Friday, September 11, 2026
AI 인프라 · 뉴스 & 분석
논평

Core Scientific Texas Grid Approval, September 2026

ERCOT has conditionally approved 1,402 MW of compute load across Core Scientific's Texas sites, the latest regulatory milestone in a buildout anchored by a $14 billion AMD contract and backed by more than $1.7 billion in new financing closed since late August.

설비투자 · SEC 공시 기준전체 이력 →
최근 회계연도 설비투자
$729M (FY2025)
전년 대비
667.7% · $95M → $729M
투자 / 매출
229% (FY2025, $319M)
최고 기록 기간
$389M · 2026-03-31

On September 11, 2026, ERCOT granted Core Scientific conditional approval for up to 1,402 megawatts of compute load across multiple Texas sites — a grid-capacity milestone that removes one of the most significant near-term constraints on a company that has assembled, in roughly eighteen months, one of the largest signed AI data center pipelines in North America. The approvals are conditional, meaning Core Scientific must still satisfy unspecified technical and commercial requirements before full interconnection is confirmed. But the scale itself is a signal: Texas grid operators have accepted the company's buildout projections as credible, and the clearance provides the regulatory foothold on which detailed site engineering can begin.

The significance of that clearance is inseparable from the contract backlog Core Scientific has built since pivoting away from Bitcoin mining. The most recent anchor is the AMD agreement: 529 megawatts, 15 years, up to $14 billion, spanning five campuses, confirmed in late August 2026, with AMD receiving warrants for up to 30 million Core Scientific shares as part of the transaction. That deal lifted total signed AI capacity to approximately 1.1 GW, with aggregate potential revenue exceeding $24 billion across the full portfolio, according to disclosures cited across multiple sources. Earlier in 2026, a CoreWeave arrangement saw 200 megawatts delivered ahead of schedule — an execution proof point that strengthened the company's credibility with the hyperscale customers who followed.

Core Scientific has simultaneously been building scale outside Texas. In August 2026 it completed the $444 million acquisition of Polaris DS, securing approximately 440 megawatts of additional capacity and advancing plans for a Muskogee, Oklahoma campus targeted at 1.5 gigawatts of gross power. A concurrent Oklahoma data center initiative, described by multiple sources as reaching gigawatt scale, drove the stock to what several outlets characterized as a record high in early September. The geographic diversification is deliberate: ERCOT's historically congested interconnection queue makes a second major state footprint a hedge against grid delays, not merely incremental acreage.

The buildout is expensive, and the capital structure reflects it. Slicast's compilation of SEC XBRL filings shows capital expenditure of $729 million in fiscal year 2025 — a 667.7% increase from $95 million in fiscal year 2024 — and a capex-to-revenue ratio of 229% against $319 million of FY2025 revenue, placing Core Scientific fifth among the seven neoclouds Slicast tracks by intensity. The single quarter ended March 31, 2026 recorded $389 million in capex, the highest in the company's filing history. To fund this pace, Core Scientific has assembled a multi-tranche capital stack: a $1 billion facility arranged through Morgan Stanley, a $600 million senior secured credit facility closed in late August, and a $125 million note offering priced in early September. Market reception to the new debt has been mixed — the stock rallied sharply on the initial credit announcement, then fell 5.6% as investors weighed the incremental leverage, before recovering on subsequent contract news. Director Elizabeth Crain's open-market purchase of 6,000 shares on September 4 was cited by analysts as a signal of internal conviction.

The macro backdrop adds complexity. Rising U.S. Treasury yields have pushed up financing costs across the digital infrastructure sector — a dynamic reported in early September to be weighing on Applied Digital, Cipher Mining, and Core Scientific alike. Core Scientific's capex intensity means the company is structurally dependent on long-duration contracted cash flows to service its debt, and any slippage in AMD deployment or ERCOT condition clearance could widen the gap between spending and revenue. The Texas expansion also drew investor concern about regulatory scrutiny in August — a risk that had not materialized into a specific constraint as of early September but has not been formally resolved.

Core Scientific enters the fourth quarter of 2026 with a formidable contractual foundation: 1.1 GW of signed AI capacity, a 15-year AMD anchor, more than $1.7 billion in newly committed financing, and now conditional ERCOT approval for 1.4 GW that, if converted to firm interconnection agreements, underwrites a multi-year construction and revenue cycle. Three signals will determine whether that foundation translates into durable returns. First, the pace at which ERCOT converts conditional approvals to firm interconnection rights — the conditions attached to the 1,402-megawatt status remain unspecified in public disclosures and could carry meaningful timelines. Second, the deployment cadence and first revenue recognition from the AMD contract across five campuses. Third, the trajectory of free cash flow as Polaris DS integration costs roll through the income statement and debt service scales up alongside the expanding capital stack. The bull case is legible; the execution risk at this pace of spending is equally real.

Based on 117 archived reports · Core Scientific
Core Scientific Texas Grid Approval, September 2026 · Slicast