Bit Whitman (@BitWhitman) header
Bit Whitman (@BitWhitman) profile picture

Investor @ Macro | AI Infra | Bitcoin | "New Joule Order"

Tampa, FL
Joined February 2013
Photo by Bit Whitman (@BitWhitman)Photo by Bit Whitman (@BitWhitman)Photo by Bit Whitman (@BitWhitman)Photo by Bit Whitman (@BitWhitman)Photo by Bit Whitman (@BitWhitman)Photo by Bit Whitman (@BitWhitman)Photo by Bit Whitman (@BitWhitman)Photo by Bit Whitman (@BitWhitman)
5
886
Bit Whitman retweeted
Treasury’s $95 Billion Buyback Plan Leaves Door Open to More Treasury plans to buy back up to $95 billion of debt over the next three months, versus the $63 billion ceiling initially announced in August. The wording also leaves room for more: Treasury says it is evaluating “possible enhancements,” including “changes to maximum purchase amounts” and operation frequency. That signals flexibility to expand the program—not a commitment to exceed $95 billion home.treasury.gov/news/press…
4
41
2
171
13,756
This is actually what’s happening in the market at large. The emperor has no power.
The biggest overlooked insight from $IREN earnings call is that the company is preserving the option to buy cheaper GPUs Near the end of the call, Jefferies asked whether GPU purchase prices might rise enough to eliminate today's attractive rental economics. Daniel Roberts gave a rather unexpected answer. IREN has deliberately avoided purchasing all the GPUs it expects to need next year. Why? Because if financing remains tight and datacenter construction keeps slipping, GPU prices could actually fall. Think about that for a moment. The conventional AI bull thesis says GPUs are scarce, so buy as many as possible. IREN's argument is more sophisticated: the economic value of a GPU depends on having somewhere to power and operate it If the industry has more chips available than deployable power and datacenter infrastructure, the scarcity premium can migrate from the GPUs toward the power connections. Here is management explicitly positioning for the possibility that financial scarcity makes one part of the AI supply chain cheaper while another remains scarce. IREN is effectively trying to own the long duration infrastructure option without committing prematurely to every generation of depreciating compute. This is not the first tine the management takes a non consensual direction. In the recent past, they avoided signing too many deals, and in retrospect it now seems a smart move (look at Nscale).
3
507
Bit Whitman retweeted
There’s a revolution happening in the Trump administration right now. I am shocked about who I am hearing from - senior leaders throughout the government who are ready to take off the shackles and get BIG things done for the American people - which is what @StephenM and @POTUS want. I can’t stress this enough: some of President Trump and RFK’s strongest campaign promises deal with institutional reform that democrats ran on too - not innately partisan issues. Don’t let the media talk us down. We can get BIG things done in the next year - and Americans want that.
1,547
4,028
194
30,398
1,100,163
Bit Whitman retweeted
Agree IREN is the next hyperscaler, @Joseph_Invests. A challenge for the smart $IREN bulls: IREN has >5GW of secured power and a >35GW pipeline, and it isn't slowing down. On Thursday's call Dan was asked about NVIDIA's modular DSX PODs. His answer: IREN has had teams on modular and prefab construction for nine months, is working with NVIDIA on designs, and is still deciding whether to run the DSX POD exactly as presented. The goal he named: simpler, faster builds with less labour on site. My read: that's the path to factory-built PODs that deploy in weeks and are quickly upgradeable at any time. Remember who they bought in 2020: PodTech. The name wasn't an accident. See photo from 2019. While most of the space faces headwinds, I think IREN has huge tailwinds. What else does this unlock? Add your tailwinds below. (Disclosure: large IREN shareholder, PodTech co-founder.)
$IREN WILL BE THE NEXT HYPERSCALER, HERE'S WHY: 35 GW Acquired 20 M/MW Deals Long Term 35 GW at 20 M/MW Equals 700 Billion ARR 700 Billion X 4 P/S Is a 2.8 Trillion Market Cap Yes I know this is crazy, yes I know this is a long term out but I am illustrating that $IREN could very well be the next hyper scaler
24
20
2
257
16,597
I feel like there’s going to be a competition for who can be more populist. Aka who can print more free money.
