Satoshis, 2.1 quadrillion. Well, only 7 million of those, 6.9 million, are the first Satoshi of each block. So we have over 200 of those Satoshis. And then if you have rare, that's the first Satoshi at the beginning of a difficulty epoch. Epic is at a halving event. There's only 32 of those, et cetera. So anyways, you know, Hive having been mining for years and years and years, we were fortunate to have accumulated some uncommon Satoshis. And now my CTO is working on doing transactions whereby we're going to be able to start mining our own Ordinals and Miners. uh transacting with individual sats we might sell some of these we do have a buyer that's interested and we've um have some an expression of interest to sell some of the premiums it's very exciting uh next slide please HPC. So last but certainly not least, you know, Hive is known as being a best-in-class Bitcoin miner. But of course, of course, we are known for having our world-renowned fleet of NVIDIA GPUs. And just to put it into context really quickly, Hive had 150,000 GPUs running during the ETH mining days. About 110,000 of those were actually AMD GPUs. gaming-grade GPUs, which were great for mining Ethereum. And those had been mining since 2018, and those had ROI'd four times over by the time the merge happened. So they had repaid themselves four times over. Great investment. Those profits went to expand the Bitcoin business. It was a very high profit margin business, about 80%, 90%. And after the merge, we sold those AMDs, but we kept them. the 38 000 nvidia data center grade gpus and those are the cards that have the capacity for doing hpc and ai computing and so now we're embarking on this very exciting journey which i'm going to dive into on our ai infrastructure now this data center that you see doesn't look like a crypto mine it looks like a data center it looks like you know a tier 3 data center and that's ours this is not a stock photo that's marin and johanna in our facility in Boden, Sweden. And so we have six years of history operating GPUs. So we have some of the best experts on the planet to run large fleets of GPUs. I mean, a lot of people mine Ethereum doing retail grade, you know, very small quantities. I've had one of the largest Ethereum mines in the world. And so all of that expertise is now being funneled into our AI infrastructure business. It's super exciting. Next slide. So this is a section on our AI outlook. So let's jump into the next slide. You will notice that we rebranded Hive Digital Technologies, of course. And so, you know, AI and cryptocurrency are pillars of Web3. And you've heard about the metaverse and DAO. These are all technologies that have been developed and developed. You know, communities and researchers and developers have been making contributions. But Web3 will be an adaptive web, whereas Web2 was a semantic web. Web2 relied on a lot of metadata, a few big Internet companies owning your information. Web3 will be more trustless and permissionless, which is allowed through blockchain technology. Crypto is going to be the Internet native currency of that. And AI will play a big part of that as well, so it will be a more adaptive web. Next slide. So where does high fit in? Well, first of all, it's pretty well renowned now that there's a large amount of processing power needed for AI applications, specifically large language models. I'm going to get to that in a second. Additionally, there's security concerns about data if you're uploading to a public device. GPT like OpenAI's chat GPT. And so companies are not able to upload client data or their own sensitive data because OpenAI will own it. It's kind of like uploading a photo to Facebook in 2012 where Guess what? Now Mark Zuckerberg has all your photos at your birthday party or your ski vacation. And so now companies need to have enterprise-grade AI compute, which we can provide through Hive Cloud, where we can run a large language model and even fine-tune it for clients. And so they have GPUs that they know exactly which GPUs and where they are. They can own the inputs, the processing, and the outputs of their own AI model. And so this will be our enterprise-grade offering to HiveCloud. Again, we've got a six-year track record as a best-in-class crypto miner operating a large fleet of GPUs. And by the way, our beta test GPU service is currently on track to do approximately a million dollars a year run rate revenue. As you can see, this quarter we did about $220,000 in revenue as that beta test has continued. Let's look at the next slide. So what is a GPT anyways? Well, a GPT is a generative pre-trained transformer. It's a type of large language model. It's a very prominent framework in generative AI. And so you can see some of the architecture details on the right. But really what you want to know is that GPTs actually emerged way back in 2018. Next slide. The thing is, you hadn't really heard of them until recently. Why? Because the intelligence, the apparent intelligence of the GPTs have not evolved until it was more remarkable with GPT-3.5 and moreover, more notably, GPT-4. you will see on this chart that illustrates the exponential growth. GPD-1 was in 2018 at 117 million parameters. It wouldn't have seemed that smart if you try to play with it. GPD-2, about 10x that. A year later, GPT-3, 175 billion parameters. A year later, 150x the previous. And GPT-3.5 came out last year, and then that was a big improvement. There was some architectural upgrades. And, of course, GPT-4 is 1.8 trillion parameters. And it's rumored that GPT-5 is going to have 17 trillion parameters. So as there's this exponential growth in demand, that is where you need all the processing power for these GPUs. which, by the way, are really good at multiplication and addition. And so that is very necessary for training the large language models and running inferences. Next slide. As I mentioned, GPT-5 may use 17 trillion data points. And it's not just about GPT-5. I mean, that's OpenAI's product. What we're finding and where we believe the immediate opportunity is actually open source GBTs, such as LAMA2, 70 billion parameter, where tests are now showing, you know, with companies such as Meta, research institutes at Stanford, et cetera, that having smaller, large language models, but with data sets that are bespoke and specific to your use of are producing results that are as good as GPT-4. And so my, the point here to understand is you don't need a trillion parameters that covers everything from, you know, um, you know, soup recipes to Cinderella. If you're, if you're, um, an engineering firm and you want to look at fluid dynamics data or aerodynamics, what have you, then you are obviously going to have a smaller subset. And so with a $70 billion parameter model, you can fine tune it. We can vectorize your existing data, feed it into the LLM. fine-tune it to your application and then you can have a much more focused and robust model and we can do that on our stack of gpus which we're going to get into shortly here next slide So right now we have spec'd out a great super micro server with 96 threads, 7.6 terabytes of memory, 512 gigabytes of RAM. If we use 10 GPUs, they will have 480 gigabytes of VRAM. And so these are powerful. And we can cluster them. And before we even got to this next generation, we did a beta test, and that did 230,000 last quarter and about 220,000 this quarter. But keep in mind, that is only 1% of our 38,000 fleet of NVIDIA data center-grade GPUs.

Next slide.

With our new incoming super micros, which we just press release, we just tripled our footprint. We had some arrive in Stockholm. We're operating these out of a tier three data center. It's very exciting. With just 12% of our NVIDIA chips, that's about 4,600 GPUs. Our target is this year in December. We will realize, based on demand that we're seeing in the marketplaces we work with, where we lease out our GPUs for HPC computing, about $15 million in revenue. And the capex for this would be about $6.4 million. Again, we have 38 million GPUs, and if you looked at that whole suite of NVIDIA GPUs on a blue sky basis, and they were all repurpose for AI compute based on the indicative pricing we've seen in these marketplaces, it could be as much as $100 million of revenue. So it's a very exciting time and we'll keep the market posted for updates.

Next slide. And that concludes our presentation.

So thank you very much. Please check out our new social handles as we've rebranded Hive Digital Technologies. Also, I'm on Twitter now. So go ahead and follow me. It's simply Iden Killick, Hive. So, yeah, you know, I'm entering the Twitterverse or the X-verse now, I suppose. Look forward to meeting you all there in cyberspace. And we'll be keeping the market posted with exciting developments on our AI journey. Thank you so much.