Before Fred Thiel had to find electricity for a fleet of Bitcoin miners, he had a more conventional employer: a bank in London. He was still in high school. His first technology job was writing software, in the late 1970s, inside the sort of institution that cryptocurrency would eventually give him a great deal to discuss. The family connection was unusually close. His father and stepmother were bankers. Questions about money moving between countries were familiar territory long before he became a public advocate for Bitcoin.
Thiel was born in France to Swedish parents and educated across France, the United States, Britain and Sweden. His stepmother worked as a senior economist at the OECD, helping develop regulatory frameworks for Eastern European countries. In his telling, banking was part of the household conversation. A childhood spread across borders also supplied a practical introduction to the friction involved in crossing them. Money might be represented by numbers, but the institutions moving it had jurisdictions, procedures and opening hours.
That background gives his enthusiasm for Bitcoin a particular accent. He talks about settlement, the process by which obligations between parties are resolved, as readily as he talks about coins. The teenager writing bank software and the executive discussing a decentralized network are interested in a related problem: how information travels, and what people can do once it arrives. The machinery has changed considerably. The old fascination with the plumbing has survived.
A career in things that have to connect
Thiel’s route to MARA included leadership roles at Lantronix, GameSpy and Local Corporation, as well as private equity and advisory work. Lantronix put him in device networking; its stock began trading publicly in August 2000. GameSpy put him in the business of online games. Local Corporation brought search and digital advertising. It is an untidy résumé if the objective is to remain in one industry, and a rather coherent one if the interest is connecting machines, users and markets.
By 2015, he was explaining an industrial sensor problem: a pump operating within its normal temperature range need not send every observation back to a central store. A nearby gateway could assess the information and report when something mattered. The example was modest, almost aggressively unglamorous. It also anticipated the enduring question of where computing should happen. Moving everything everywhere is expensive. Sometimes the useful intelligence belongs beside the machine making the noise.
His consulting business, Thiel Advisors, offered strategy and leadership development. One client testimonial describes his attention to executive teams and the value of time. His professional network also includes YPO, where he has held technology and fintech leadership positions. These are less photogenic activities than a Bitcoin mine, but they help explain his working life: repeated involvement with people deciding how to reorganize a business before the market reorganizes it for them.
“time is the most precious resource”
A saying attributed to Thiel by a client of Thiel Advisors
The paint came off the walls
Cryptocurrency first drew him toward trading infrastructure. Around 2015 and 2016, he explored a European platform intended to exploit differences between markets. Sprocket, which he co-founded, sought to reduce friction across exchanges. He then joined Marathon Patent Group’s board in April 2018. The company was moving from patent licensing into mining digital assets. Merrick Okamoto, his predecessor as CEO, had the unenviable preliminary task of repairing the balance sheet.
The early Canadian mining operation supplied an anecdote no corporate strategy diagram would dare invent. Thiel has recalled that when the machines were switched on, their fans drew loose paint from the building into the equipment and damaged it. The digital economy had encountered a decorating problem. There is something useful about that detail: Bitcoin’s accounting may live on a distributed network, but the computers doing the work remain very much inside buildings.
On April 26, 2021, Thiel became CEO. Okamoto initially remained executive chairman; Thiel added the chairmanship on January 1, 2022. Those dates matter because he inherited a transformation already underway. His task was to operate and enlarge it. The expanding fleet needed somewhere to run, and MARA initially relied heavily on third parties to host machines. Buying hardware was one decision. Depending on somebody else to keep it productive was another.
The attraction of owning the socket
As Thiel tells it, the company eventually wanted more control over the infrastructure beneath its equipment. Hosting arrangements put operational performance partly in other hands. The shift toward owning sites made uptime, power and the surrounding land part of the same management problem. By April 2024, MARA had closed the acquisition of a 200-megawatt mining data center in Garden City, Texas, adjacent to a wind farm. Electricity was becoming something to organize the business around.