3
1
3
387
Bit Whitman retweeted
BofA Flow show is always hilarious with the Zeitgeist: “My base case now is an AI bust, stocks cause a recession, then a fiscal crisis. But my predictions mean squat right now and I could be buying chip stocks in another hour.”
12
8
93
13,839
Bit Whitman retweeted
A huge print is starting to come into view. We’re obviously not there yet, not even close, but the eventuality is taking shape. Traders should be developing a gameplan in advance -- things to keep on the screen and an understanding of historical precedent. Before a print happens, there can be a window in which inflation protection can be bought cheaply, because protection initially gets sold along with everything else. A good template for this (and everyone should get familiar with it) is gold’s decline in mid-late 2008, then its surge over the next few years as the Fed’s policy response to the crisis took hold. And note: In 2008, gold bottomed before stocks. The main questions in these situations are timing and price: How long does the window stay open for acquiring protection before a policy response takes hold, and how cheap does protection get? Big prints are rare (and history doesn’t repeat exactly) so there are no set answers or rules. Gold and BTC may have already seen most of their selling. But if opportunities do develop, investors with dry powder will be best positioned to take advantage of them (as always).
26
53
9
620
80,395
IREN IS NOT A NEOCLOUD. IT IS A VERTICALLY INTEGRATED AI SUPERFACTORY BUILDER AND PROVIDER @chamath
5
6
1
175
12,589
The vibes coming off my Sixers are just…😢
95
This remains the biggest market to watch.
AFTERMARKET SHARE RAID Goodman and $DLR just bid for ~10% of NEXTDC in an "opportunistic" transaction at a 45% premium. Implies a mid-teens multiple of EBIDTA once you add the debt to build their contracted pipeline. For context WULF is guiding to $1.5B in normalized NOI for Lake Mariner + Justified. Call it $1.4B after corporate costs. At15x implies $21B of enterprise value. Finishing the build may take $8B of net debt, which leaves $13B for equity vs. today's $5B market cap. Would be $26/share, not counting convert dilution. WULF's timeline is also better defined. About half of their contracted MW is targeted live by Q1 2027, whereas NXT just pushed M3 out into 2028. NXT +73% this week and hiring advisors to try to prevent a takeover. For weaker balance sheet names, I recommend hiring the advisors before your shares get sniped like this.
1
259
Reposting for the culture. $IREN
The Flywheel vs Treadmill — $IREN vs $NBIS: It's important to remember that over the long arc of a credit cycle, and specifically in a high-capex environment, terminal value goes to the player that manages its capital stack most efficiently. On that basis, it's worth comparing two diametrically opposed strategies. It's also important having a healthy debate about which stands the test of time, so I welcome everyone's opinion. Both $IREN and $NBIS buy the same chips but run very different capital strategies. The difference exists beneath the chip layer. $IREN owning its land, power and buildings outright is a structural advantage for building a flywheel. That layer doesn't depreciate, and once a building is stabilized it can be refinanced — equity comes back out and funds the next site. Rinse and repeat. $NBIS rents its buildings (though it's starting to build its own), and has only its GPUs and the contracts on them to offer as collateral. When the chips melt there's nothing underneath holding value, and financing the next buildout gets harder each cycle. Oversimplifying but it's easy to see IREN as a long megawatts trade and NBIS as long GPU residual value. To be clear, both are long the latter but IREN has the asset base that doesn't decay. To be clear, the chip market is strong right now and NBIS funds its fleet fine today — prepayments, operating cash flow, securitization. But these chips lose value every year no matter whose rack they sit in. That part is identical for both companies. Today's market is pricing appropriately the execution risk of $IREN and also the breakneck acceleration from $NBIS based on their "asset-lite" model. IREN's floor steps up every time a new site turns on. In $NBIS has to outrun its primary asset base from depreciating just to stay where it's at. The truth is chips are going to melt for everyone. The question is what you're standing on while they do. Also.. NO CRYING IN THE CASINO! @brianfry01 @_Sgr_A_Star @bitcoinbutcher1 @jiahanjimliu @SmallCapSnipa @MarkosAAIG @BitcoinAIGuy
8
280
Internalizing this.
