tsp

Holding 36,350 $MARA shares at an average price of approximately $16.50.

[tsp @tpcgip]

@AdamLeeoff7 @darkerhumors No, now that $BTC has started moving, I actually hope the approval takes a bit longer so the value of mining has more time to rise — and with it, the value of the MW $MARA would be giving up.

[tsp @tpcgip]

$MARA A normal FERC approval process takes around 90 days. We’re now on day 108 since the Long Ridge application was filed on May 8.

[Robert Samuels @RobSamuelsIR]

Is Morgan Stanley getting incrementally more positive on @MARA? "We recently had the opportunity to interview MARA CEO and Executive Chairman Fred Thiel, and we were encouraged by Fred's bullishness on the potential to sign two HPC data center transactions before year-end . . . the feedback that the AI industry has a heightened sense of urgency to secure power, plus MARA management commentary, implies that we could be wrong in our conservatism."

[tsp @tpcgip]

@darkerhumors In any case, time is on $MARA’s side. The longer Long Ridge takes, the more valuable the MW they would be giving up become. $BTC is now taking care of that. 🤗

[Cryptoklepto @CK_Cryptoklepto]

$MARA should be at least $17.50 a share right now and would NOT be the most valuable $BTC / AI / HPC play in this grouping of stocks unless the price was over $331 a share

Also, none of these peers have the reflexivity of 35,577 bitcoin upside built into their balance sheets

👀

[tsp @tpcgip]

I don’t think $MARA expected Long Ridge to still be pending at this point.

The FERC filing was submitted exactly 90 days before earnings — roughly the normal review window. My guess: they hoped to have approval and possibly closing news ready for earnings week.

[Moe @Crypto_Moe84]

$BTC

Many many are sidelined and waiting for a new bottom or a deep retrace. I dont think the market will give them a deep retrace.. They can be lucky if they even get 72.8K

What i can tell you - 65.7K was THE LEVEL ... IN THIS PERIOD YOU WILL NEVER CLOSE back below it!

V-Recovery isnt FUN. Things can go very fast

Thank you for your attention to this matter

[tsp @tpcgip]

@jakeski75791164 Exactly. Right now the market is effectively assigning zero value to $MARA’s AI future. If a tenant deal dropped today, the stock might spike initially, but in this AI tape I wouldn’t be surprised to see most of that move fade by the close or the next session. The sector is dead.

[tsp @tpcgip]

$MARA We’re actually lucky that the tenant deal hasn’t been made yet. With BTC strengthening while AI valuations are getting hit, MARA doesn’t have to rush into a mediocre lease. It can mine BTC, monetize the power, and wait for a deal worth giving that capacity up for.

[tsp @tpcgip]

@jakeski75791164 Yep, you probably noticed how $RIOT’s stock reacted to the top-tier lease deal it announced a couple of weeks ago during this phase of the AI rotation.

[tsp @tpcgip]

$MARA Also note that as power becomes available at Long Ridge and Matagorda, MARA can monetize unused capacity through Bitcoin mining while the sites are being developed for AI/HPC — instead of leaving MW idle for years waiting for tenants to use it.

[tsp @tpcgip]

Are you getting any sleep at all over there on the other side of the Atlantic? $BTC $MARA

[tsp @tpcgip]

$MARA is starting to outperform $RIOT, $IREN, $CLSK, $CIFR and the other usual suspects.

[tsp @tpcgip]

The future of $MARA shorts?

[tsp @tpcgip]

The future of $MARA shorts?

[tsp @tpcgip]

$MARA is starting to outperform $RIOT, $IREN, $CLSK, $CIFR and the other usual suspects.

[tsp @tpcgip]

$MARA is especially interesting here: a huge mining fleet, massive BTC treasury, and an aggressive AI/data center strategy. If $BTC’s breakout starts an $80–100k expansion, miner/BTC correlation could snap back fast — and violently.

[matthew sigel, recovering CFA @matthew_sigel]

Bitcoin's correlation with the miners has collapsed to an all-time low (chart below). Intuitive given the AI pivot but arguably has gone too far. I think there's a case it reverses in a BTC bull market -->

One major reason the miners have fallen so much is that every lease they sign requires a bond issuance and some equity contribution. Given negative earnings with BTC at $60k and the rising 10-year, the market has been pricing in dilution. That could be very wrong in a Bitcoin bull.

Assume 80/20 debt to equity, roughly where these projects were originally guided: A critical megawatt costs about $11M. That puts the equity check on the visible unfunded pipeline at $774M for HUT's Beacon Point Phase 2 (352 MW), $385M for CLSK's Sandersville (175 MW), $475M for RIOT's Rockdale AI lab plus the remaining AMD capacity (216 MW). CIFR has all three contracted projects financed already, but carries 477 gross MW at Odessa, Reveille and Ulysses that sit outside the ERCOT batch process and are leasable now. At 20% I estimate the equity portion of that build is roughly $734M at 80% LTV (conservative).

