The mining rig makers are eating the miners
That arrangement created a strange industry. Public miners raised billions in equity and debt to buy machines from private Chinese firms that captured most of the economic surplus of the network. Bitmain alone was reportedly generating gross margins north of 70% on its S19 series at the peak. The miners themselves ran on thin operational spreads, hostage to hash price and power costs, one halving away from insolvency. Whenever bitcoin ran, Bitmain got rich. Whenever bitcoin crashed, the listed miners went bankrupt. Core Scientific. Compute North. Iris Energy nearly. The pattern repeated every cycle.
Something changed in 2024 and it has been building steam ever since. A few operators, watching Nvidia's data center business go vertical and watching Trump's tariff regime hit Chinese electronics, decided the vertically integrated approach was not optional anymore. If you wanted to survive, you had to own the rig, own the power, own the site, and increasingly own an AI compute business bolted onto the same substation. The 2024 halving cut block rewards in half. The 2028 halving will cut them again. The only way to still be alive by then is to stop being a customer of your most important input.
One company saw this earlier than the rest. It was founded by Jihan Wu, the same Jihan Wu who co-founded Bitmain, then left after a boardroom war in 2021 with his co-founder Micree Zhan. Wu is not a subtle operator. He is the man who ran the Bitcoin Cash fork in 2017 and picked the fight with Roger Ver that split the community. He understands ASIC design at the transistor level because he helped invent the modern bitcoin miner. When he left Bitmain, the smart money assumed he would start another Chinese fabless designer. Instead he built something more ambitious.
He built a Singapore-headquartered, US-listed, vertically integrated operation that designs its own chips at Samsung and TSMC, builds its own rigs, runs its own data centers across Bhutan, Norway, Ethiopia, Texas and Ohio, sells hash rate as a service, and last year started shipping AI cloud infrastructure using the same power and cooling footprint. The chip design side is the hidden asset. This year the company taped out a 3nm bitcoin mining ASIC (called SEAL03) that on paper delivers around 5 joules per terahash. Bitmain's flagship S21 XP does about 13.5. The math on that is not subtle. If those specs hold in production and at scale, the entire installed base of North American mining hardware becomes obsolete inside 18 months, and the only non-Chinese company with a chip that can replace it sits in Singapore.
The second layer of the thesis is where it stops being about bitcoin. Every large-scale bitcoin miner in the US is now trying to convince the market they are actually an AI infrastructure play. Core Scientific signed with CoreWeave. TeraWulf signed with Fluidstack and Google backstopped the lease. Applied Digital signed with CoreWeave for HPC hosting. Hut 8 spun out American Bitcoin with the Trump family. The pivot is real, but most of these companies are landlords. They own the substation and the shell, and they rent it to someone else who owns the GPUs and the customer.
The company Wu built is doing something different. It designs the silicon. It manufactures the systems. It operates the sites. It sells the hash rate and the AI compute. It is the only listed bitcoin infrastructure company that owns every layer of the stack from transistor to kilowatt-hour to customer invoice. When Jensen Huang talks about Nvidia's moat being system-level integration rather than any single chip, he could be describing this business almost word for word, just applied to bitcoin instead of GPUs.
The market has not figured this out yet. The stock trades at a fraction of what a pure-play AI infrastructure name at similar revenue would fetch, because most analysts still classify it as a bitcoin miner and mark it to the hash-price cycle. That is the mispricing.
The only way to still be alive by the 2028 halving is to stop being a customer of your most important input.
You're reading the open chapter.
The full Deep Dive — including the contrarian thesis, the chain reaction, the stocks we rate True North, and the catalysts we're tracking — is reserved for Compass members.
- The full contrarian thesis with our supporting data
- 3–5 specific stocks we rate True North on this theme
- Catalysts and dates we're watching for confirmation
- Saturday Deep Dive every week (52/year)
- Free Tuesday & Thursday e-letters included
- Full Compass Ratings on 4,000+ stocks
$9/mo · $89/yr · $499 lifetime. Cancel anytime. 7-day money-back guarantee.