If all you do is buy from forced sellers and never become a forced seller you’re going to be a very rich guy
1
85
Bit Whitman retweeted
$IREN dropped a site video with the Q1 print. went through it frame by frame as a civil engineer. here's what the build tables actually show Fully handed over: ✅ Prince George data halls 1-4 ✅ Childress Horizon 1 Substations are complete on every row of the 0.8GW table except canal flats. power goes in first, that's the right build order The rows in final stages: ✅ Horizon 2 and 3: structure and electricals done, cooling and gpu install in progress ✅ Mackenzie and childress air cooled halls 1-3: electricals, cooling and gpus all running in parallel ✅ Horizon 4: substation, foundations and structure done, electricals underway Next phase is already moving: ✅ Sweetwater 1 phase 2: high voltage substation complete ✅ Prince George data hall 4 liquid cooled: civils and both substations complete ✅ Badajoz, bundey and sweetwater 2: design done, procurement underway The hard part of a data center is the power and the substation IREN has that done across the 0.8GW and is already building it for what comes after Most of the remaining work on the 0.8GW is fit out, and that's the part that moves fastest
2
4
39
2,675
Bit Whitman retweeted
SpaceDCs: While Elon is Spamming SpaceDC Posts While $SPCX is Quietly Hoarding Powered Land Developments. $SPCX is proposing a 6GW site in Arizona. Development will not even start until 2028 and take several years to finish. $SPCX is projecting 10GW as 1-2GW across several sites. Filling a 6GW site will take 3 years at $SPCX speed and 6-9 years for others. Arizona is not in space last time I checked. Elon is making long term commitments to terrestrial DCs confirms what I've been saying. SpaceDC is completely feasible for $SPCX. The hard part is doing it economically at scale. I'm a believer in FSD and SpaceDCs, but let's make it clear, scaling FSD is much easier as it's software. SpaceDCs economic ramp will be much harder than FSD and FSD is going on year 12. By 2028, SpaceDCs will be in their early roadster phase.
4
5
71
9,769
Who did this shit lmao
343
2,052
359
13,781
981,915
Bit Whitman retweeted
Abundance is often built on unpopular infrastructure. @Melt_Dem on why today’s backlash to data centers and AI looks a lot like yesterday’s backlash to energy infrastructure.
The companies building superintelligence have the business model of an oil refinery. GPUs turn energy into compute, labs turn compute into tokens. As these commodities scale, new financial markets get built around them. My latest Slow Hunch is w/@CrucibleCap founder @Melt_Dem. Meltem started in oil and gas, was early to Bitcoin, and has spent years thinking about the relationship between energy, compute, materials, and markets. We get into the AI infrastructure boom, the coming neo-cloud shakeout, compute as a new asset class, and what happens when agents become economic actors. Its markets all the way down, and maybe all the way forward. Chapters 00:00 AI is a commodities business 01:50 Meltem’s slow hunch: energy, compute, and markets 06:25 From Bitcoin mining to the energy-to-compute thesis 10:03 The founding thesis behind Crucible 15:03 Why commodities and capital markets matter 20:18 The economics of the AI infrastructure boom 26:06 Debt, neo-clouds, and the coming compute shakeout 30:58 Compute and energy as the commodities of the AI era 34:43 Agents as traders and economic actors 37:47 What an agent-native economy might look like 40:34 Why markets may be the best way to distribute AI’s gains
4
3
1
45
3,556
And importantly, they started many many years ago. I’d expect that 1/3 of this pipeline is close to being available. Now go look at the deal RIOT just signed.
$IREN: ~35 GW GLOBAL PIPELINE UNDER DEVELOPMENT
5
276
$IREN +11% AH - decent quarter vs expectations given the MSFT issues. Sounds like they are preparing for possible capex cuts, isolating funding sources to hit $25B capex (guide $25-30B last quarter) without equity or converts. Sold compute at $40M/MW to Anthropic; and 90% prepay on a $22M/MW 5-year deal. Pricing keeps rising. More on call at 5pm.
18
17
3
295
16,363
Incredible numbers
3
48
1,283