Bitfury sold the shovels, then bet the winnings

Most companies that got rich on Bitcoin got rich the obvious way - they bought the coin, held it, and watched a number go up. Bitfury took a different route. It looked at the whole gold rush and decided the durable business was not the gold but the equipment: the chips that do the mining, the cooling that keeps them alive, and the software that makes the ledger useful to someone other than a speculator. For fourteen years it sold the shovels. In 2025 it did the thing almost nobody in crypto does - it put the shovels down.

Founded in 2011 by two engineers who share a first name - Valery Vavilov, the chief executive, and Valery Nebesny, the chip designer - Bitfury spent its first decade quietly assembling one of the deepest hardware-and-software stacks in the industry. It was never the loudest name in the room. But when a government wanted to put land records on a blockchain, or a Nasdaq-bound miner needed machines by the container-load, Bitfury was often the company on the other end of the phone.

The unglamorous layer

Bitcoin mining is, underneath the mythology, a hardware problem. The network rewards whoever can compute a particular kind of hash fastest and cheapest, which means the game is won on two fronts: better silicon and cheaper power. Bitfury went after both. On the silicon side it designed its own application-specific chips - ASICs - purpose-built to do one calculation and nothing else. Six generations later, the last of them, called Clarke, arrived in 2018 with the kind of efficiency numbers that only matter to people who pay very large electricity bills.

On the power side, the company built BlockBox - a data center that fits inside a shipping container. The logic is almost comically practical: if the cheapest electricity is next to a hydro dam in a remote valley, do not build a campus there and wait a year. Ship the data center to the power. Plug it in. Monitor it from anywhere.

Then there was the heat. Racks of ASICs run hot enough that cooling becomes its own engineering discipline. In 2015 Bitfury bought a Hong Kong outfit called Allied Control and leaned into two-phase immersion cooling - literally submerging the electronics in a special liquid that boils off the heat. It worked well enough that, in 2021, the whole thing was spun out as a standalone company, LiquidStack. A trick invented to keep Bitcoin miners from cooking themselves is now sold to data centers that have never touched crypto.

There is a personal thread running under all of this. Vavilov grew up as the Soviet Union came apart, and he has said in interviews that watching savings evaporate overnight taught him something about who controls money and what happens when that control fails. It is not hard to draw a line from that memory to a company built around decentralization, ledgers no single party owns, and records anchored so they cannot quietly be edited. Founders tend to build against the thing that scared them, and Bitfury reads like a long answer to money becoming paper.

Software with a paper trail

Hardware was half the company. The other half was a bet that blockchains could do something more useful than move coins around. Bitfury built Exonum, an open-source framework for permissioned blockchains that periodically anchors its records to the Bitcoin chain - borrowing Bitcoin's security without asking anyone to buy Bitcoin. The flagship use case was as concrete as they come: the Republic of Georgia used it to register land titles, eventually putting more than 100,000 property records on-chain. For anyone tired of hearing that blockchain will "change everything," it was a rare example of the technology quietly doing one thing well.

The company also built Crystal, an analytics tool that does the opposite of what most people assume crypto is for. Instead of hiding transactions, Crystal traces them - mapping the flow of funds across the blockchain for banks, regulators and law enforcement. It is a neat commercial instinct: the same public ledger that promises pseudonymity is also the most complete financial record ever written down, and someone will pay to read it.

Above: three companies that started life inside Bitfury and walked out on their own.

Who actually pays for this

Bitfury's customers were never retail traders. They were governments that needed an auditable registry, mining operators that needed machines and know-how at scale, and financial institutions that needed to know where the money went. The revenue came in several shapes: selling ASIC hardware and BlockBox units; running its own mining farms in places like Georgia, Norway and Iceland; signing long-term equipment-and-services deals with partners such as Hut 8 in Canada and Cipher Mining in the United States; and licensing software through Crystal subscriptions and Exonum contracts. Third-party estimates put annual revenue in the hundreds of millions, though the company - privately held - has never made a habit of confirming its own numbers.

Different by temperament

Bitfury's most obvious rival on the hardware side is Bitmain, the Chinese giant that dominates ASIC manufacturing, along with Canaan and MicroBT. On the analytics side, Crystal competes with Chainalysis, Elliptic and TRM Labs. What sets Bitfury apart is less a single feature and more a posture. Where many mining companies were built to ride one asset up and down, Bitfury behaved like an infrastructure firm that happened to start in crypto - full-stack, government-friendly, patient enough to sit on internal projects until they were ready to become companies of their own.

That patience turned out to be the actual product. LiquidStack, Axelera AI and Cipher Mining were all, in effect, R&D that outgrew the lab. The pattern - build the infrastructure, find the breakout inside it, spin it out - is closer to how a good holding company operates than how a typical miner does.

The second act

In November 2025 Bitfury announced it was leaving Bitcoin mining behind and launching a $1 billion initiative to fund and incubate startups in ethical AI, web3, decentralized systems, quantum computing and self-sovereign identity - with roughly $200 million earmarked for the first year. The framing shifted from "we build the infrastructure" to "we fund the people who do." Vavilov's stated reasoning was blunt: AI is taking over, and he sees a synergy between AI and the decentralized systems Bitfury spent a decade building.

It is a clean piece of self-awareness. The economics of Bitcoin mining have been squeezed by rising difficulty and consolidation; the money and the attention have moved toward AI compute. Rather than defend a business past its peak, Bitfury is doing to itself what it did to LiquidStack and Cipher - taking the useful part and letting it become the next thing.

Where it sits now

For most of its life, Bitfury occupied the layer of crypto that does not trend - the chips, the cooling, the registries, the compliance tools. That is also the layer that tends to survive when the price crashes and the noise dies down. The company's next chapter trades the certainty of a single business for the wider, messier bet of an incubator. Whether backing "ethical" AI is a durable thesis or a well-timed rebrand is the open question. What is not in doubt is the track record it is spending: a decade of shipping real hardware, running real data centers, and spinning out companies that outlived the projects that created them.

The two Valerys started with a hunch that the boring part of Bitcoin was the good part. Fourteen years later, they are making the same bet on the next machine.