Bitmain vs MicroBT: When the Chipmaker Also Mines
Bitcoin's oldest hardware rivalry is really a fight between business models, and the November 2026 crash is the stress test. Here is how Bitmain, MicroBT, and Bitdeer actually make money.
A Hardware Fight That Gets Louder in a Downturn
On November 22, 2026, Bitcoin changed hands near $60,000, having just clawed back a 9 percent bounce off a two-year low. The number that funds the whole system, hashprice, told a darker story: it had slumped to a record low near $26 per petahash a day. A week earlier, network difficulty was cut by 9.13 percent to 134.82 trillion, its third-largest drop of the year, as roughly 100 EH/s of machines went dark. When revenue per unit of work falls like that, the question of which machine you run, and which company stands behind it, stops being a spec-sheet hobby and becomes the line between a profitable site and a powered-down one.
This is the backdrop for the oldest rivalry in Bitcoin mining: Bitmain against MicroBT. One is the Beijing-born giant whose Antminer rigs and Antpool pool have shaped the industry for a decade. The other is the Shenzhen challenger whose Whatsminer machines pried loose a slice of that empire. A third name, Bitdeer, now crowds the stage with its own SEAL chips, and it was born directly out of a Bitmain boardroom feud.
What follows is not another joules-per-terahash shoot-out; the chips have been measured before. It is about business models. Bitmain, MicroBT, and Bitdeer do not only sell machines. Two of the three also mine with them, run pools, and hold Bitcoin on their own balance sheets. In a month when the coin briefly cost more to mine than it fetched, the model, not just the chip, decides who survives, and it hands an uncomfortable share of the network to a handful of private firms. The sections below cover who they are, the feud that made them, how the hardware stacks up, and why the way each company earns a living matters more than any single efficiency rating.
Who Bitmain, MicroBT, and Bitdeer Actually Are
Start with the roster, because the names blur together. Bitmain, founded in Beijing in 2013 by Jihan Wu and Micree Zhan, is the incumbent: it designs chips, builds the Antminer line, runs the Antpool mining pool, and has historically mined for its own account. It is private, China-rooted, and now building in the United States.
MicroBT, founded in Shenzhen in 2016 by Yang Zuoxing, is the specialist. Yang was a chip-design lead at Bitmain who worked on the Antminer S7 and S9 before leaving to turn Whatsminer into the only credible rival. MicroBT sells hardware and little else: no dominant pool, limited self-mining, a reputation built on durability and value rather than record efficiency.
Bitdeer is the newcomer that is not really new. Listed on Nasdaq under the ticker BTDR and headquartered in Singapore, it is the vehicle Jihan Wu kept when he left Bitmain in 2021. It designs its own SEAL chips, sells SEALMINER rigs, and runs one of the largest self-mining fleets in the world, on the order of 80 EH/s. Because it files with the SEC, it is the one member of this trio whose economics you can actually read.
| Company | Founded | Founder(s) | HQ | Hardware line | Own pool | Self-mines | Public | Est. ASIC share |
|---|---|---|---|---|---|---|---|---|
| Bitmain | 2013 | Jihan Wu, Micree Zhan | Beijing | Antminer | Antpool | Yes | No (private) | ~55 to 80% |
| MicroBT | 2016 | Yang Zuoxing | Shenzhen | Whatsminer | None | Limited | No (private) | ~15 to 30% |
| Bitdeer | 2021 (spun from Bitmain) | Jihan Wu | Singapore | SEALMINER | None | Yes (~80 EH/s) | Yes (Nasdaq: BTDR) | New entrant |
Note the pattern that the rest of this piece turns on: the biggest machine makers are also among the biggest miners. The hardware business and the mining business are not separate industries. They are, increasingly, the same companies wearing two hats.
The Founder Feud That Built a Duopoly
The three-way race has one family tree. Bitmain’s co-founders fell out spectacularly between 2019 and 2021. Micree Zhan was ousted, fought his way back, and in an early-2021 settlement bought out shareholders aligned with Jihan Wu for a reported sum near $600 million, keeping Bitmain, the Antminer brand, and Antpool. Wu walked away with Bitdeer and BTC.com. So the two men who built the dominant miner now run competing hardware companies: Bitmain is Zhan’s, Bitdeer is Wu’s.
