Two years ago, Core Scientific was in bankruptcy. Today it is closing a $1 billion credit facility backed by JPMorgan Chase and Morgan Stanley to build AI data center infrastructure. That trajectory tells you something important about where institutional capital is flowing in 2026.
The company announced it has secured a $1 billion strategic funding package, with JPMorgan contributing $500 million of that total. The credit facility carries an interest rate of SOFR plus 2.5%, giving Core Scientific competitively priced access to capital at scale. The company reports $533 million in available liquidity following the close.
The Pivot From Mining to AI
Core Scientific's story is one of the more dramatic reinventions in the digital infrastructure space. The company filed for Chapter 11 bankruptcy protection and emerged in January 2024. Rather than returning to its original business model centered on Bitcoin mining, it made a deliberate strategic decision to reposition its existing data center sites for high-density AI and high-performance computing colocation.
The logic is straightforward. Bitcoin mining data centers require significant power infrastructure and physical space - the same core assets needed to run AI workloads. The facilities can be repurposed, but the transition requires capital for equipment upgrades, early-stage development costs, and new energy supply agreements. That is exactly what this $1 billion credit line is designed to fund.
The shift is already producing results. Core Scientific has reported notable gains in colocation revenue as it has converted sites, though colocation income has not yet fully offset the decline in overall quarterly revenue that comes from winding down self-mining operations. The gap is expected to close as the conversion accelerates.
Why This Matters Beyond One Company
Core Scientific's raise is part of a broader pattern that has been building throughout 2026. JPMorgan's Security and Resiliency Initiative - the same vehicle co-leading Shield AI's $2 billion defense round and exploring participation in Reflection AI's $2.5 billion raise - is increasingly positioning the bank as a major backer of AI infrastructure across multiple categories.
The willingness of a major financial institution to extend $500 million to a company that was in bankruptcy 24 months ago reflects how strong institutional conviction in AI infrastructure demand has become. Lenders are not just betting on Core Scientific specifically - they are betting that the demand for high-density AI colocation is durable enough to justify the credit risk.
That bet has context. Data center capacity constraints are one of the most cited bottlenecks across the AI industry in 2026. Hyperscalers are building aggressively but cannot keep pace with demand. Colocation providers that can deliver power-dense facilities with reliable energy contracts are capturing premium pricing and long-term contracts from AI companies that need compute capacity now.
The Risks Are Real
The funding comes with genuine challenges worth naming. The debt load is substantial for a company still in the early stages of a business model transition. The SOFR plus 2.5% rate is competitive today but introduces exposure to interest rate movements. And the entire thesis depends on AI infrastructure demand remaining strong enough to fill capacity as Core Scientific converts its sites.
The company is also navigating the operational complexity of running down one business - Bitcoin mining - while simultaneously building another. Mining revenues are declining by design, and colocation revenues need to scale fast enough to fill that gap before the financial math becomes strained.
For business leaders and investors tracking AI infrastructure, Core Scientific's raise is a useful data point. The market for AI data center capacity is attracting capital from every direction - venture firms, private equity, banks, and now companies reinventing themselves from adjacent industries. The question is no longer whether the demand is real. It is whether the supply of infrastructure can be built fast enough to meet it.
