WhiteFiber (WYFI) Q4 2025 earnings review

Massive Anchor Deal Masks Ballooning Overhead Costs

WhiteFiber wrapped up a highly consequential 2025 with revenue accelerating 61% year-over-year to $23.6M in Q4. The company successfully executed its infrastructure pivot by securing an $865M anchor contract with Nscale for its NC-1 campus and bringing the MTL-3 facility online. While gross margins expanded impressively to 61%, the underlying corporate structure is highly capital intensive. A widening operating loss and a massive $230M post-quarter convertible debt raise highlight the heavy price of funding AI infrastructure at scale.

๐Ÿ‚ Bull Case

Nscale Contract Validates Development Pipeline

The 10-year, $865M agreement for 40MW at NC-1 fundamentally de-risks the facility. At an implied run-rate of ~$86.5M annually, this single contract exceeds the company's entire FY25 total revenue of $79.1M.

Margin Expansion is Real

Gross margins (excluding D&A) improved dramatically from 52% in 24Q4 to 61% in 25Q4. Scale efficiencies are beginning to materialize as facilities like MTL-3 ramp up utilization.

๐Ÿป Bear Case

Overhead Outpacing Top-Line Growth

Despite a massive 61% YoY jump in revenue, operating losses worsened from $1.8M in 24Q4 to $5.2M in 25Q4. General and administrative expenses skyrocketed nearly 3x YoY, completely swallowing the gross profit gains.

Severe Capital Intensity

FY25 Capital Expenditures hit $268M to generate just $79M in full-year revenue. The company had to issue $230M in convertible notes in January 2026 just to keep funding its pipeline.

โš–๏ธ Verdict: โšช

Neutral/Cautiously Bullish. The Nscale deal proves WhiteFiber can land whales, and the AI demand story remains intact. However, aggressive dilution, spiraling operating expenses, and extreme capex demands mean bottom-line profitability remains a distant hope.

Key Themes

DRIVERNEW๐ŸŸข๐ŸŸข

Colocation Services Breakout

Accelerating. Colocation revenue jumped from $1.69M in 25Q3 to $3.85M in 25Q4. This was driven directly by the MTL-3 data center coming fully online and generating immediate revenue under the Cerebras colocation agreement. With NC-1 scheduled for Q2 2026, Colocation is rapidly catching up to Cloud as the core growth engine.

DRIVER๐ŸŸข

Resilient AI Cloud Demand

Stable. Cloud services revenue grew 48% YoY to $19.3M. Management continues to cite near-term deployments and the massive imbalance between supply and demand for high-performance computing infrastructure as primary tailwinds.

MACRO๐ŸŸข

High-Density AI Infrastructure Bottleneck

Management highlighted that demand for high-density AI infrastructure continues to exceed available supply. The industry transition toward purpose-built, next-generation AI workloads heavily favors specialized operators over legacy data center generalists, allowing WYFI to secure premium 10-year lock-ups.

CONCERN๐Ÿ”ด

Operating Expense Bloat Contradicts Margin Story

Reversing. While the company boasts about gross margin improving to 61%, a deeper look at the income statement tells a darker story. Gross profit grew by ~$6.7M YoY in Q4, but combined Depreciation and G&A expenses grew by ~$10M over the same period. The business is currently displaying negative operating leverage as public company costs and overhead overwhelm top-line beats.

CONCERNNEW๐Ÿ”ด

Insatiable Capital Requirements

Accelerating. WhiteFiber burned through $268M in CapEx in FY25. With only $114M in cash exiting the year, the company was forced to execute a $230M private placement of 4.5% convertible notes in January 2026. While necessary to fund NC-1 and future sites, the constant need for external capital caps equity upside and threatens future dilution.

CONCERNNEWโšช

Single-Tenant Concentration Risk at NC-1

The $865M Nscale agreement is transformative, but it also concentrates a massive amount of counterparty risk into a single anchor tenant for 40MW of capacity. Any execution or financial issues on Nscale's end over the 10-year term could jeopardize the economics of the entire North Carolina campus.

Other KPIs

Adjusted EBITDA (25Q4)$5.8 million

Stable. Adjusted EBITDA came in at $5.8M, representing a slight increase from $5.5M in 24Q4. Importantly, this shows a significant sequential recovery from the heavily depressed $2.3M reported in 25Q3, proving that unit economics are stabilizing post-IPO.

Net Loss (25Q4)-$1.5 million

Accelerating recovery. While technically a wider loss than 24Q4 (-$1.0M), it marks a dramatic sequential improvement from the -$15.7M net loss seen in 25Q3, which was heavily skewed by massive one-time stock-based compensation hits related to the IPO.

FY25 Capital Expenditures$268 million

Massive capital outlay primarily tied to the acquisition of the NC-1 development site and the lease-to-own arrangement for MTL-3. This intense upfront spend underscores the barrier to entry in the AI data center space.

Guidance

NC-1 Phase 1 Service CommencementQ2 2026

Accelerating. The company tightened its timeline, stating the initial 24-megawatt phase for Nscale will commence in Q2 2026, advancing from prior Q3 commentary of 'early / first half of 2026'. This locks in the timeline for a major step-function increase in colocation revenues.

Nscale Contract Value$865 million over 10 years

Stable. While not traditional quarter-by-quarter revenue guidance, this firmly anchors future projections, implying an eventual sequential add of roughly $21.6M per quarter once fully ramped. This provides excellent revenue visibility through the end of the decade.

Key Questions

Margin Profile of Nscale Agreement

The $865M headline number for Nscale is impressive, but how does the gross margin profile of this massive colocation deal compare to the 61% margins currently seen in the Cloud business?

G&A Expense Runway

G&A expenses were $11.4M in Q4, down from the IPO-heavy Q3 but still up nearly 3x YoY. What is the normalized quarterly G&A run-rate investors should model for FY26?

Pipeline Expansion Beyond NC-1

With NC-1 secured and $230M in fresh capital, how imminent is the formalization of the next development site, and are power constraints gating these negotiations?