Via corescientific.com
The former Bitcoin miner's pivot to AI infrastructure is paying off faster than expected, with nearly half a gigawatt now billing customers
Core Scientific is billing customers for nearly 440 megawatts of AI computing capacity, beating its own delivery timeline. For a company that emerged from bankruptcy in early 2024 as a Bitcoin miner, that’s a remarkable second act.
The NASDAQ-listed company (ticker: CORZ) has been steadily transforming itself from a crypto mining operation into a high-density colocation provider for artificial intelligence and high-performance computing workloads. Getting 440 MW online and generating revenue ahead of schedule suggests the operational side of that transformation is working.
From mining rigs to GPU racks #
Core Scientific’s pivot wasn’t subtle. The company recognized that its existing power infrastructure, originally built to run thousands of Bitcoin mining machines, could be repurposed for something potentially more lucrative: housing the GPU clusters that AI companies desperately need.
The math is straightforward. Bitcoin miners need cheap power and cooling. AI data centers need cheap power and cooling. The real estate and electrical infrastructure overlap significantly, which gives former miners a head start over building from scratch.
What makes the 440 MW milestone notable isn’t just the number itself. It’s the “ahead of schedule” part. Data center construction timelines have become one of the industry’s biggest bottlenecks, with projects routinely slipping by months or even years. Delivering early is the exception, not the rule.
The company has a track record of beating deadlines. In 2024, it delivered 16 MW of capacity for CoreWeave more than a month ahead of schedule. That kind of reliability matters when your clients are AI companies racing to train and deploy models before competitors do.
Scaling up through acquisition and construction #
Core Scientific isn’t relying solely on organic growth. The company completed its acquisition of Polaris DS LLC, a former competitor in the Bitcoin mining space, for approximately $444 million around mid-August 2026. That deal added roughly 440 MW of existing grid-connected power capacity to Core Scientific’s portfolio.
So in one move, the company essentially doubled its capacity profile. Buying operational power infrastructure is significantly faster than building it, especially when permitting and grid interconnection can take years.
The bigger ambition centers on Muskogee, Oklahoma. Core Scientific is expanding its campus there to approximately 1.5 gigawatts of gross power capacity, with around 1.0 GW available for leasing to clients. An initial 82.5 MW building is slated for delivery in Q4 2027.
To put 1.5 GW in perspective, that’s roughly the output of a large nuclear power plant. Concentrating that much power capacity at a single campus reflects just how electricity-hungry modern AI workloads have become.
Between its existing operations, the Polaris acquisition, and the Muskogee expansion, Core Scientific now claims more than 1,300 MW of contracted power. Future lease payments from AI clients are projected to exceed $6 billion.
The client roster and competitive landscape #
CoreWeave remains one of Core Scientific’s anchor tenants, with contracts spanning several sites and exceeding hundreds of megawatts. CoreWeave itself went public in 2025 and has been on a spending spree to secure GPU capacity for its cloud computing platform, making it one of the most aggressive buyers of data center space in the market.
Core Scientific has also established a collaboration with AMD for potential capacity of up to 2.5 GW. That figure is enormous, roughly the equivalent of powering a small city, and signals how chipmakers are increasingly getting involved in securing the physical infrastructure their hardware will run in.
The market has responded favorably. When Core Scientific announced the Muskogee campus expansion plans in May 2026, shares jumped approximately 7%. That kind of reaction suggests investors view the company’s AI infrastructure play as credible rather than speculative.
The competitive field is getting crowded, though. Other former Bitcoin miners like Riot Platforms, Hut 8, and Cipher Mining have all explored or executed similar pivots toward AI hosting. Traditional data center operators like Equinix and Digital Realty still command dominant market positions. And hyperscalers like Microsoft and Google continue building their own facilities at a pace that dwarfs everyone else.
Core Scientific’s edge, at least for now, is speed. The company can point to existing power infrastructure that was already built and permitted, a proven ability to deliver on time (or early), and long-term contracts that provide revenue visibility. In a market where demand for AI compute far exceeds supply, being able to turn on capacity quickly is worth more than almost any other competitive advantage.
What to watch next #
The $6 billion in projected future lease payments is an impressive pipeline number, but pipeline and revenue are different things. Execution risk remains real, particularly as the Muskogee campus scales from plans to operational reality.
Integration of the Polaris acquisition will also be worth monitoring. Absorbing $444 million worth of infrastructure while simultaneously building out new capacity requires disciplined capital allocation and operational focus. Companies in growth mode sometimes stumble when they try to do too much at once.
Power availability could become a constraint. Utilities across the US are struggling to keep up with data center demand, and some regions have begun pushing back on new large-scale power requests. Whether Core Scientific can secure reliable power at attractive rates for its full expansion pipeline will be a key variable.
For investors trying to gauge the AI infrastructure buildout, Core Scientific offers an interesting case study. The company went from bankruptcy to Bitcoin mining to billing 440 MW of AI capacity in roughly two years. Whether that trajectory continues upward depends on factors largely outside its control: how long AI spending remains elevated, whether hyperscalers keep outsourcing capacity needs, and whether the power grid can keep up with an industry that seems to have an insatiable appetite for electricity. Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our