Product1 distinct publisher3 min readUpdated
The AI cloud firm is paying about $1m per megawatt of initial capacity for 120 acres and 500,000 square feet in Yadkin County. It has run this trade with the same seller before.
The Product Desk · Product desk

Compiled by The Product DeskSomething wrong?How this is made
WhiteFiber has agreed to pay $60 million for two industrial properties in Yadkin County, North Carolina, which it plans to retrofit into data center campuses called NC-2 and NC-3 [1]. The seller is textiles manufacturer Unifi, the same company that sold WhiteFiber the site now operating as NC-1, roughly 55 miles away [3][6].
The price works out to about $120 per square foot across roughly 120 acres and 500,000 square feet of warehouse space [3][1]. The two properties are expected to provide a combined minimum of 60MW of initial gross utility capacity, 30MW each [2], putting the entry cost near $1 million per megawatt of initial capacity [2]. WhiteFiber says the pair could support up to approximately 200MW combined over time [4]; if that arrives, the acquisition basis falls to roughly $300,000 per megawatt [2], with about 3.3 times the initial capacity as headroom [3]. Added to the 40MW NC-1 campus in Madison, the state footprint would start at about 100MW and could reach 240MW [4].
The schedule is the part worth studying. The deal runs through Enovum Data Centers, WhiteFiber's wholly owned data center subsidiary, and is expected to close in Q4 2026 [5], with initial ready-for-service capacity targeted for Q3 2027 [2] - roughly three quarters after closing [6]. WhiteFiber has already demonstrated the conversion once with the same counterparty: NC-1 was previously a Unifi manufacturing site and went live over the summer [6]. CEO Sam Tabar framed the purchase as building on capabilities and relationships already established in the region [7].
Unifi's side is a balance sheet transaction. CEO Eddie Ingle said the carve-out makes Unifi "a leaner and more profitable organization," is expected to have minimal operational impact with no downtime, will leave US production capacities unchanged, and on closing would let Unifi retire a substantial amount of debt [8]. That alignment is why these deals keep clearing: an industrial owner with idle square footage and debt, and a buyer that needs a roof and a utility path rather than a factory.
Demand is stated but not yet contracted. WhiteFiber says it is in advanced discussions with prospective customers and has received non-binding letters of intent with investment-grade credit support around the Yadkin sites [9]. Non-binding is the operative word, and it will be nine months of closing risk before anything is signed against steel.
The financials argue for cheap capacity. In Q2 2026 WhiteFiber reported revenue of $28.8 million, up 54 percent year on year, with an operating loss of $9.3 million and a net loss of $15 million [10]. Cloud Services revenue of $23.8 million included approximately $12.3 million tied to a previously disclosed customer termination [11]; excluding that, the underlying figure is around $11.5 million, about 31 percent below the prior-year $16.6 million [5]. Colocation revenue, the business these retrofits feed, was $4.7 million, up 173 percent [11]. The company also signed a partnership with Krambu aimed at adding 100MW of capacity over 2027 and beyond [12].
Watch whether the Q4 2026 close holds [5], whether the letters of intent convert to binding contracts with the credit support described [9], and whether the utility interconnection path actually stretches from 60MW to 200MW [4]. Also watch the colocation line against cloud services stripped of termination revenue [11][5].
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
WhiteFiber said NC-2 and NC-3 could have the potential to support up to approximately 200MW of combined gross utility capacity over time.
WhiteFiber said it is in 'advanced discussions' with prospective customers and has received non-binding letters of intent with investment-grade credit support around the Yadkin sites.
WhiteFiber has entered into a definitive agreement to acquire two industrial properties in Yadkin County, North Carolina, for $60 million, and intends to retrofit them into data center campuses to be known as NC-2 and NC-3.
The two properties are expected to provide a combined minimum of 60MW of initial gross utility capacity (30MW each), with WhiteFiber targeting initial ready-for-service capacity in the third quarter of 2027.
The sites were acquired from textiles manufacturing firm Unifi Manufacturing, Inc. Unifi said the deal includes approximately 120 acres of land and 500,000 square feet (46,450 sqm) of warehouse space across two separate locations in Yadkin County.
The deal was made through WhiteFiber's wholly owned data center subsidiary, Enovum Data Centers Corp, and is expected to close in Q4 2026.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Concrete numbers, one outlet, all of it announcement-derived
The physical and financial specifics are unusually crisp for a pre-close deal: price, acreage, square footage, per-site megawatts, close quarter, ready-for-service quarter, distance to NC-1, and a full quarter of segment revenue. But every figure traces to the two principals' own announcements relayed by a single trade publication; there is no utility, permitting, engineering, lender or independent-analyst corroboration, and the transaction has not closed. That caps evidentiary strength below the midpoint despite the specificity.
Model proven once at NC-1; the new capacity is pre-close and uncontracted
There is genuine, banked adoption behind the strategy: NC-1 is a 40MW retrofit of the same seller's plant that went live over the summer, colocation revenue nearly tripled to $4.7m, and named tenants (Cerebras, Nscale, Modal Labs via Hyperbolic) exist at group level. For NC-2 and NC-3 specifically, adoption is prospective only — a Q4 2026 close, a Q3 2027 first-service target, and non-binding letters of intent with no disclosed counterparty or contracted megawatts. Underlying cloud revenue excluding the termination amount actually fell year over year, so demand momentum is concentrated in the smaller colocation line.
Headline megawatts and growth rates run ahead of what is committed
The overstatement is structural rather than fabricated. A $60m purchase buys 60MW of minimum initial capacity — about $1m per megawatt — yet the framing leans on ~200MW of potential (about $300k per megawatt, roughly 3.3x the committed figure) and, with NC-1, a ~240MW footprint narrative. Demand is non-binding, nothing closes until Q4 2026 and nothing serves until Q3 2027. The 54 percent revenue growth headline is inflated by ~$12.3m of terminated-contract recognition, without which cloud revenue fell about 31 percent. The reporting does relay the qualifiers ('minimum', 'potential', 'non-binding'), which keeps the gap moderate rather than severe.
Both principals gain from the announcement; only their words are on the record
The story is constructed almost entirely from two interested parties' announcements. WhiteFiber is a public AI infrastructure company taken public by Bit Digital that benefits from a scaling narrative and from signalling investment-grade demand while its cloud line weakens. Unifi explicitly states it would retire a substantial amount of debt and improve financial flexibility on closing, so it needs the sale to look clean and non-disruptive. The Krambu partnership announcement in the same month adds another capacity headline. No counterparty, lender, utility or community voice offsets these incentives.
Facts of the announcement are solid; the outcome is not
Confidence is moderate. What was announced — price, asset scope, capacity tranches, target dates, quarterly financials — is reported with enough specificity that it is unlikely to be wrong, and the seller's separate announcement corroborates the asset details. What remains uncertain is everything that matters for outcome: whether the Q4 2026 close happens, whether utility capacity supports 60MW by Q3 2027 or 200MW ever, whether the non-binding LOIs convert, and how the retrofit is financed. A single publisher with no independent verification keeps this near the middle.
product
Kog's pitch: the cheapest inference upgrade is the H200s you already bought1 distinct publisher
invest
Nvidia is brokering the Nordic build-out, not just supplying it2 distinct publishers
invest
One Anthropic order, six times the price: Fractile's $6.5B mark arrives two years before its chips3 distinct publishers
product
Cerebras's CS-4 is three old wafers in a new rack: price the packaging, not the silicon2 distinct publishers
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 17, 2026