TeraWulf AI Infrastructure Expansion, October 2026: 1 GW Muskie Campus and the Mining-to-Neocloud Pivot
TeraWulf closed at $12.60 on October 14, up 9.2%, with no published report identifying a specific catalyst; the company's October 5 SEC 8-K confirmed an amended Kentucky Power agreement doubling the Muskie campus to 1 GW of contracted capacity, with a second 500 MW phase targeted for 2029.
- 최근 회계연도 설비투자
- $1.06B (FY2025)
- 전년 대비
- 295.7% · $268M → $1.06B
- 투자 / 매출
- 629% (FY2025, $168M)
- 최고 기록 기간
- $523M · 2026-03-31
TeraWulf's shares closed at $12.60 on the October 14 session, up 9.2% on the day — a move for which no published report has identified a specific catalyst. The gain arrives against a dense backdrop of corporate developments that have accumulated through October, including an amended power agreement confirmed by SEC filing, a new analyst coverage initiation, and multiple reports describing an accelerating AI-compute commercial pipeline.
The most substantive recent development is the expansion of the Muskie AI data center campus in Kentucky. In an 8-K filed on October 5, 2026, TeraWulf disclosed that its wholly-owned subsidiary, DB Long Hunts Data LLC, had executed amended and restated transmission infrastructure and energy service agreements with Kentucky Power Company, an American Electric Power company. Multiple reports describe the amendment as doubling the campus's contracted power capacity to 1 gigawatt, with a second 500-megawatt phase now targeted for 2029 — reportedly one year ahead of the prior schedule — supported by a new generation unit that Kentucky Power is adding at its Big Sandy plant. The immediate market reaction was cautious: several reports recorded WULF shares declining between 5% and 11% in the wake of the announcement, with at least one headline explicitly questioning whether the investment thesis had changed. The October 14 session's 9.2% gain, for which no published report has identified a direct cause, follows a month in which that same news generated a notable negative price response.
That expansion is one chapter in a wholesale strategic redefinition. TeraWulf began as a bitcoin mining operator, anchored by the Lake Mariner facility in upstate New York; it has since repositioned as an AI-compute infrastructure provider. Reports from September 2026 confirmed that HPC and AI revenues had formally overtaken bitcoin mining in the company's income mix. According to multiple reports, TeraWulf has signed a 20-year capacity lease with Anthropic, with the company's chief executive quoted as describing that relationship as "the tip of the iceberg" for its broader AI infrastructure ambitions. One report from Pluang puts the aggregate value of long-term contracts supporting the business at $27 billion; as a single-source figure, that claim warrants caution, but it is directionally consistent with the scale of capacity that has been publicly announced. The company has additionally disclosed a 3-gigawatt AI infrastructure target and is advancing a Hancock, Kentucky project that local projections associate with 300 jobs and $536.8 million in cumulative local tax impact.
The capital demands of this ambition are visible in SEC-derived data. Slicast's compilation of XBRL filings shows TeraWulf's capital expenditure reached $1.06 billion in fiscal year 2025, a 295.7% increase from $268 million in fiscal year 2024, against revenue of $168 million in that same year — a capex intensity of 629% of revenue. Among the seven neoclouds tracked by Slicast, that intensity ranks second, behind only NBIS; by absolute capex, TeraWulf is fifth, trailing leader CoreWeave. The quarter ended March 31, 2026 was the largest single quarter in Slicast's dataset at $523 million. Capital allocation at this scale, relative to current revenue, is characteristic of infrastructure platforms deploying ahead of contracted demand, but it concentrates execution risk significantly.
Equity coverage has broadly affirmed the strategy. TD Cowen initiated coverage in October 2026 with a Buy rating and a $21 price target. Citi, per reports, sees approximately 40% additional upside even after a roughly 500% price gain over the prior twelve months. Morgan Stanley and B. Riley have also expressed constructive views on TeraWulf and select peers. Yet risk factors remain material. Community opposition is an active constraint: the Town of Lansing, New York has moved to draft a data center ban following local public opposition to a TeraWulf facility, and the company has stated it will pursue legal action if the ordinance advances — an outcome whose precedential impact on New York State permitting could extend well beyond a single site. A reported fire at the Lake Mariner facility, described in a single published report, raised operational safety questions that have not been resolved at the level of public SEC disclosures reviewed by Slicast. And in August 2026, CEO Paul Prager sold 137,500 shares for approximately $2.3 million — a disclosed transaction that, while individually unremarkable, sits alongside bullish public messaging about the company's AI trajectory.
Three near-term developments will test whether the current valuation is grounded. The Q3 FY2026 earnings call, scheduled for November 6, is the first opportunity for management to quantify the HPC-versus-mining revenue split, detail the funding structure for the Muskie second phase, and address operational matters at Lake Mariner with specificity. Second, the regulatory environment in New York State — spanning the Lansing moratorium effort and the pending Responsible Data Center Development Act — will define what capacity TeraWulf can realistically deploy east of its Kentucky footprint. Third, actual delivery against the Muskie 1-gigawatt commitment, with the second 500-megawatt phase targeting 2029, is the central execution metric: for every neocloud, the gap between contractual capacity and operational, revenue-generating megawatts is where investment cases succeed or fail, and for a company carrying a capex intensity that ranks among the highest in its peer group, the tolerance for delivery slippage is narrow.