Author: MacKenzie Sigalos, CNBC; Translation: Golden Finance xiaozou
Abstract:
Houston-based technology company Lancium and Denver-based Crusoe Energy Systems announced a multibillion-dollar partnership Thursday morning to build a 200-megawatt data center outside of Abilene, Texas, designed to "meet the unique needs of AI companies."
This is the first phase of a larger 1.2-gigawatt expansion.
Ali Fenn, president of Lancium, told CNBC that this will be one of the largest AI data center parks in the world, further proving that the competition to support the development of artificial intelligence is becoming increasingly fierce, while Bitcoin is left behind.
Right in the heart of West Texas, just off Interstate 20, is a town of 125,000 called Abilene. Once a stop on the cross-border cattle roads of the old American West, this small frontier village is now entering the booming artificial intelligence industry.
On Thursday morning, Houston-based technology company Lancium and Denver-based Crusoe Energy Systems announced a multibillion-dollar partnership to build a 200-megawatt data center outside Abilene designed to “meet the unique needs of AI companies” — for example, providing advanced cloud computing services for applications such as medical research and aircraft design. It’s the first phase of a much larger 1.2-gigawatt expansion.
Ali Fenn, president of Lancium, told CNBC that at full capacity, this would be one of the largest AI data center campuses in the world, further proving that the competition to support the development of artificial intelligence is becoming increasingly fierce, while Bitcoin is being left behind.
“The data center space is rapidly evolving to support modern AI workloads, which require new levels of high-density rack space, direct-to-chip liquid cooling, and unprecedented total energy demands,” said Chase Lochmiller, co-founder and CEO of Crusoe.
There are a lot of synergies between Bitcoin mining and the AI infrastructure business.
Mining companies have massive data centers with access to fiber-optic lines and large amounts of electricity across the U.S. They’re exactly the type of facilities needed for computationally intensive AI operations, which means their sites and technology are in high demand.
At the same time, mining companies need to diversify. In April, the bitcoin price halved, an event that occurs approximately every four years, after which the business of generating new tokens has become less profitable. “Some operators are feeling the financial pressure of the recent block reward halving, which cut industry revenues in half, and are actively exploring exit strategies,” JPMorgan Chase analysts wrote in a June report.
The booming AI industry requires more production capacity, and at the same time, Bitcoin miners are looking for new ways to generate returns on their huge investments, which in turn promotes various mergers, financings and cooperation.
Bitcoin miners turn to artificial intelligence
Lancium and Crusoe are among a host of miners looking to switch from Bitcoin to AI, and so far the strategy appears to be working.
A research report released by JPMorgan Chase on June 17 showed that on June 15, the total market value of 14 major Bitcoin miners listed in the United States reached a record high of US$22.8 billion, an increase of US$4.4 billion in just two weeks.
It is estimated that about 27% of Bitcoin miner Bit Digital's revenue currently comes from artificial intelligence. The company said in June that it had reached a cooperation agreement with a customer to provide Nvidia GPUs to a data center in Iceland for three years, and the cooperation is expected to generate $92 million in revenue per year. The cost of general processing units will be paid by liquidating some crypto assets.
Miami-based Hut 8 said it raised $150 million in debt from private equity firm Coatue to help build out its AI data center portfolio.
Hut 8 CEO Asher Genoot recently told CNBC that his company has “established commercial agreements in several emerging AI verticals under the GPU-as-a-service model, including a customer agreement that provides fixed-fee infrastructure and revenue sharing.”
The shift toward AI has gone particularly smoothly for Core Scientific, which emerged from bankruptcy in January.
B. Riley on Tuesday upgraded its rating on the stock to buy from neutral and raised its price target on the stock to $13 from 50 cents, citing the company’s recent series of partnerships with CoreWeave, an Nvidia-backed startup that is one of Nvidia’s main suppliers of technology for running AI models.
Last month, CoreWeave offered to buy Core Scientific for $1.02 billion, shortly after the two companies announced an expansion of their existing collaboration. Core Scientific, which is now valued at about $2 billion, rejected the offer.
Strengthening the power grid
Over the years, Crusoe has become almost synonymous with the Bitcoin mining industry.
Crusoe's technology helps oil companies turn wasted energy into useful resources. With the help of Crusoe, many Bitcoin miners have installed machines near these energy sources in order to take advantage of this cheaper energy. For example, starting in 2021, Exxon Mobil began working with Crusoe to conduct Bitcoin mining operations in North Dakota.
But Crusoe’s Lochmiller told CNBC that AI infrastructure has actually been part of the company’s vision since its founding six years ago.
“We are reimagining AI infrastructure from the ground up – from our energy solutions, to the design, engineering and construction of our purpose-built AI data centers, to our manufacturing capabilities for critical electrical data center infrastructure in partnership with Crusoe Industries, and ultimately to our purpose-built AI compute stack,” he said.
The AI data center in Abilene is expected to be operational in 2025 and is also planned to be powered primarily by renewable energy.
“Our power orchestration technology is designed to ensure that large AI data center campuses become an asset to the grid, not a liability,” Lancium’s Fenn told CNBC.
Lancium's patented technology translates energy buyers' needs into a dashboard that can gradually adjust power up or down in as little as five seconds. This helps balance the grid's own volatile energy sources, such as wind and solar.
“The original vision of Lancium was to bring large-scale loads to where there was the best and most abundant renewable energy to facilitate the energy transition,” Fenn said.
Back in 2018, Fenn said that the only workload suitable for this scenario was Bitcoin mining.
One of the best features of Bitcoin is that it is completely location-independent. Miners only need power and an internet connection, unlike other industries that require proximity to their end users.
In some cases, the built-in benefits of cryptocurrency mining provide a large enough economic incentive to make it worthwhile to build the infrastructure needed to harness untapped power resources — especially in Texas, which is known as a mecca for renewable energy sources such as wind and solar.
Bitcoin miners are also flexible consumers of electricity — essentially, they act like buyers who will use as much power as they can at any time of day and are happy to accept a power outage with just a few seconds’ notice.
But since then, Lancium's strategy has shifted to artificial intelligence.
“Traditional data centers were — and still are — optimized primarily for proximity to urban areas and users,” Fenn said. “That’s all changed now, and AI data centers are optimized for energy availability, cost, and green cleanliness at scale. Our vision, campus, and technology are perfectly positioned to address this larger, more expansive opportunity.”
Needham analysts estimate that the power capacity of large publicly traded bitcoin miners is expected to more than double over the next one to two years, including expansion plans for mining and high-performance computing operations.
The Electric Power Research Institute estimates that data centers will account for 9% of the nation’s electricity use by 2030, up from about 4% in 2023. Many see nuclear power as a solution to meeting this demand.
TeraWulf uses nuclear energy to power its mining farms and is looking to move into machine learning. So far, the company has 2 megawatts of dedicated high-performance computing capacity, although it plans to shift its energy infrastructure toward AI and HPC.
OpenAI CEO Sam Altman told CNBC last year that he is "very confident" that nuclear power can meet the needs of AI workloads.
“I don’t think there’s a way without nuclear,” Altman said. “I mean, maybe we can do it with solar and storage. But from my personal perspective, I feel like nuclear is the most likely and best way to do it.”
