Digi Power X acquires $20M of Nvidia B300 GPUs from Super Micro Computer

Digi Power X agreed to acquire approximately $20 million of next‑generation NVIDIA B300 GPUs from Super Micro Computer to deploy in Tier III AI data centers and offer via its NeoCloudz GPU‑as‑a‑Service platform, with initial customer availability targeted for March 2026 and onboarding in H1 2026. The move is aimed at generating recurring revenue from enterprise and AI customers and the company is targeting a positive return on invested capital within ~30 months of deployment; separately it settled a compensation dispute with H.C. Wainwright, agreeing to pay ~$840,000 cash and issue 269,231 warrants exercisable at $2.85 for five years.

Analysis

Market structure: Winners are NVIDIA (NVDA) and Super Micro (SMCI) as OEM and GPU vendors gain pricing power from sustained enterprise GPU demand; Digi Power X (DGXX) can capture niche recurring revenue but the $20M order is modest against hyperscaler demand so DGXX won’t move market share of cloud leaders. Pricing power will favor NVDA/SMCI for the next 6–18 months if B300 scarcity persists; legacy CPU-focused colos and low-density hosting providers are most exposed to margin compression. Cross-asset: stronger capex for AI hardware supports tech equities and potentially copper/energy equipment demand, may modestly raise credit spreads for smaller colo operators while improving sentiment in high-beta tech and industrial suppliers.

Risk assessment: Tail risks include sudden export-control restrictions on high-end GPUs (weeks-months), a failure to reach targeted utilization (>60% sustained) that pushes ROIC beyond 30 months, and a large customer default concentrated in DGXX’s early book. Near-term (days–weeks) price moves will be muted; medium-term (Mar–Jun 2026) onboarding risk and energy-cost sensitivity (electricity >$0.08/kWh can flip model economics) are critical; long-term (12–36 months) scale and recurring-contract traction determine survivorship. Hidden dependencies: grid capacity, long-term power contracts, and OEM lead times (6–12 weeks) are second-order constraints that can delay revenue recognition. Catalysts: NVDA pricing announcements, SMCI supply cadence, DGXX’s first 60-day utilization reports.

Trade implications: Establish a tactical, small-sized exposure: consider a 1–2% long position in DGXX (high risk) sized to hedgeable capital with a hard stop at -40%, and scale to 3–5% only if DGXX reports >60% utilization for three consecutive months or signs 12-month committed contracts. Add a 3–4% long in SMCI as a levered play on server deployments (target +20–35% in 6–12 months) and a <1% allocation to NVDA 3–6 month call spreads to capture continued GPU pricing upside while limiting premium. Pair trade: long SMCI / hedge with a small short or put on DGXX to express supplier strength vs risky reseller economics. Avoid large allocations to legacy colo REITs until clear demand shift is proven.

Contrarian angles: The market may overrate DGXX’s strategic importance—$20M of GPUs is operationally meaningful for DGXX but immaterial relative to NVDA’s TAM; upside for DGXX depends on execution, not hardware access. Historical parallels to early GPU hosting entrants (2017–2019) show long sales cycles and pricing declines—hourly GPU ASPs could fall 20–40% if competition scales, harming ROIC. An unintended consequence: proliferation of small GPU-as-a-Service providers could accelerate standardized pricing and compress margins for newcomers while consolidating value with NVDA/SMCI. Investment trigger: if DGXX secures multi-quarter committed revenue covering 75% of incremental depreciation and energy costs within 6 months, re-rate to a full position; absent that, treat as speculative.

Key Decisions for Investors

• Establish a tactical 1–2% long position in DGXX (speculative) with a stop-loss at -40% and plan to scale to 3–5% only if DGXX reports >60% GPU utilization for three consecutive months or signs 12-month committed contracts by Q3 2026.
• Initiate a 3–4% long position in SMCI to capture server OEM upside tied to B300 deployments, target +20–35% return over 6–12 months, set a trailing stop at -20% and reassess on NVDA supply cadence updates.
• Deploy a <1% notional NVDA 3–6 month call spread (buy-call/sell higher strike) to capture continued GPU price tailwinds while capping premium; close or roll after NVDA pricing or export-control announcements.
• Execute a pair hedge: long SMCI (3–4%) and buy protective puts or short 1–2% notional in DGXX to express supplier strength vs reseller execution risk until DGXX proves contracted revenue; unwind hedge if DGXX posts contracted ARR covering >75% of incremental opex within six months.
• Do not increase exposure to legacy colo/REITs; reallocate 1–3% of portfolio from low-density colo names into SMCI/NVDA exposure unless those colos secure multi-year GPU-focused contracts by Q4 2026.