Money Qubit

Who i am and what i own. Civil engineer. co founder of a construction company. i run building sites for a living. That matters here. i look at a data center company like one of my sites. how much power, how much land, cooling, and is the schedule real or just a slide. Investing is not my day job. i follow a few and i post about those only. The book: Ai infrastructure $IREN $NBIS Physical ai and sensors $OUST $ONDS Space $RKLB $ASTS Photonics and chips $MRVL $COHR Crypto exposure sits in $BLOX. How i pick: ✅ i look at what the company has built first. numbers come after ✅ i like companies that built something real but are not earning from it yet ✅ i buy only on red days. i sell only on green days ✅ after i finish reading, i wait weeks before i buy What you will not get here: ❌ price targets ❌ buy or sell calls ❌ opinions on stocks i do not own ❌ paid promos, ever Positions change. when they do i update this post. my own money, my own research. not advice. Last updated 18 aug 2026

$IREN ceo on demand: "any GPU you can bring online, you have got multiple people all wanting it" ✅ 2026 sold out ✅ 2027 conversations cover multiples of the capacity they can bring online ✅ 2028 already coming into focus And customers aren't just buying, they're helping fund the build: ✅ ~$22m per MW on the new 5yr deal with 90% of gpu capex prepaid ✅ $40m per MW on the new short term deal Pricing isn't going down. the CEO says the evidence suggests it's going up Demand is there. now customers are paying to help build it

$IREN carries $6bn of data center assets on its balance sheet, and management has previously described its data center portfolio as unencumbered. Owning infrastructure creates another potential funding route as the cloud business scales. That flexibility matters when capital gets expensive. Additional financing still needs to be secured.

Anthropic made $33bn of revenue in 9 months. its IPO filing shows $895bn of compute already signed, much of it take-or-pay $70bn of that is due in the next 12 months Every dollar ends up as GPUs, power and data centre rent. who gets paid first?

Bitcoin's hashrate grew every year from 2016 to 2025. this november it's flat to down 4% on a year ago Mining pays less, and miners are moving their power to AI racks Fewer machines chasing the same $BTC. relief for the miners left, or a security worry?

One chart: in Q4 2024, 91% of $CORZ revenue came from bitcoin mining. last quarter it was 7% AI colocation filled the gap, growing from $15m to $215m a quarter in a year. management says mining ends by year end Power was the asset all along. which miner flips next?

$QNT revenue jumped 233% to $9.1m. the first nine months are down 7%, and R&D ran $10.80 per $1 of revenue The real report card is the roadmap: 100 logical qubits in 2027, fault tolerant in 2029 That's the date $BTC should circle. post quantum ready by then?

Imagine paying 10x more for the exact same thing. that's zcash ETFs right now What's an ETF? a way to own a coin through your normal stock account. no wallet, no keys, no crypto app So what's the deal? grayscale's zcash fund is the only one you can buy in the US today. it charges 2.5% a year. the winklevoss twins just applied to launch one that charges 0.25% What does that mean in real money? put $10,000 in. grayscale takes $250 a year. winklevoss would take $25. same coin. 0.25% is what the biggest $BTC fund charges Can you buy the cheap one yet? not yet. it still needs a green light from the SEC Here's the catch though. the whole point of $ZEC is privacy. you can send money without the world seeing who paid whom or how much. buy it through an ETF and you skip that part. the winklevoss fund says its coins will sit in addresses anyone can check on the blockchain And you don't get a vote either. zcash holders vote on big decisions, like how its development money gets spent. but only with coins in their own private wallet. ETF owners can't vote My read: the cheaper fund is great news for buyers. but through an ETF you own the price, not the privacy Would you buy zcash through an ETF, or hold it yourself?

$GOOGL's waymo did 50,000 paid rides a week in may 2024. by spring 2026: 500,000. 10x in two years Still burning cash: the alphabet unit that houses it lost $2.0bn last quarter on $372m of revenue Every robotaxi is a GPU customer, on the road and in the cloud. 1m by year-end?

$IREN signed a 5 year deal with customer prepayments covering 90% of the associated gpu capex. My bullish takeaway: customers are helping finance expansion. If repeatable, that could reduce the new shareholder capital needed to grow.

$IREN said it contracted H100 capacity at double the price of its previous H100 contract at Prince George. That is encouraging evidence that older gpus can retain commercial value as newer chips arrive. One contract cannot prove lifetime returns, but it strengthens the case for useful revenue beyond the first deal.

$WULF booked a $710m profit on $76m of revenue. the business itself lost $88m The gap is a $728m paper gain on warrants. diluted EPS was minus 3 cents The real progress is rent: AI leases are 83% of revenue now. MW online or headline profit, which do you watch?

