pennyether

I've only listened to a few casts from a year ago. At the time he was bitching about how nobody uses AI, that it provides little real world utility, and that it's being crammed down our throats and nobody wants it. Also that $CORZ wasn't capable of building out a DC because in his little mind they were some sketchy crypto related post-bankruptcy mystery. (They've completed almost all the MWs on time, and stock is soaring.) I don't care to relisten and find bytes, because it's cringe listening to him. Early on I thought maybe he was onto something, but at this point it's clear he's just a charlatan placing a bet on an AI crash-out so he can collect reputation points. Essentially a conspiracy theorist. I actually suspect he is aware of this and doesn't actually drink his koolaid.. it's a measured bet with little downside and there's high demand for "anti AI" FUD.

This is kind of my point. You are trusting that raw demand for MW's, and Starwood being the seller/operator, is better than Fred and co. He will still be pompous and boring, but at least it will be more performative and less executive. Despite however he frames things... what I think really happens is $MARA will donate their MWs to somebody that knows what to do with them, in exchange for equity in that venture.

More nuanced take: Former $MARA was bullshitting ivory-tower mgmt slinging narrative spaghetti at the wall and being late to every part of the game. Awful competency at investing and monetizing MWs. Market cap leadership and Fred's grandstanding drew in boomer "I want BTC exposure" dumb money. But they luckily chased one narrative -- buying power assets. Despite this pursuit never really improving their mining costs, those assets have appreciated in value. New $MARA will still be bullshitting and ivory tower, but at least the strategy and narratives will be implanted into Fred's head by intelligent sources and not himself, and the competency of building/leasing/monetizing MWs will come from Starwood. Now less BTC, and since market cap is lagging peers, a simple "$/MW valuation" discount will draw in AI fomo'ers.

$MARA - I have Jun and Aug calls. Annual shareholder meeting June 18th. I think Starwood lands a tenant by then. Sure would be nice to have something non-bullshit to present, for once. $MARA shifting from a PR / crap-mgmt grift towards a consultant-run shell; deserves a rerate.

A year ago: @edzitron whining that $CRWV can't grow because $CORZ is "unproven" and "formerly bankrupt" etc etc. ~250MWs delivered and latest guidance was 450MWs by this summer.. which is almost exactly on schedule. $CORZ up 3x since then, and @edzitron still fudding. BAU. https://t.co/XP05x8qjXJ

A year ago: @edzitron whining that $CRWV can't grow because $CORZ is "unproven" and "formerly bankrupt" etc etc. ~250MWs delivered and latest guidance was 450MWs by this summer.. which is almost exactly on schedule. $CORZ up 3x since then, and @edzitron still fudding. BAU. https://t.co/XP05x8qjXJ

$MARA - Revenue came in a little higher than expected (more non-MARAPOOL revenue), costs lower than expected. Adj EBITDA (ex-ΔBTC) at -$22m vs Q4's -$39m. Gotta admit mining improved... from terrible to just bad. I had guessed -$1.46b in net income (assumed the same costs as in Q4)... it came in at -$1.26b, for a diff of $200m. Some ($17m) was from the better-than-expected mining ops, most was from depreciation dropping from $285m (Q4) to $192m, a diff of $93m. And "other" moving from -$24m (Q4) to $74m (due to gain from debt extinguishment), a diff of $95m. Anyway, there you have it. Pretty bad but not as terrible as expected. They'll burn cash 'til they get a deal. All eyes on HPC, good luck.

$MARA - Revenue came in a little higher than expected (more non-MARAPOOL revenue), costs lower than expected. Adj EBITDA (ex-ΔBTC) at -$22m vs Q4's -$39m. Gotta admit mining improved... from terrible to just bad. I had guessed -$1.46b in net income (assumed the same costs as in Q4)... it came in at -$1.26b, for a diff of $200m. Some ($17m) was from the better-than-expected mining ops, most was from depreciation dropping from $285m (Q4) to $192m, a diff of $93m. And "other" moving from -$24m (Q4) to $74m (due to gain from debt extinguishment), a diff of $95m. Anyway, there you have it. Pretty bad but not as terrible as expected. They'll burn cash 'til they get a deal. All eyes on HPC, good luck.

Correction: $MARA produces more like 44,000,000,000 petahashes per day. So a "cost per petahash per day" of $30 would bankrupt them in under an hour. (So yeah they mean PH/s/day) Attached is the footnote that explains how they calculate "cost per petahash per day", wherein the weasel word "online" is used and apparently includes hashrate that is not producing BTC. In 2025Q1 you can divide $131.5m of costs by ($28.50/PH/s x 90 days) to get to 51.25 EH/s that they used. But 51.25 EH/s should've mined about 2,644 BTC and they only mined 2,226 BTC... which is what 43.15 PH/s would've mined. Meaning their hashcost is fluffed down by ~19%. If you want to claim they just got unlucky that quarter and mined 2,226 BTC instead of 2,644 BTC, the odds of that are nearly impossible. 2,226 / 3.17 = 702 blocks 2,644 / 3.17 = 834 blocks difference of 132 blocks I'll let ChatGPT explain the rest

