Evan Wilburn (@bobleewaggeris) header
Evan Wilburn (@bobleewaggeris) profile picture

Husband. Father. Builder. Energy × Digital Scarcity × AI. long $MARA $TSLA $MSTR $BTC Personal opinions. NFA. DYOR.

Joined May 2021
Photo by Evan Wilburn (@bobleewaggeris)Photo by Evan Wilburn (@bobleewaggeris)Photo by Evan Wilburn (@bobleewaggeris)Photo by Evan Wilburn (@bobleewaggeris)Photo by Evan Wilburn (@bobleewaggeris)Photo by Evan Wilburn (@bobleewaggeris)Photo by Evan Wilburn (@bobleewaggeris)Photo by Evan Wilburn (@bobleewaggeris)Photo by Evan Wilburn (@bobleewaggeris)Photo by Evan Wilburn (@bobleewaggeris)
Evan Wilburn retweeted
What does $MARA need to show by Dec 31 for you to keep holding?
50%Two signed leases
20%One signed lease
6%Any LOI or tenant news
24%Nothing, I'm holding
86 votes • Final results
1
5
708
What does $MARA need to show by Dec 31 for you to keep holding?
50%Two signed leases
20%One signed lease
6%Any LOI or tenant news
24%Nothing, I'm holding
86 votes • Final results
1
5
708
Fred has had every chance over the last few months to soften the year-end target. He’s not backing down. I like it. $MARA
4
2
31
2,604
Evan Wilburn retweeted
Reminder: $MARA has already covered FULL acquisition costs of Long Ridge from $FIP by obtaining approval to transfer asset debt and by raising $600 million via collateralized $BTC loans Once FERC grants final sale approval, this will be a non-dilutive catalyst for AI contracts
AI demand isn't waiting 7-10 years for infrastructure to catch up. MARA's agreement to acquire Long Ridge shortens that timeline, at below-replacement cost. Learn more: bit.ly/4wdcSWL
3
9
46
4,896
Flip the $MARA Starwood deal around and look at it from Starwood's chair for a minute. On Starwood's August call, Barry Sternlicht said everyone in data centers is chasing "the same half a dozen tenants." On the same call he said there had been "unbelievable tenant interest" in the MARA properties. Think through the implication. If you are chasing a handful of hyperscalers, you need more than a site. You need inventory across markets, and a way to carry it without bleeding capital and time. Hyperscaler requirements are specific. One wants Texas, one needs 300 MW in a particular market, another needs 100 MW now with a path to 500 later. A developer with one property gets one shot at the match. And one property can fail for reasons you do not control. Starwood spent years assembling land, power and approvals for a 1.2 GW campus in Delaware, and the state said no. Years of work, nothing deliverable to show a customer. MARA changes the starting point. The power is already there, the infrastructure is operating, and the site is producing revenue. Bitcoin mining is the holding income on Starwood's development inventory while the higher value use gets built. So when one of the six calls, Starwood has a menu "Tell me the geography, power profile, scale and delivery date you need. We have a portfolio." And the inventory is elastic. The agreement lets Starwood add land, power and infrastructure around whichever site a customer picks, so Kearney's 100 MW start has a permitted path to 368 and Granbury's mine has hundreds of acres assembling around it. MARA can add new sites to the shelf, which it did with Matagorda. And mining comes off the shelf in tranches as a tenant needs it, with the rest still earning. The site grows to the requirement instead of the requirement shrinking to the site. $MARA's job is to keep growing the inventory. Starwood's job is to turn the menu into leases.
10
3
1
43
2,934
Starwood needs to hurry up and get us a tenant or we’re going to have to find a new real estate agent $MARA
3
1
2
47
2,860
$MARA power pipeline as of Sep 29 with key updates. Roughly 820 MW owned and operating today. 4.8 GW pro forma if Long Ridge closes and Matagorda fully energizes. Corrections welcome.