Zcash is in a fight right now over a $120m grants fund, and it comes down to one question. who should get the new coins? Where do new $ZEC go today? 80% of every block reward goes to miners, 8% to the zcash community grants committee and 12% to a fund coinholders vote on. on $BTC, miners get all of it How much is 8%? about 144 ZEC a day, roughly $190k. the committee also holds about 91,500 ZEC in its treasury, near $120m What did the committee offer? on oct 29 it proposed moving 36,200 ZEC into the new miner reserve from this week's upgrade, and taking less in future. about 98,000 ZEC in total. critics like helius ceo mert mumtaz called it "nowhere near good enough". it pulled the plan this week What's on the table now? co-founders sean bowe and zooko wilcox back a draft that ends the 8% stream, parks it in a dev and legal defense lockbox, and has the committee send 50,000 ZEC to the miner reserve When does it get decided? any change likely needs a network upgrade and there's no date. the current funding ends at the 2028 halving anyway unless a new ZIP extends it My read: every proof-of-work chain ends up arguing about this. once the subsidy shrinks, every coin sent somewhere other than miners is security you aren't buying Should a coin fund its own developers from block rewards, or should miners get everything like bitcoin?

$ZEC just got 3x faster. the NU7 upgrade went live today at block 3,543,405, and blocks now land every 25 seconds instead of 75 Does that mean 3x more new coins? no. the reward per block was cut by the same 3x, so issuance stays the same. halvings are now spaced 3x more blocks apart, so the 21m cap and the schedule don't move Is it actually faster? since the switch the chain has averaged about 27 seconds a block. three confirmations now take about a minute and a quarter, down from almost four What changed with fees? 60% of every block's fees now go into a public reserve. miners keep 40%. the reserve starts paying back out through block rewards around 2031 Why do that? same problem $BTC has. rewards keep halving, so one day fees have to pay for security. this smooths the fee side out before that day comes Anything lost? yes. sprout, the oldest privacy pool from 2016, is now shut. about 12,200 ZEC never moved out and can't be spent, around $16m. a recovery idea is on the table but nothing is approved My read: the fee reserve is the part to watch. every proof-of-work chain, bitcoin included, hits the same security budget question once the subsidy shrinks Would you want something like this on bitcoin, or is that a line you don't cross?

A UK power station with 4GW of grid access wants 1GW+ of it for AI. drax is teaming up with brookfield Phase 1 is 100MW from early 2028, and it still needs a tenant and planning Owning the grid hookup is the moat, same play as miners turned AI clouds. which plant is next?

$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

A year ago bitcoin mining was 97% of $IREN revenue. this quarter AI cloud is 88%, $200m of $228m Revenue barely moved, $240m to $228m. the business underneath is new, with $20.3bn of backlog

$IREN as a civil engineer this is the slide i'd frame 14 months from starting civil works to an NVIDIA Exemplar tested 19k GB300 cluster at Horizon 1 Industry benchmark: 18 months for the data center alone, no GPUs ✅ That 14 months includes gpu deployment. the 18 month benchmark doesn't ✅ My read: speed comes from building on land and power they already own. the hard part is done before the first pour ✅ Operating capacity now 0.1GW IT, +127% QoQ (company says) Next stop 0.3GW by january, then 0.8GW targeted by end of 2027 And the land bank behind it: ✅ >30,000 acres secured ✅ >5GW secured power Every new GW still has to go through the same civils, the same grid connection, the same build $IREN has shown it can do that faster than the benchmark Speed of delivery is the moat money can't rush

$IREN is carrying $2.5bn of deferred revenue Sep 30: $2,544m June 30: $1,842m That's customers paying ahead of delivery With AI infra names getting sold on debt worries, this matters: ✅ Q1 operating cash flow $529m. Deferred revenue inflow alone was $645m ✅ Post quarter 5yr deal: company says the customer prepays 90% of the GPU capex ✅ My read: customers only pay ahead like this when they can't get the capacity anywhere else. That's the power moat showing up in the cash flow statement ❌ Take the deferred revenue inflow out and operating cash flow was roughly -$116m ❌ Net debt went from $216m to $4.43bn in one quarter ❌ FY27 capex still to spend: $15bn. $9bn covered, $6bn is still a target "subject to market conditions" ❌ ARR target of >$4bn slipped from year end to jan 31, may run to end of feb Deferred revenue is a liability. IREN pays it back by delivering Every month of delay it sits there as a liability instead of turning into revenue Delivery is the whole trade now $IREN

$ZEC NU7 activates at block 3,543,405 🔐 • Blocks: 75s → 25s (3x faster confirmations) • Same daily issuance, 21M cap intact • Sprout pool closed for good • V4 transactions retired Not a new pool. That’s Tachyon, still in development.

Buy the dip, I said. 
The dip: also buying the dip. 📉 
$IREN $NBIS $CRWV