MicroBT is the other branch. Yang Zuoxing led chip design on the S7 and S9, then left Bitmain in 2016 after being refused equity, and started a rival. Bitmain did not take it quietly. It sued MicroBT and Yang in 2017 over an ASIC patent; the Beijing Intellectual Property Court dismissed the case in 2018 and revoked the patent. Yang was briefly detained in late 2019 in a separate embezzlement matter tied to his Bitmain years. Then, on April 7, 2021, Bitmain filed again at the same court, this time a trade-secret claim seeking as much as 99 million yuan (about $15 million) over designs linked to the S7 and S9.
The dates matter, because those suits are old news, not a fresh 2026 clash. The lasting point is the irony buried in them: two of the three top ASIC makers trace straight back to Bitmain, one through its estranged co-founder and one through its former chief chip designer. The duopoly, and the trio it has become, grew out of a single company’s internal fights.
The Efficiency Frontier: S23 XP, A4 Ultra, and Whatsminer
On to the hardware, where the unit that matters is J/TH, joules per terahash. Lower is better, because electricity is the dominant cost of mining; a machine that does the same work for fewer watts simply outlives its rivals when margins thin. In late 2026 the frontier sits below 10 J/TH, and Bitmain holds the crown again.
Bitmain’s Antminer S23 XP Hyd runs 600 TH/s at 8.9 J/TH (about 5,340 watts), unveiled at the World Digital Mining Summit in August 2026 and shipping from November at $25.50 per terahash, around $15,300 a unit. It is the first Antminer flagship under 9 joules. The earlier S23 Hydro does 580 TH/s at 9.5 J/TH, and its dense 3U chassis packs roughly 1,160 TH/s; in a November profitability snapshot, that S23 Hydro 3U was the single most profitable machine on the market, near $8.74 a day at ten-cent power.
Bitdeer is right on its heels. The SEALMINER A4 Ultra Hydro does 886 TH/s at 9.45 J/TH and briefly held the efficiency record when it launched in April 2026; on that same November list it placed second, around $6.79 a day. MicroBT competes a step back. Its M70 Whatsminer series, launched at Bitcoin MENA in Abu Dhabi in December 2025, tops out near 12.5 J/TH on its best hydro tier, with much of the fleet closer to 13.5, and its petahash-class M79S pushes past 900 TH/s but draws roughly 14 kilowatts. MicroBT, in other words, trails the frontier by three to four joules.
| Model | Maker | Hashrate (TH/s) | Efficiency (J/TH) | Cooling | Status (late 2026) |
|---|---|---|---|---|---|
| Antminer S23 XP Hyd | Bitmain | 600 | 8.9 | Hydro | Shipping from Nov 2026 |
| Antminer S23 Hyd 3U | Bitmain | ~1,160 | 9.5 | Hydro | Shipping |
| SEALMINER A4 Ultra Hydro | Bitdeer | 886 | 9.45 | Hydro | Shipping |
| SEALMINER A4 Pro Hydro | Bitdeer | 680 | 10.9 | Hydro | Shipping |
| Whatsminer M70 series (top tier) | MicroBT | ~540 to 600 | ~12.5 | Hydro | Shipping |
| Whatsminer M79S | MicroBT | ~930 to 1,040 | ~13.5 | Hydro | Shipping |
The gap has not stopped MicroBT from gaining ground, because efficiency is only half the sale. Bitmain, as Blockspace Media founder Colin Harper wrote for Forbes, long held the upper hand on price, but compressed margins after the 2024 halving and the new tariff regime have flipped what he called a seller’s market into a buyer’s market, with customers now negotiating from strength. When nobody is forced to buy the absolute best chip, a durable, cheaper, easier-to-service machine finds plenty of homes.
Market Share Nobody Can Audit
How much of the market does Bitmain really own? Nobody can say with authority, because none of the three publish audited figures. By installed hashrate, the common shorthand puts Bitmain around four-fifths of the network, MicroBT near 15 percent, and Canaan a couple of percent. By recent shipments and revenue, the picture is less lopsided: research from analytics firm Spark estimates Bitmain at 50 to 65 percent, MicroBT at 20 to 30 percent, and Canaan at 5 to 10 percent, with the three together controlling more than 90 percent of the market.
