WhiteFiber (WYFI) is a 14-month-old neocloud/datacenter operator spun out from Bit Digital (BTBT, 70.1% owner). The thesis is a single $865M / 10-year colocation contract with Nscale, anchored at NC-1 (Madison, NC), with Phase 1 (27 MW) targeting Q2 2026 billing commencement. The Q1 2026 10-Q filed May 14 forced two reads: what the filing actually confirms, and what it doesn't.

What the filing says

The $923.7M Remaining Performance Obligations table is the primary data point: $920.9M in colocation services, schedule running $50M in 2026, ≈$90M annually 2027-2030, $503M thereafter. Nscale sent $65.7M in Q1 prepayments and was billed $67.8M in non-recurring construction charges. Cash $75.8M. A $230M convertible note (4.50%, due 2031, conversion $25.91 — deeply out-of-the-money) issued January 2026 added ≈$102M of net runway after a $120M capped call hedge.

NC-1 billing has NOT commenced. Management guides Q2 start, Q3 full revenue. The Duke Energy 40 MW deadline (April 1, 2026) was met silently in the filing — neither confirmed nor denied. Q1 G&A came in at $17.8M (annualizes to $71M, worse than FY2025's $52.5M full year). Cloud revenue +13% YoY despite losing the "Initial Customer" — a new 256 H100 contract at $50.2M / 24 months (deploying April) materially replaces the loss.

What the market thinks

The stock implies ≈30-35% probability of NC-1 on-time execution. Sell-side disagrees: 8 buys / 0 holds / 0 sells, mean target $34.62 (≈80% implied probability). Short interest 17.1%. Sell-side and tape have diverged — uncommon. Insiders bought $687K of personal money across three C-suite officers (CEO $425K, President $205K, CTO $57K) in six weeks during the drawdown.

Range of fair value across three approaches (run-rate × peer multiple, sum-of-parts, market-implied probability decode): $15-37/share. Wide range; the dispersion IS the finding. NC-1 execution determines which end of the band.

Why the gap exists

WYFI was spun out from BTBT in March 2025. Public float is only 11.3M shares; 70.1% sits with BTBT, which is at a 52-week low ($1.47) burning -$205M / 9mo of operating cash. Factor regression shows BTBT explains 50% of WYFI variance (β=0.89, R²=0.55, 156d). Idiosyncratic variance is only 45% — fails the 75% target by 30 percentage points. The stock has been dragged by its parent, not by its own fundamentals.

Cross-sector data adds a meta-finding worth flagging on its own. CRWV Q1 2026 call (May 7): "Average pricing A100s, H100s, H200s, L40S all increased quarter-over-quarter, largely sold out." NBIS Q4 2025: "Sold out capacity once again Q4, high utilization, strong pricing." The consensus narrative that Hopper-generation GPUs commoditize as Blackwell ramps is contradicted by sector data through May 2026. This validates WYFI's $11/hr-equivalent enterprise contract pricing as in-band, not anomalous — Q1 cloud gross margin 59.5% triangulates with bundled enterprise economics, not pure compute (which would carry materially higher margins).

Risks (ranked by impact)

1. NC-1 slip beyond Q2. G&A $71M annualized + $10.4M/yr convert interest beginning August + $75.8M cash = compounding squeeze. A 2-quarter slip compresses runway below 6 quarters.
2. BTBT secondary into 11.3M float. A 5M+ share registration creates 20-40% technical hit independent of fundamentals. No disclosure in the 10-Q, but BTBT's cash position creates pressure.
3. Duke ESA termination. Thesis-break event. Filing silence on the April 1 milestone is the watchable risk.
4. Initial Customer litigation. $30M+ stuck in disputed early-termination negotiation. Legal cost overhang.
5. Hopper pricing reverses. If Blackwell ramps faster than expected in H2 2026, the cross-sector tailwind reverses and the cloud rebuild thesis degrades.

Catalysts

• June 15, 2026 — Duke Energy 40 MW power delivery confirmation
• July 15, 2026 — NC-1 Q2 billing commencement 8-K (primary catalyst)
• Aug 14, 2026 — WYFI Q2 10-Q (G&A trajectory test)
• Aug 15, 2026 — CRWV Q2 earnings (Hopper pricing validation)
• Nov 14, 2026 — WYFI Q3 earnings (NC-1 revenue recognition test)

What would change our mind

• NC-1 8-K silent through July 15 → conviction haircut, NC-1 P drops from ≈55% toward ≈35%
• Q2 G&A annualized >$80M → cash runway concern moves to front of frame
• CRWV reports H100 pricing DOWN QoQ → Hopper resilience disproved, sector tailwind reverses
• BTBT secondary filing >5M shares → technical destruction overrides fundamentals near-term
• Q3 NC-1 revenue <$2M → thesis broken regardless of guidance language

Structural note on expression

Naked long WYFI is mathematically a BTBT bet with an NC-1 overlay — 50% of variance is uncompensated parent-company factor exposure. Backward-looking regression suggests long $1 WYFI / short $0.89 BTBT triples the Sharpe ratio (0.58 → 1.72) by isolating the residual idio. The hedged structure expresses the catalyst; the naked structure does not.

Evidence