Digi Power X's AI Pivot Shows Up In Revenue, But Thin Core Profit, Cash Burn And Financing Dependence Define The Setup

Digi Power X reported third-quarter earnings of $0.01 per share on revenue of $12.25 million. Earnings rose 133.3% year over year and revenue grew 50.3%. No consensus estimate was available, so the quarter cannot be scored as a beat or a miss against the Street. The more useful comparison is sequential. Revenue jumped 85% from $6.63 million in the second quarter, and net income of $1.0 million reversed a $14.36 million loss in the prior quarter. That is the central tension. The AI transformation is now visible in reported revenue, but the profit behind the headline came mostly from mark-to-market gains rather than operations. The near-term story also depends on financing that has not yet closed.

The quality of the earnings deserves scrutiny. Management's own reconciliation shows how much non-operating help the quarter received:

• a $7.1 million digital currency revaluation gain;
• a $1.5 million warrant fair-value gain.

Strip those out, along with share-based compensation and other items, and Adjusted EBITDA was just $0.6 million. According to the call, that is down from $3.3 million in the second quarter, even as revenue nearly doubled. On a GAAP basis, the company still posted a gross loss of $2.41 million, although that narrowed from $5.48 million. Gross profit before depreciation and amortization swung to $3.08 million from a $1.46 million loss, which is a genuine operating improvement. Still, it falls well short of the self-funding economics the headline net income implies.

The revenue mix is where the quarter delivered. AI revenue reached $5.66 million, up more than 400% sequentially, and made up 46% of the total, compared with nothing a year ago. Within that:

• Tier III colocation contributed $3.20 million;
• NeoCloudz GPU rental grew 127% to $2.46 million.

Legacy Bitcoin activities fell to roughly 27% of revenue from about 63% a year earlier. Energy sales rose 68% sequentially. However, the 85% sequential growth fell short of the more-than-100% increase management guided on the second-quarter call, so even the strongest number came with an asterisk. Management sounded more confident and execution-focused. It pointed to a 30% sequential cut in G&A and to the launch of the NeoCloudz Token Factory, which sells AI inference directly. Management claims the Token Factory will earn 85% to 90% margins, but that is an assertion, not yet a reported result.

The forward catalysts are substantial and better defined than before. Management has quantified the ten-year take-or-pay Cerebras agreement at Columbiana:

• roughly $402 million for Phase 1 (15 MW);
• about $1.07 billion if Phase 2 (25 MW) proceeds;
• an option that could lift total potential value to about $2.5 billion.

Phase 1 timelines have held across three consecutive calls. Tenant access is set for December 15, with billing beginning in January 2027, and Phase 2 access is targeted for March 2027. Financing has also advanced. A $200 million Goldman Sachs-syndicated facility has progressed from a term sheet to a commitment letter, with closing expected in the fourth quarter.

Management's first formal 2027 outlook is a $250 million to $300 million annualized revenue run-rate by the end of the third quarter of 2027. Only about $143 million of that is supported by current operations and the Cerebras contract, and that figure includes the still-conditional Phase 2. The call also framed the target as roughly in line with, or slightly softer than, the roughly $300 million goal discussed in the first quarter. It is explicitly contingent on financing and market conditions.

Bears have legitimate ammunition. Year-to-date capex climbed to $168 million, and the 2027 plan calls for about $350 million more. Management intends to fund that principally through customer deposits and contract-backed debt. It noted that deposits in new negotiations have risen to 30% to 70%, from about 15% six months ago.

Liquidity reporting needs careful reading because the figures come from different sources and dates:

• The release shows cash, cash equivalents and digital currencies of $119.6 million at quarter end, down from $142.4 million.
• The call cited cash of roughly $150 million in August falling to about $100 million in November.
• Working capital slipped to $103 million from $131 million.

There is also a discrepancy over fixed assets. The release's sequential table lists second-quarter net fixed assets at $73.24 million, while the call cited $127.5 million.

Several other risks are significant:

• Customer concentration in Cerebras is heavy.
• The New York AI conversion now targets the third and fourth quarters of 2027, and a state moratorium blocks power expansion there.
• The earlier plan for about $65 million of 2026 AI revenue and 55 MW deployed by the fourth quarter of 2026 has effectively been abandoned.
• The stake in US Data Centers continues to dilute, falling from 55% to 48%.

Market context is limited. Investor sentiment readings and a current price were not available. Management did acknowledge a sharp share-price drop and broader market turmoil on the call. For reference, the stock opened at $3.92 after the August report, essentially at the current 200-day moving average of $3.91. Shares came into this report under pressure despite the revenue progress.

The bottom line is that Digi Power X delivered visible proof of its AI pivot, with revenue up 85% sequentially and AI now nearly half the business. The earnings, however, were manufactured largely by valuation gains, core EBITDA thinned sharply, and cash is being consumed quickly. The investment case now hinges on three execution milestones over the next two quarters:

1. closing the $200 million facility;
2. starting Cerebras billing on schedule in January;
3. showing that the Token Factory's promised margins appear in reported results.

This earnings review was originally published on Earnings Whispers.

This article is for informational purposes only and is not investment advice. Earnings Whispers and its staff may hold positions in the securities mentioned. It is AI-assisted, based on proprietary Earnings Whispers data and company reports, and edited by Earnings Whispers staff.