Both readings can be true at once. Bitmain dominates the installed base built over a decade, while MicroBT and Bitdeer win a growing share of the newest orders. Either way, a single private company makes the plurality, and quite possibly the majority, of the machines that secure a trillion-dollar asset, and its exact weight is a matter of estimate rather than disclosure. Treat every precise percentage in this industry, including the ones here, with suspicion; the uncertainty is itself a governance fact.
Three Business Models, Not One Rivalry
Calling the contest Bitmain vs MicroBT frames it as a fight over chips. It is better understood as a fight between three business models.
Bitmain is the empire. It designs chips, builds rigs, runs a top-two pool, mines for itself, and has pushed into artificial-intelligence silicon through an affiliate. Its revenue comes from selling the frontier machine at a premium and from the network effects of an integrated stack that touches almost every stage of mining. When hardware sales slump, as they have in this downturn, its pool and mining arms keep cash moving.
MicroBT is the specialist. It sells Whatsminer and little else: no dominant pool, limited self-mining, a reputation for machines that keep running and are easy to service. Its edge is being the clean, focused alternative to a sprawling incumbent, and it has leaned hardest of the three into American manufacturing. The trade-off is that it feels a hardware slump with no second revenue leg to soften the blow.
Bitdeer is the public integrator. Like Bitmain it is vertically integrated, chips plus rigs plus a large self-mining fleet, but unlike Bitmain it answers to public shareholders and discloses its numbers. It sells SEALMINERs and mines with them, and it can point investors to monthly production updates rather than ask them to trust a private black box.
These are not just different revenue mixes; they drive different behavior. A pure vendor like MicroBT has every reason to sell its best machine to anyone who will pay, as early as it can. An integrated maker that also mines faces a subtler calculus: it can deploy the newest, most efficient rigs in its own data centers first and sell the prior generation to the market, timing releases to its own fleet-refresh cycle. None of the makers say they work this way, and it is hard to prove from the outside, but the incentive is baked into the structure, which means the company you buy from is partly the company that decides when you get the best silicon.
The through-line is simple: two of the three biggest names in mining hardware are also among the biggest miners, and that dual role, more than any efficiency rating, shapes how they price machines, when they release them, and who ends up holding power over the network. That is where the interesting, and uncomfortable, questions begin.
When the Chipmaker Also Mines: Antpool and Vertical Integration
Bitmain’s version of integration is the sharpest case, because it makes most of the machines and runs Antpool, one of the two largest mining pools, at roughly 19 to 21 percent of network hashrate. Analysts at Spark describe this as both Antpool’s primary advantage and its most concerning feature: a hardware maker that runs a dominant pool can steer hashrate from its own mining operations, shape pool defaults through firmware, and use its market position to defend pool share. New Antminers, the same research notes, often ship pre-configured to point at Antpool, a quiet default-bias pipeline, and several smaller pools are argued to route through Antpool’s infrastructure as proxies, which would make its real influence larger than the public number suggests.
Bitmain has flirted with separating the two halves before. Back in 2021, during the founder split and one of its periodic pushes toward an initial public offering, the company told shareholders it had completed the legal steps to spin Antpool into a standalone overseas entity. Years on, the two remain closely associated in practice, and the concern has not faded.
Put plainly, when the company selling you a machine also competes with you for block rewards and directs where a fifth of the network points its hashrate, its incentives and yours are not identical. None of this is illegal, and Bitmain rejects the idea that it abuses the position, but the structure is unusual. In most industries a dominant supplier that also competed downstream and controlled distribution would draw a hard regulatory look.
The Pool Problem Hardware Makes Worse
Zoom out from one company to block production, and the concentration compounds. Foundry USA, owned by the crypto conglomerate Digital Currency Group, sits around a quarter of network hashrate; Antpool sits around a fifth. Together, two pools build close to half of all Bitcoin blocks. Hardware concentration and pool concentration stack on top of each other, because the dominant chipmaker also operates a top-two pool.
