WhiteFiber reported $28.8 million in revenue, up 54% year over year, but roughly $12.3 million of that came from a one-time payment tied to a previously disclosed customer termination. Strip it out and revenue was down about 11% year over year and about 25% versus Q1, with Cloud Services down roughly 31% either way. The clear positive is NC-1: colocation revenue nearly tripled year over year, though it was flat sequentially, and the company signed more than $540 million in new multi-year AI contracts since May. Net loss widened to $15.0 million from $8.8 million, driven mainly by new interest expense, and cash and restricted cash fell $19.7 million from Q1. NC-1's next phase still depends on financing that has not closed. The quarter proves the contract pipeline is real. It does not yet prove the business can fund and operate it profitably.
A one-time payment flattered the headline beat. Cloud Services revenue of $23.8 million included approximately $12.3 million tied to a previously disclosed customer termination, a settlement-style payment rather than repeat business. Section 3 breaks down what that means for growth and margins.
NC-1 reached initial billing, but the step-up is still a Q3 test. Colocation revenue grew 173% year over year to $4.7 million as the site moved from construction into customer billing, but sequentially it was essentially flat versus Q1's $4.8 million. CEO Sam Tabar said NC-1 expects full contracted run-rate billing across its 40 megawatts later this month.
The contract pipeline strengthened, and losses widened, for a specific reason. WhiteFiber signed more than $540 million in new multi-year AI contracts since its May call, including its first Nvidia Vera Rubin deployment. Net loss nonetheless grew to $15.0 million from $8.8 million, driven mainly by new interest expense below the operating line, detailed in Section 5.
WhiteFiber does not report an "organic" or adjusted revenue figure, so the numbers below are a simple analytical subtraction, reported revenue less the disclosed termination-related amount, not a company-defined metric.
Total revenue excluding the $12.3 million termination payment would have been approximately $16.5 million: about 11% below Q2 2025's $18.7 million, and about 25% below Q1 2026's $21.9 million. Cloud Services revenue on the same basis would have been approximately $11.5 million, down roughly 31% both year over year and sequentially. Reported Cloud Services growth of 43% becomes a meaningful decline once the payment is removed.
The $12.3 million payment likely made a material contribution to Q2's $5.5 million adjusted EBITDA, but WhiteFiber did not disclose the associated costs or the exact EBITDA contribution, so a precise recurring figure cannot be calculated from the release. The six-month trend is more telling: H1 2026 adjusted EBITDA was $8.5 million, down about 8% from $9.3 million in H1 2025, despite the termination payment landing inside this year's total. Q2's year-over-year improvement looks weaker once read alongside a declining first half.
The $932.9 million colocation RPO gives long-dated revenue visibility, but it is not a diversified backlog. It is tied primarily to one facility and one agreement: a 10-year, 40-megawatt colocation deal with Nscale Global Holdings worth approximately $865 million, signed in December 2025. WhiteFiber's direct counterparty is Nscale, not the hyperscaler itself; Nscale separately holds an offtake agreement with an unnamed, investment-grade hyperscale customer that strengthens the credit profile behind the contract, but WhiteFiber has no direct agreement with that end customer. Successful NC-1 commissioning and Nscale's own performance remain central to whether this RPO converts as expected.
These values are spread over three- to five-year terms; they strengthen long-term visibility but should not be read as revenue available next quarter. The timeline below shows when each deployment is targeted to start billing, not the full multi-year term:
WhiteFiber also demonstrated 111.2 terabits per second of cross-data-center throughput at roughly 0.9 millisecond latency, technology it plans to launch commercially in Q3 2026, though it is not yet tied to a signed customer or disclosed revenue.
Operating loss was nearly flat year over year, $9.3 million versus $9.2 million: the $5.0 million software impairment pulled results down, largely offset by the termination payment sitting in revenue. The wider net loss came almost entirely from below the operating line, where Q2 interest expense totaled roughly $6.0 million ($4.6 million third-party, $1.4 million related-party), against no comparable interest expense a year ago.
Cash and restricted cash fell to $60.4 million from a combined $80.1 million at Q1 2026 ($75.8 million cash plus $4.3 million restricted), a decline of about $19.7 million, or 25%, in one quarter. The Q2 release does not split cash from restricted cash, so the freely deployable portion should be confirmed once the Q2 Form 10-Q is available.
Two financing sources sit behind the buildout. Post-quarter, WhiteFiber expanded an RBC credit facility to up to CAD $115 million, plus a possible CAD $25 million accordion, with CAD $36.8 million drawn by mid-July. In May, WhiteFiber also signed a $100 million delayed-draw term loan, expandable to $150 million, with Bit Digital Capital, a subsidiary of Bit Digital, Inc. WhiteFiber's CEO, Sam Tabar, and CFO, Erke Huang, hold the same roles at Bit Digital, making this a related-party facility, meant partly to bridge the period before permanent NC-1 project financing closes.
WhiteFiber is separately in exclusivity with a different lender group for a proposed secured financing tied to NC-1. Management said this could return a significant share of invested capital to the balance sheet for reuse in future projects, while leaving the terms, leverage, and any future equity requirements unresolved. The deal is not guaranteed to close, or to close on favorable terms.
Weighted-average diluted shares rose 43% to 38.7 million from 27.0 million a year ago, softening the per-share impact of the wider net loss; $3.7 million of Q2 costs was share-based compensation.
WhiteFiber does not publish formal revenue guidance. The company-stated marker is full contracted run-rate billing across NC-1's 40 megawatts by late August 2026. Colocation revenue was $4.7 million in Q2 on partial billing and essentially flat sequentially, so Q3 is the first quarter that should show whether full billing produces the step-up the story requires. WhiteFiber has not disclosed a dollar run-rate per megawatt, so that scale cannot be independently verified in advance.
This quarter proved NC-1 can bill: it moved from construction to active customer revenue, and WhiteFiber can still win large, multi-year AI infrastructure contracts against real competition for GPU capacity.
It did not prove the business can grow without a one-time payment propping up the headline number. Recurring revenue contracted from both comparison periods, the wider net loss traces mainly to new interest expense, cash fell by a quarter in three months, and the financing NC-1's next phase depends on remains unsigned.
Can WhiteFiber turn NC-1 and its multi-year contract pipeline into recurring revenue fast enough to offset weaker Cloud Services revenue, rising interest costs, and continued dependence on external financing?