1
2
30
1,325
$MARA: I think the overlooked benefit of Long Ridge is the expertise coming with it. MARA plans to retain the operating team. This is important because the AI power constraint is about getting usable power to the right location on a timetable the customer can work with. Grid connections are already limiting how fast data centers can expand. So if usable power is the bottleneck, then knowing how to turn a power opportunity into capacity a customer can actually use is the valuable skill. You have to know what a site needs and what it will cost, then decide whether the customer will pay enough to justify the work. This is why Long Ridge and Matagorda are interesting together. One is an operating energy business with integrated fuel and generation. The other is a much earlier development with a lot of work still ahead. I think understanding the finished system changes how you look at the unfinished ones. You see the expensive problems earlier and make better calls about which sites are worth the work. That improves the throughput and economics of the whole development portfolio. Think of it like a megawatt refinery. Owning available megawatts lets you benefit from scarcity, but learning how to make more megawatts useful gives you a process you can repeat. The plant has good economics already. What interests me is whether the expertise coming with it makes MARA better at resolving the constraint holding back the AI buildout.
Most investors think $MARA MARA has to wait for its first AI tenant before Long Ridge creates value. They are wrong: -The $1.5 BILLION deal includes a 505 MW natural gas plant and more than 1,600 acres of land for a future AI data center campus -The power plant already produces roughly $144 million of annualized adjusted EBITDA, with more than 70% of its electricity contracted under long-term agreements -Long Ridge operates at less than $15 per MWh and has its own integrated natural gas supply, giving MARA access to reliable power at an extremely low cost -The site already has 200 MW of MARA capacity, a path to 600 gross MW of AI workloads, and more than 1 GW of total potential capacity
1
2
34
2,213
The more I look at $MARA’s strategy, the more I think the real advantage is what it isn’t committing to. If the AI boom keeps ripping, MARA participates through ownership in the infrastructure layer that survives multiple GPU generations. If the GPU layer gets oversupplied, MARA hasn’t loaded its balance sheet with tens of billions of rapidly depreciating hardware or committed itself too heavily to a single compute strategy. And if the cycle turns, the valuable thing may suddenly be patient capital attached to power. Don’t be the guy who maxes out the balance sheet at peak pricing. Be the guy with power, sites and optionality when everyone else needs capital.
5
3
1
29
5,589
The other interesting scenario is where $MARA gets extremely good at refining megawatts, Bitcoin rips, AI continues to boom, while at the same time higher layers of the stack get more competitive and margins compress. Competition up there doesn't shrink the demand for power, it multiplies it. So you've got a situation where more operators are fighting over the same megawatts, each earning less, and the margin they're losing is moving down to whoever owns what they all need. Meanwhile MARA's balance sheet is still compounding, and they not only own the scarcest layer of the stack, but they're getting good at creating more of it, at the exact moment everyone above them needs more of it.
309
The other interesting scenario is where $MARA gets extremely good at refining megawatts, Bitcoin rips, AI continues to boom, while at the same time higher layers of the stack get more competitive and margins compress. Competition up there doesn't shrink the demand for power, it multiplies it. So you've got a situation where more operators are fighting over the same megawatts, each earning less, and the margin they're losing is moving down to whoever owns what they all need. Meanwhile MARA's balance sheet is still compounding, and they not only own the scarcest layer of the stack, but they're getting good at creating more of it, at the exact moment everyone above them needs more of it.
The more I look at $MARA’s strategy, the more I think the real advantage is what it isn’t committing to. If the AI boom keeps ripping, MARA participates through ownership in the infrastructure layer that survives multiple GPU generations. If the GPU layer gets oversupplied, MARA hasn’t loaded its balance sheet with tens of billions of rapidly depreciating hardware or committed itself too heavily to a single compute strategy. And if the cycle turns, the valuable thing may suddenly be patient capital attached to power. Don’t be the guy who maxes out the balance sheet at peak pricing. Be the guy with power, sites and optionality when everyone else needs capital.