Two pools at roughly half the hashrate does not mean a 51 percent attack is looming; pools are coordinators, and the hashrate they point belongs to thousands of independent miners who can switch in minutes. But the arrangement concentrates softer forms of power: which transactions get mined, which standards get adopted, and whose infrastructure the network leans on day to day. Concentration at the hardware layer and concentration at the pool layer are different problems, and Bitmain is one of the few entities that sits near the center of both.
Pools matter because whoever assembles a block decides which transactions go in and in what order, the same ordering power that fuels MEV across the rest of crypto. On Bitcoin that leverage has long been understated, but the rise of Runes, Ordinals, and real fee competition has turned block construction into a live question rather than an academic one. A pool operator that also sells the machines has more levers to press than one that merely coordinates hashrate.
The pattern of a single well-capitalized group quietly accreting influence over a network that sells itself as leaderless is not unique to mining. HOGE Wire has traced the same shape in AI, where DCG’s orbit built a center in a protocol that promised none. The counterweight in mining is an industry push toward open block-construction standards: pools representing about 75 percent of hashrate have joined a working group around Stratum V2 and related protocols that let individual miners, not the pool operator, choose the transactions in their blocks. If it holds, it blunts the worst of the concentration without having to break up the pools.
Silicon’s Single Point of Failure
Beneath the rivalry sits a dependency both companies share and neither can escape. Every Antminer, Whatsminer, and SEALMINER is only as available as the foundries that print its chips. Bitmain fabricates its advanced silicon at TSMC in Taiwan; MicroBT and Canaan have leaned on Samsung in South Korea. Two Asian foundries, in short, sit upstream of nearly all Bitcoin security, and a disruption at either, whether a natural disaster, a blockade, or an export rule, would ripple straight into hashrate.
Export controls already bite. In January 2025 the US Commerce Department added Sophgo, an AI-chip venture tied to Bitmain’s Micree Zhan, to its Entity List after a TSMC-made chip matching a Sophgo design turned up in a Huawei processor; TSMC halted the shipments and Sophgo denied any Huawei link. The action does not ban Antminers, but it shows how exposed the duopoly’s move into AI silicon is to Washington, and how quickly a chip-supply question can become a geopolitical one.
The dependency is deepening, not easing. Each turn of the efficiency race, from 5nm toward 3nm-class nodes, pushes the makers onto more advanced and more concentrated fabrication that only a handful of production lines on Earth can run. That is what buys the headline J/TH gains that sell machines, but it also tightens the industry’s reliance on a sliver of global semiconductor capacity it does not own and cannot replicate quickly. The better the chips get, the narrower the base of foundries able to make them.
The slowest-moving risk is in the cryptography itself. The same advanced process nodes that print these mining chips are the frontier where longer-term questions about Bitcoin’s cryptographic durability eventually land, the kind of exposure that global regulators have begun to flag and that HOGE Wire examined through Bitcoin’s quantum-and-Taproot lens. None of it is urgent for a miner deciding what to buy this quarter, but it is a reminder that the hardware layer, not just the protocol, carries systemic weight.
The Reshoring Race: Made in the USA
Tariffs redrew the map. Since August 2025, Southeast-Asia-assembled rigs face about 21.6 percent in US import duty, and China-origin machines up to roughly 57.6 percent, against an old baseline near 2.6 percent. Ethan Vera, chief operating officer of mining-services firm Luxor, put the stakes bluntly to The Block: at 21.6 percent tariffs, he said, the United States is now one of the least competitive jurisdictions to bring machines in, and miners are looking at Canada and other markets instead.
Both makers answered by building stateside. MicroBT moved first and hardest, starting US assembly in 2021, striking a supply deal with Luxor, and opening a US online shop in 2025 with domestic warehousing; its pitch, in its own words, is American-made, tariff-free mining rigs. Bitmain opened its first US production line in December 2024 and says it is scaling toward a full American factory; at the S23 launch, Bitmain vice president Irene Gao said the company was working to expand its manufacturing capabilities in the United States. The asterisk on all of it: assembly is not fabrication. The chips still cross the Pacific, so American-made describes the final assembly work, not the silicon.
