WhiteFiber recorded $10.4 million of interest expense in the third quarter against $24.0 million of revenue, according to its Form 10-Q filed on 10 November. By our arithmetic, interest equalled 43% of revenue; a year earlier the line was nil.
The AI cloud and data-centre operator reported a net loss of $101.4 million, against $15.8 million a year earlier. On the same day it disclosed, in an 8-K, a $300.0 million term loan for its NC-1 data centre in North Carolina, arranged by Morgan Stanley Senior Funding.
Business Pill · Tokens per Watt
A short explainer on tokens per watt, a measure of compute efficiency. It teaches the general idea only and does not discuss WhiteFiber.
The key insight: By our arithmetic from the 10-Q, WhiteFiber’s $10.4 million of third-quarter interest expense equalled 43% of its $24.0 million of revenue, and the minimum rate on the new $300.0 million NC-1 loan is 8.00% (a 2.00% SOFR floor plus 6.00%).
Revenue rose to $24.0 million from $20.2 million, up 19% by our arithmetic. Cloud services, the GPU rental business, brought in $18.1 million against $18.0 million a year earlier. Colocation, renting data-centre space and power, rose to $5.6 million from $1.7 million.
Adjusted EBITDA, the company’s non-GAAP measure, was a loss of $24 thousand, against a profit of $2.3 million a year earlier. General and administrative expenses were $18.0 million. The adjusted EBITDA reconciliation adds back $3.9 million of share-based compensation. Depreciation rose to $8.1 million from $6.4 million.
Interest expense was $6.4 million to third parties and $4.0 million to related parties. The related-party lender is a subsidiary of Bit Digital, which the 10-Q says owns about 59.7% of WhiteFiber. That facility bears interest at an initial 9.5% a year.
WhiteFiber’s third quarter of 2026: revenue against general and administrative expense, interest expense, depreciation and adjusted EBITDA (the company’s non-GAAP measure). Source: WhiteFiber Form 10-Q filed 10 November 2026. Interest as a share of revenue (43%) is our arithmetic.
Where the $101.4 Million Loss Came From
Most of the loss is one charge. WhiteFiber booked an $80.8 million “induced conversion expense” in the quarter. In August it settled $198.15 million of its 2031 convertible notes for $118.0 million of cash and 6,271,534 shares with a fair value of $169.8 million.
The charge is what that package was worth above what holders would have received under the original conversion terms, the filing says. By our arithmetic, it is 80% of the quarter’s net loss; without it, the loss would have been about $20.7 million.
In August WhiteFiber also issued $310.0 million of 5.00% convertible notes due 2032. At 30 September, $31.9 million of the 2031 notes and all $310.0 million of the 2032 notes were outstanding. By our arithmetic, debt on the balance sheet (convertible notes plus third-party and related-party loans, net) totalled about $402.6 million, against cash of $138.5 million.
The $300 Million NC-1 Loan and Its Price
The 8-K describes $300.0 million of commitments: a $200.0 million initial term loan and a $100.0 million delayed-draw loan. Morgan Stanley Senior Funding is administrative and collateral agent, lead arranger and lead bookrunner, alongside a consortium of lenders. No loans had been funded as of the report.
Loans bear interest at Term SOFR, with a 2.00% floor, plus 6.00% a year. By our arithmetic, the lowest possible rate is 8.00%. On the full $300 million, that is $24.0 million of interest a year, the size of WhiteFiber’s entire third-quarter revenue.
Initial funding depends on “readiness and customer acceptance of the first 40 megawatts of capacity” at NC-1. The loans mature five years after funding and amortize 2.50% a quarter. The 8-K adds: “The loans are not guaranteed by WhiteFiber, Inc.”
That 40 MW is contracted to Nscale under a service order the 10-Q values at about $865 million over 10 years. The first 20 MW went into service on 23 September, and the filing says the remaining 20 MW is now operational. Events of default include “specified project- and customer-agreement defaults”. Our earlier piece Nscale Puts 73% of Capacity at Owned Sites covers the tenant’s side.
Customers, Backlog and the New Risk Factor
“As of the date of this Form 10-Q, we have five existing customers.” The largest accounted for about 50% of cloudrevenue in the first nine months. Three customer orders were discontinued in the period, a $5.1 million revenue impact, including the company’s Initial Customer.
New orders contracted this year total about $1,447 million over six months to five years, the filing says. Remaining performance obligations were $729.0 million, of which $716.3 million is cloud services. Deferred revenue reached $306.0 million by our arithmetic ($54.6 million current plus $251.4 million non-current).
The 10-Q adds one new risk factor: “Increased scrutiny of data center development and changes in governmental or utility policies could delay or restrict our expansion, limit power availability and increase our costs.” It names NC-1’s planned expansion and the Yadkinville properties as depending in part on additional power.
The Structural Read
As we read the filings, the quarter’s debt cost and the NC-1 loan describe the same build-out from two sides: the 10-Q shows interest already at 43% of revenue, and the 8-K ties the next $300.0 million to the first 40 MW of capacity being ready and accepted by the customer.
WhiteFiber Form 8-K, 10 November 2026
“The loans are not guaranteed by WhiteFiber, Inc.”
Three Implications
FOR THE INCOME STATEMENT Of the $101.4 million net loss, $80.8 million was an induced conversion expense on the August note settlement, 80% by our arithmetic.
FOR THE BALANCE SHEET Debt on the balance sheet totalled about $402.6 million by our arithmetic, against cash of $138.5 million at 30 September.
FOR CUSTOMERS The 10-Q lists five customers; the largest accounted for about 50% of cloudrevenue in the first nine months.
The framework argues: “the capex race is no longer a corporate spending cycle. It is a sovereign-scale financing problem dressed in tech-company clothing”.
As we read it, WhiteFiber’s quarter shows that financing at the scale of a single neocloud: convertible notes, a related-party loan and a project loan secured on one data centre. The framework predates these filings and does not discuss WhiteFiber.
What Is Not Established
The 8-K does not state the size of the upfront and funding premiums, the commitment fee or the covenant levels. Initial funding remains subject to conditions, and the filing says there is no assurance it will occur.
The 10-Q does not name the customer that provides about half of cloudrevenue. We have not seen a statement from WhiteFiber beyond its filings, and we did not contact the company for this piece.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
By our arithmetic, WhiteFiber recorded $10.4 million of interest expense on $24.0 million of third-quarter revenue, and none of the new $300.0 million NC-1 loan had been funded as of the 8-K. That loan is secured on substantially all of the Borrower’s assets, Holdings’ equity in the Borrower and a mortgage on NC-1, and is not guaranteed by WhiteFiber, Inc. For more on how this layer is funded, see How Neoclouds Work, Neoclouds Are Becoming Manufacturers and Anthropic’s $34.9B Lambda Compute Deal Is Now Terminable.
A note on sourcing. Every reported figure comes from WhiteFiber’s Form 10-Q for the quarter ended 30 September 2026 and its 8-K on the NC-1 term loan, both filed with the SEC on 10 November 2026. Figures labelled “by our arithmetic” are our own calculations from those filings. Adjusted EBITDA is the company’s non-GAAP measure. We did not contact WhiteFiber or its lenders for this piece.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