1
6
2
39
3,999
Evan Wilburn retweeted
Great news for $MARA and also great to see that the bullshit ambulance chaser lawsuits regarding "data center noise" are all being dropped
Big news for Bitcoin miners and data centers: 31 of 37 plaintiffs dismissed their federal claims with prejudice in litigation challenging alleged noise and vibration from @MARA Bitcoin mining & data center facility. A reminder: headlines are not evidence. Virality is not causation. Litigation asks: Is it true? and can you prove it? This is much more nuanced than one post can capture. Community concerns are taken seriously and investigated, measured & mitigated. But once concerns become legal claims, particularly claims of substantial nuisance or medical injury claims seeking money, allegations have to withstand legal scrutiny. That means evidence. Experts. Source attribution. Science. Causation. After a year in litigation, 31 cases dropped and the lawyer withdrew from representing the remaining 6. That speaks volumes and you can draw your own conclusions about the merits of the claims. Taking a complaint seriously does not mean assuming every allegation is true.
1
2
29
2,939
The thing everyone missed in the SemiAnalysis $IREN section is the recommendation. They told IREN to stop selling managed clusters and just sell bare metal. I agree with them. IREN is trying to be a power developer, a data center builder, a GPU owner and a cloud company at the same time, before it has mastered any one of them. Microsoft is paying for the first three. It never bought the managed cloud. It contracted dedicated GPU capacity at Childress, in a building to its own spec, ran its own acceptance tests, and accepted Horizon 1 in August. Every layer you own is another thing you have to be great at. IREN would be a better company doing three things well rather than trying to do all four at once.
2
382
Evan Wilburn retweeted
I think I found one of the most mispriced AI infrastructure stocks in the market. It controls billions of dollars worth of power infrastructure. I broke down exactly why I’m buying it here:
18
36
11
126
59,514
The AI power shortage is being described wrong. The scarce resource is AI-ready power. $MARA is building an AI megawatt refinery. -> acquire sites -> de-risk them into AI-ready assets -> contribute them for equity -> recycle capital -> repeat. Raw MW in. AI-ready MW out.
weekend reading for investors underwriting the AI power trade: who controls the scarce megawatts, what are they worth, and where is the market mispricing them? @matthew_sigel @Patrick_Bush_VE vaneck.com/us/en/blogs/digit…
1
8
41
3,154
Evan Wilburn retweeted
Replying to @Ploutos97
I like the double catalyst possibility for $MARA. So MARA for me, $MSTR is a more pure BTC play ofcourse
1
3
12
587
One clarification on my Kearney math. 368 MW is gross facility power, not critical IT. So I simplified the model by applying the build cost and lease assumptions to the full 368 MW. critical IT will be lower based on PUE, so both build cost and lease revenue come down together. Doesn’t change the thesis though. The important numbers are still the site contribution value and how much new cash MARA actually has to put in at each ownership level.
1
77
Good coverage on Velaura @CK_Cryptoklepto. The more I look at it, the more interesting $MARA’s positioning gets. Power. Data centers. Sovereign compute. Bitcoin. Now even the silicon efficiency layer. $MARA keeps building exposure across the AI infrastructure stack, specifically where the bottlenecks are.
Interesting article on Velaura AI's revolutionary Titan Core technology Article: eejournal.com/article/new-se… Titan Core: velaura.ai/low-power-ai-comp… Note: MARA has an investment stake in Velaura AI and $MARA CEO Fred Thiel along with $INTC CEO Lip-Bu Tan are on the Board of Directors
1
9
36
2,085
Exaion aging well $MARA
Hate to be the guy, but I told you! Anthropic & OpenAI Hunt for smaller Campuses …. They realised they want to train each model independently so they can have access to the kill switch for each data center without hurting the public models. This is happening, it’s real. Sovereign AI is emerging
7
47
2,741