Here Bitdeer’s structure is a quiet edge. US-listed and Singapore-based, it sidesteps the China-origin stigma that clings to its two rivals, one more way the corporate model, rather than the chip, shapes who wins which order.
The AI Escape Hatch
Every miner is eyeing the same hedge, and so are the companies that sell to them: artificial intelligence and high-performance computing. Bitmain’s Sophgo tie is an explicit bet that the chip-design muscle behind Antminer can also serve AI accelerators, which is precisely why the Entity List action stung. The makers’ own customers, the public miners, are converting Bitcoin sites to AI data centers at speed, which changes what they want from a rig and how long they plan to run it.
For the manufacturers, AI is both threat and outlet. It is a threat because every megawatt a customer gives to AI is a megawatt not buying Antminers or Whatsminers. It is an outlet because the same silicon expertise can chase far fatter margins than Bitcoin mining offers at a $26 hashprice. Expect the roadmaps to split: ever-more-efficient SHA-256 machines for the miners who stay, and a pivot toward AI and HPC chips for everyone hedging their energy bets.
The shift is already visible on the ground. Several of the largest public miners have signed multi-billion-dollar deals to convert Bitcoin sites into AI and high-performance-computing data centers, chasing tenants with far deeper pockets than block rewards provide. For the hardware makers that is a double-edged trend: it thins the pool of buyers for new SHA-256 rigs even as it validates the bet that their chip and data-center expertise can travel. Bitmain and Bitdeer, which already run their own sites, are better placed than a pure vendor to straddle both markets; MicroBT, the specialist, has the narrowest hedge if mining demand keeps leaking to AI.
The Stress Test: Which Model Survives a $26 Hashprice
Return to November 2026. With hashprice near a record low and many rigs underwater, the business model becomes the shock absorber. Fred Thiel, chief executive of MARA, has long described mining as a zero-sum game in which, as more capacity joins, margins compress and the floor is your energy cost. When that floor climbs above revenue, the integrated players have cushions a pure hardware vendor does not.
Bitmain and Bitdeer self-mine, so when machine sales slow, mined coins and pool fees keep revenue flowing, and both can deploy their own newest, most efficient rigs first. MicroBT, the pure vendor, feels the sales slump most directly, yet it also carries no mining losses and no pool-centralization baggage, and its value positioning plays well precisely when buyers hunt for the cheapest working terahash. The machines themselves sort fast: at ten-cent power and a high-$20s hashprice, only the sub-10 J/TH frontier clears more than a few dollars a day, while anything above roughly 15 J/TH is bleeding.
| Rig | Efficiency (J/TH) | Approx. profit/day at $0.10/kWh | Illustrative breakeven power price |
|---|---|---|---|
| Antminer S23 Hydro 3U | 9.5 | ~$8.74 | ~$0.12/kWh |
| SEALMINER A4 Ultra Hydro | 9.45 | ~$6.79 | ~$0.12/kWh |
| Whatsminer M70 (top tier) | 12.5 | thin | ~$0.09/kWh |
| Older S19-class | ~21.5 | negative | ~$0.05/kWh |
Daily-profit figures come from a November snapshot at ten-cent power; the breakeven column is illustrative, computed as hashprice divided by 24 hours times efficiency at a roughly $28 hashprice. The takeaway is the same spread HOGE Wire detailed when the median miner’s all-in cost pushed through the spot price in a $57,000 stress test: a downturn does not kill the industry, it sorts it. The makers with a second revenue leg are best placed to buy distressed hashrate on the way down, which is how each washout tends to leave the biggest players bigger.
What It Means for Miners Buying Today
For anyone actually buying machines, the vendor debate collapses into a procurement checklist, and in this market every line on it is a survival decision rather than a preference.
- Efficiency first: at a high-$20s hashprice, a rig much above 12 J/TH needs nearly free power to make sense. The frontier (Bitmain’s S23 XP at 8.9, Bitdeer’s A4 Ultra at 9.45, the S23 Hydro at 9.5) buys the most runway.