The following February, MARA closed a separate Texas wind-farm acquisition with 114 megawatts of nameplate generation and 240 megawatts of interconnection capacity. These are different measurements: a generator’s rated output and its connection to the grid. The plan included giving older mining machines additional useful life with excess wind energy. A machine that looks uneconomic under one electricity bill can look different under another. Obsolescence, in this business, has a utility account.
The argument extends to what comes out of a computer. Heat usually appears as a nuisance to be removed. In Finland, MARA developed district-heating projects that reuse it. Its 2024 annual report described two pilots serving communities with a combined population of approximately 80,000. That figure describes the communities, rather than a count of individual households supplied. Still, the underlying proposition is tangible: sell the computation, then find a customer for its warmth.
A megawatt gets a second audition
Thiel has carried the energy argument into public life. He testified before a House Financial Services subcommittee in February 2025 about banking access for digital-asset businesses. At the Hill Nation Summit that July, his remarks emphasized Bitcoin mining’s relationship with energy infrastructure. He wants policymakers to see mining as a flexible customer for electricity, including power that might otherwise go unused. It is a case he makes as an interested operator, with a company whose costs depend on the outcome.
By February 2026, the range of possible customers had widened. MARA announced a strategic agreement with Starwood Capital Group and Starwood Digital Ventures to develop selected sites for enterprise, hyperscale and AI computing. The announcement set out approximately one gigawatt of expected near-term IT capacity and a pathway beyond 2.5 gigawatts. These were development ambitions. The agreement did not turn every proposed megawatt into an occupied data center on announcement day.
The division of labor is revealing. MARA contributes sites and energy infrastructure; Starwood brings development, tenant sourcing, construction and operating capabilities. A permitted connection and land have value before a new tenant arrives, but somebody still has to secure that tenant and deliver the building. Thiel’s bet gives his earlier operating experience a new setting. The product offered to the customer is increasingly a place where computing can reliably happen.
“predictable access to energy at scale”
Thiel on what MARA’s sites offer customers, February 2026
The arithmetic keeps its own counsel
The latest numbers resist an effortless success story. For the second quarter of 2026, MARA reported energized hashrate of 70.3 exahashes per second, up from 57.4 a year earlier. Revenue nevertheless fell 27 percent to $174.9 million, and the company reported a $611.3 million net loss. More computing capacity can coexist with worse financial results. A miner works within a competitive network and a volatile asset market; its machines cannot negotiate a kinder outcome.
The AI expansion also has conditions attached. In its August update, MARA described lease discussions alongside Starwood and an objective of signing at least one lease before year-end. It was still awaiting regulatory approval for the Long Ridge acquisition. Rights to a Texas site with up to two gigawatts of potential capacity also depended on approvals. The attractive numbers belong beside those verbs: discussing, awaiting, developing. They describe work to finish.
Another piece was already further along. MARA completed its majority acquisition of Exaion in February 2026, adding private-cloud and high-performance computing capabilities. The combination asks Thiel to connect several kinds of expertise: energy assets, physical development, enterprise customers and digital operations. His earlier career makes the range less surprising. It also makes execution the interesting question. Familiarity with several businesses does not remove the obligation to make them function together.
What remains of the schoolboy programmer
There is a continuity running through these apparently different careers. Bank software, connected industrial equipment, trading systems and data centers all force someone to examine the arrangements beneath an ordinary transaction. Thiel tends to approach technology through those arrangements. His public explanations return to capacity, operating costs and where a machine can be put to work. It is a practical vocabulary for an industry that often prefers to discuss destiny.
His aspiration for MARA is now wider than producing Bitcoin. He wants energy and computing infrastructure that can serve different workloads and capture more value from the same underlying assets. The attraction is understandable. A site can outlast a generation of equipment; a power connection can become valuable to a customer who has never wanted to mine a coin. Whether that possibility becomes durable earnings will be decided by construction, contracts and operations.
The early London job and the Canadian paint episode belong in the same story. One is about understanding a system from inside an institution. The other is about discovering that a building can deliver an immediate and expensive objection. Thiel has spent decades moving between such abstractions and particulars. For all the vocabulary surrounding Bitcoin and artificial intelligence, his next chapter still depends on a wonderfully ordinary question: where, exactly, can you plug it in?