Hut 8 CEO Asher Genoot recently told CNBC his company “finalized commercial agreements for our new AI vertical under a GPU-as-a-service model, including a customer agreement which provides for fixed infrastructure payments plus revenue sharing.”
The pivot to AI has been going especially well for Core Scientific, which emerged from bankruptcy in January.
On Tuesday, B. Riley upgraded its stock to buy from neutral and raised its price target on the shares to $13 from 50 cents, citing the company’s recent spate of deals with CoreWeave, an Nvidia-backed startup that’s one of the main providers of the chipmaker’s technology for running AI models.
Last month, CoreWeave offered to buy Core Scientific for $1.02 billion, not long after the pair announced an expansion of their existing partnership. Core Scientific rejected the bid. The company is currently worth about $2 billion.
Beefing up the grid
For years, Crusoe’s work has been virtually synonymous with the bitcoin mining industry.
Crusoe’s technology helps oil companies to turn wasted energy, or flare gas, into a useful resource. Many bitcoin miners, with the help of Crusoe, have set up machines adjacent to these sites in order to capitalize on this cheaper power source. Starting in 2021, for example, Exxon Mobil began working with Crusoe to mine bitcoin in North Dakota.
But Crusoe’s Lochmiller told CNBC that AI infrastructure has actually been part of the vision since the company’s founding six years ago.
“We’re reimagining AI infrastructure from the ground up — from our energy solutions, to the design, engineering and building of our specially designed AI data centers, to our manufacturing capabilities with Crusoe Industries for key electrical data center infrastructure and ultimately to our purpose-built AI computing stack,” he said.
The facility in Abilene, which is expected to go live in 2025, is also planning to draw primarily from renewable energy sources.
“Our power orchestration technology is positioned to ensure that mega-scale AI data center campuses can be assets to the grid, not liabilities,” Lancium’s Fenn told CNBC.
Lancium has patented technology that allows it to turn the demand of energy buyers into a sort of dial that can be incrementally turned up or down in as little as five seconds. This helps to balance out a power grid that has inherently volatile energy sources like wind and solar power.
“Lancium’s original vision was to bring large-scale loads to locations with the best, abundant renewable energy in order to facilitate the energy transition,” said Fenn.
Back in 2018, Fenn says that the only load that was a good fit for this was bitcoin mining.
One of bitcoin’s greatest features is that it is totally location agnostic. Miners only require a power source and an internet connection, unlike other industries that must be relatively close to their end users.
In some cases, the built-in proceeds from minting cryptocurrency have offered enough of a financial incentive to make it worth it to build the infrastructure necessary to harness previously untapped sources of power — especially in Texas, which is known for being a mecca for renewable energy sources like wind and solar.
Bitcoin miners are also flexible electricity consumers — essentially, they function as buyers who will take as much power as they’re given, whatever the time of day, and are just as willing to power down with a few seconds’ notice.
But Lancium’s strategy has since shifted to AI.
“Traditional data centers were — and still are — optimized principally for proximity to urban areas and users,” said Fenn. “That has all changed now, with AI data centers optimized for massive scale energy availability, cost and greenness. Our vision, campuses and technology are perfectly positioned for this significantly larger, expanded opportunity.”
In the next one to two years, Needham analysts estimate that large publicly traded bitcoin miners are expected to more than double power capacity, including both their mining and HPC business expansion plans.
The Electric Power Research Institute estimates that data centers could take up to 9% of the country’s total electricity consumption by 2030, up from around 4% in 2023. Tapping into nuclear energy is seen by many as the answer to meeting that demand.
TeraWulf powers its mining sites with nuclear energy, and is looking to get into machine learning. So far, the firm has 2 megawatts dedicated to high-performance computing capacity, though it has plans to transition its energy infrastructure toward AI and HPC.
OpenAI CEO Sam Altman told CNBC last year that he’s a big believer in nuclear when it comes to serving the needs of AI workloads.
“I don’t see a way for us to get there without nuclear,” Altman said. “I mean, maybe we could get there just with solar and storage. But from my vantage point, I feel like this is the most likely and the best way to get there.”