- Price and resale: weigh Bitmain’s premium frontier against MicroBT’s cheaper value tiers, and remember that secondary-market prices fell with hashprice, so any rig is a fast-depreciating asset, not a store of value.
- Supply chain and duty: where a machine ships from decides your landed cost, 21.6 percent versus 57.6 percent versus US-made, and MicroBT’s and Bitdeer’s structures can mean a cleaner import.
- Durability and support: MicroBT’s serviceability reputation and US repair hubs matter most to operators who cannot babysit a fleet.
- Monoculture risk: a fleet running entirely on one vendor’s firmware inherits that vendor’s defaults, including pool bias, so spreading orders across makers is a hedge small miners often overlook.
The honest summary for a buyer: Bitmain and Bitdeer sell the best economics if you can run them, MicroBT sells the fewest headaches, and the right answer is usually a mix, chosen for your power price rather than for a brand.
What It Means for the Network
Beyond any single operator’s profit and loss, the rivalry decides something larger. Bitcoin’s security rests on hardware from a near-monopoly supplier, fabricated at two foreign foundries, with the dominant maker also running a top-two pool and mining for itself. That is a lot of single points of failure for a system whose entire pitch is having none.
The counterweights are real but partial. MicroBT is a credible challenger that keeps Bitmain honest on price and availability. Bitdeer is a vertically integrated rival that at least discloses its numbers. The Stratum V2 push gives individual miners a say over block construction. Reshoring spreads the geography of assembly, even if not of fabrication. None of it dissolves the concentration; it only dilutes it.
The honest read for 2026 is that the Bitmain-versus-MicroBT rivalry is healthy for the network precisely because it is a rivalry; the day it stops being one is the day to worry. The downturn, brutal as it is for miners, does the network a backhanded favor: a record-low hashprice forces exactly the competition, efficiency, and balance-sheet discipline that a comfortable monopoly never would. When the chipmaker also mines, the best check on its power is a rival chipmaker that also mines, and right now Bitcoin still has two of them.
Frequently Asked Questions
What is the difference between Bitmain and MicroBT?
Bitmain, founded in Beijing in 2013, makes the Antminer line, runs the Antpool mining pool, and mines Bitcoin for itself; it is the market leader by a wide margin. MicroBT, founded in Shenzhen in 2016 by a former Bitmain chip engineer, makes the Whatsminer line and focuses on hardware without a dominant pool. Bitmain generally leads on peak efficiency, while MicroBT competes on value, durability, and US-made supply.
Which is more efficient in 2026, Antminer or Whatsminer?
Bitmain’s Antminer holds the efficiency crown. Its S23 XP Hyd runs at 8.9 J/TH, the first Antminer flagship under 9 joules per terahash, while MicroBT’s best Whatsminer tier sits around 12.5 J/TH. Bitdeer’s SEALMINER A4 Ultra lands between them at 9.45 J/TH. Lower J/TH means less electricity per unit of mining, which is decisive when hashprice is low.
Why does it matter that Bitmain runs a mining pool?
Because a company that makes most mining machines and also operates a top-two pool (Antpool, around a fifth of network hashrate) has unusual influence: it can steer its own hashrate, set pool defaults through firmware, and shape block construction. Critics call it a structural conflict of interest and a centralization risk, while Bitmain disputes that it abuses the position.
Is Bitdeer a competitor to Bitmain and MicroBT?
Yes. Bitdeer is a Nasdaq-listed company (ticker BTDR) founded by Bitmain co-founder Jihan Wu after he left in the 2021 split. It designs its own SEAL chips and sells SEALMINER rigs that now rival Antminer and Whatsminer on efficiency, and it runs a large self-mining fleet of roughly 80 EH/s, which makes it both a hardware vendor and a major miner.
Did Bitmain sue MicroBT?
Yes, more than once. A 2017 patent suit was dismissed in 2018, and in April 2021 Bitmain filed a trade-secret case at the Beijing Intellectual Property Court seeking up to about $15 million over designs tied to the older S7 and S9 chips. MicroBT’s founder was also briefly detained in late 2019 in a separate matter. These disputes are several years old, not recent events.
Marcus Okafor covers mining hardware and network infrastructure for HOGE Wire.