$MARA - My thoughts pre-Q1 results, and what questions I'd ask them. Bluntly, this quarter will be a disaster. They'll have mined at very thin gross margins (unprofitably for hosted), and burnt a chunk of cash. Narrative will 100% be on AI/HPC prospects, though they really should provide some answers as to wtf they are going to do with the awful mining business. Just some bulleted thoughts on mining: * For context, Q1 averaged a hashprice of ~$34.30 * MARAPOOL seems to have mined at ~51 EH/s. This produced ~2,060 BTC, or ~$157m in revenue. Generally they have a little bit more BTC/revenue from non-MARAPOOL.. I'm guessing they'll print ~$165m in revenue. Below consensus, apparently. * As for gross costs, 51 EH/s is about the same as Q4, in which they spent ~$158m on cost of revenue. So hardly any gross margin... * ...On top of CoR, there are massive cash expenses. SG&A (-$55m in Q4) and R&D (-$4m in Q4) and "other" (-$24m in Q4, though it fluctuates Q to Q).. that's a lot of overhead cash burn to support mining BTC at thin gross margins. (Generally, hashprice trends downward...) * Self-owned mining: in Q4 cost of revenue alone was $31.57 as hashcost. Compare to Q1 average of $34.30 and that's not much margin. * Hosted mining: In Q4 this costed $36.20 in hashcost, which is less than hashprice was in Q1. So if they kept hosted hashrate on, they'd have burnt money in Q1. And it appears they had the same total hashrate as in Q4, so it appears likely they kept it all on. Why? * Bottom line: Revenue of $165m, minus roughly same costs as Q4, and you land at $50m of cash burn, plus whatever they spent on R&D ($4m?) and "other" ($10m?) * That's a lot of cash to be burning after having sold a ton of BTC, having huge debt coming due, halving in two years (ffs just quit mining at that point), etc. * AI/HPC is figuratively a life raft for them, and it could very well be an entire ship. Insane to me Fred was poo-poo'ing this for so long and insisted $MARA was a competitive miner. ----------------- So that's the bottom line stuff. Next part is just a curiosity about their self-defined "cost per petahash per day", as they call it in their filings. (aka, hashcost) Firstly, the term makes no sense. $MARA produces something like 73,000,000 Petahashes per day. If they were paying ~$30 for each one they'd be bankrupt in a week. I believe they mean "cost per petahash per second per day" which is the same units as hashcost. But, whatever... tomato tomato right? At any rate, I believe they are fluffing hashcost downward intentionally, by using inflated hashrate numbers in the denominator. If they aren't, then they really need to clearly describe what "daily average operational hashrate online during the period" means... because the hashrate they divide the costs by is: 1) Very significantly higher than what one would expect, given the amount of BTC they produce. It's nearly impossible to produce such few BTC if hashing at that rate. (You can claim pool luck all you want.. there's math to show how unlikely it is and the inflating is egregious.) 2) In Q4 periods (where it takes more effort to back out the "online hashrate" number) it is somehow a nice, round number. Coincidence, I guess? My guess is they are including hashrate that is not actually hashing... which is bullshit, given the whole point of "hashcost" is to determine the cost of producing hashes... not having hashrate "online" and not hashing. It's even more confusing in the context of "energized hashrate" being a term they use. Apparently there is "energized hashrate" that is not "online hashrate". Then there is "online hashrate" that is not hashing, and that's what they divide costs by. ------------------ Questions for the earnings call * Most important: Regarding the Starwood agreement... exactly how much is each of $MARA's assets worth when folding them into a JV? Eg, if the JV is valued at $5b, and $MARA is pitching in 500MW of assets, how much equity would that get them? * How will $MARA fund additional JV equity buy in, to get to 50% * How will the JV's earnings appear in future earnings reports? I expect it to be non-consolidated.. meaning we're left with a terrible cash-burning mining business, then some sort of line-item that is the JV's GAAP income. So we don't even get clarity into the rev/cost/margins of the JV... it's just a "trust us bro" situation for what is realistically the only upside of the company. * Referring to the hashrate used in your "cost per petahash per day" calculation: I notice it is not producing as much BTC as what a 3rd party mining pool would. Is there a reason for this discrepancy and/or what does "online hashrate" mean? Is there room for improvement here? * Regarding hosted mining: It appears that in Q1 the average hashprice was below what you typically pay to have your hosted miners online, yet it appears you hashed the same amount as in Q4. Do you have the ability to shut these down opportunistically, or is it "take or pay"? ------------- That's all for now. Overall conclusion is that $MARA has a terrible mining business, but could very well be saved by AI/HPC... though even in that circumstance I wouldn't want to own any due to debt, and how management has historically spoken from an ivory tower made of poop, is late to literally everything, pays themselves too much, and perpetually distorts every thing they do. Disclosure: Have a bunch of $12P. Willing to get burnt if they announce HPC news, and would be happy for all the roped-in shareholders. But I suspect the headline miss to revenue, lack of a firm deal, and terrible "adjusted EBITDA" will rip them a new one.