Bitcoin averaged $71,680 in Q2. Here's what each earned in mining gross profit, revenue less power and direct cost, and what it annualizes to at $60k, $80k and $100k over the next year:

>CLSK made $52.5M, a $210M run rate. Cash cost is $44,591 a coin on 7,668 coins a year, so at $60k/80k/100k that's $118M, $272M, $425M annual cash profit. Covers 31%, 71% and 110% of Sandersville equity needs.
>RIOT made $34.5M, a $138M run rate. Cash cost $49,912 on 6,348 coins, so $64M, $191M, $318M. Covers 13%, 40% and 67% of Rockdale.
>CIFR made $9.8M, a $39M run rate with the fleet down from 23.6 EH/s to 11.6 at Odessa. So $23M, $51M, $78M. Against $734M, only 3%, 7% and 11% of equity needs.
>HUT made nothing. It moved all mining into American Bitcoin last year. But see below.

Then the coin stack, which nobody is crediting. Borrow at 50% LTV against unpledged coins instead of issuing a share:
>CLSK holds 13,931 with 4,070 posted on derivative collateral, so 9,861 free. That covers 77% of the necessary equity check at $60k, 102% at $80k, 128% at $100k. And the capacity is already committed: $400M of undrawn bitcoin-backed lines at Coinbase Prime and Two Prime, drawable against those same coins.
>RIOT holds 11,380 with 5,821 pledged against a $200M Coinbase draw, so 5,559 free. A new 50% LTV facility on those would cover 35%, 47% and 58% at $60k/$80k/$100k.
>HUT holds 4,567 free coins plus 4,291 look-through at American Bitcoin ($ABTC +15% today), less $200M drawn on FalconX. 34% at $60k, 46% at $80k, 57% at $100k.
>CIFR holds only about 640 coins after selling into the build, so the stack adds 3% to 4%

Those are borrowing numbers. Sell the coins (or the ABTC shares) instead and you double them, so CLSK covers 205% of Sandersville at $80k, RIOT 94% of Rockdale, HUT 92% of Beacon Point Phase 2. Before a dollar of mining profit.

CLSK has already run this play. From the Q3 release: "Anticipated equity portion of Sandersville project has been fully funded." Out of $203M of cash and $815M of bitcoin, with all long-lead items pre-paid. Share count went from 284.3M to 256.8M over nine months. They were buying back stock while the market priced dilution.

And 20% equity contribution could be quite conservative. HUT River Bend's actual out-of-pocket was $184M on 245 MW, about 7% of cost, and the bond proceeds then reimbursed their initial costs at closing.

Anyway, given the reflexivity here, I think these correlations may start to rise again soon.

(Happy to be corrected on math if someone gets to different figures. We are assuming network hashrate holds flat - it keeps falling. And we assume they all keep mining through the next year: RIOT's first AI lab delivery is December 2027, CLSK doesn't give up Sandersville until Q4 2027, and Odessa holds mining flexibility through July 2027. RIOT's own framing is that mining is a flexible load monetizing capacity until the leases finalize, and it's selling monthly production to fund the data center equity, with no common equity issued in Q2.)

[tsp @tpcgip]

$MARA’s network is becoming pretty unique:

→ GW-scale power/data centers
→ Velaura: AI silicon
→ Upscale: GPU/XPU networking
→ Aurascape: AI security
→ Exaion: enterprise/HPC/cloud
→ Starwood: data center development

+Fred Thiel and Lip-Bu Tan both sit on Velaura’s board.

[Matt Allen @investmattallen]

$MARA has a hidden AI infrastructure bet most investors aren’t paying attention to:

-MARA seed-funded Auradine, now rebranded to Velaura AI, and owns ~15%

-Velaura has spun out Aurascape (AI security) and Upscale AI (AI networking)

- $MARA CEO Fred Thiel sits on the board alongside $INTC CEO Lip-Bu Tan

-Velaura AI is focused on building next-gen AI infrastructure, including energy-efficient compute systems and data center technology

-Upscale AI is focused on AI networking, the layer that connects GPUs and moves data across AI systems

-AI isn’t just about compute, it’s about how fast data can move between chips, and when it can’t, GPUs sit idle and performance drops, making networking one of the biggest bottlenecks in scaling AI

-Aurascape is focused on securing AI systems, data, and agents as enterprises deploy AI at scale

-Through one investment, $MARA has exposure to compute, networking, and security, and while most investors still see it as a Bitcoin miner, moves like this show it’s positioning into the core infrastructure behind AI

[tsp @tpcgip]

@GigaMara_ @FredThiel @rob_samuels

[tsp @tpcgip]

@CK_Cryptoklepto @tbone4121 Note! In February, the news release and the quarterly earnings report were published just five minutes apart in the after-hours session. $MARA