APLD Slides 36% in 3 Months: Should Investors Exit or Hold the Stock?

Applied Digital shares have fallen 35.6% over the past three months, underperforming its sector and AI data-center peers, as investors focus on capital intensity, leverage and dilution risk. The company spent $2.87B on property and equipment in fiscal 2026 and carries $5.31B of scheduled debt obligations plus an estimated $1.78B of lifetime interest payments. Although APLD has $36B of contracted revenue tied to 1.41 GW of capacity, most projects will not begin operating until 2027-28; consensus estimates for its fiscal Q1 2027 loss worsened to $0.26 per share, while fiscal 2027 EPS estimates fell 11.2% in 30 days. Zacks rates APLD a Rank #5 Strong Sell, citing the delayed revenue ramp and an elevated 4.46x price-to-book valuation.

Analysis

APLD is best viewed as a levered construction-duration asset rather than a near-term AI revenue beneficiary. The key equity sensitivity is not contracted headline value but the spread between construction completion/cost escalation and the yield on stabilized lease cash flow; with substantial fixed financing costs ahead of the ramp, even modest commissioning delays can disproportionately impair equity value. Equity issuance is particularly punitive if the stock weakens, creating a reflexive funding-risk premium that peers with operating capacity and lower development exposure do not face.

Near term (days to 3 months), downward estimate revisions and any incremental capital raise can sustain underperformance. The more relevant catalyst window is the next two earnings reports: investors need evidence that energized capacity, customer acceptance, construction spend and unrestricted liquidity are tracking plan. A further miss on deployment timing, rising capex per MW, or higher financing costs should widen the valuation discount; conversely, third-party project-finance commitments or non-recourse asset monetizations would materially weaken the short thesis.

The contrarian case is that the market may be valuing APLD against crypto-linked infrastructure names rather than against scarce, power-secured AI capacity. If lease counterparties are investment-grade and contracts include meaningful take-or-pay protections, de-risking one or two large campuses could trigger a sharp rerating. That is a 6-18 month optionality argument, however, and does not resolve the intervening liquidity and execution gap. Hyperscaler capex strength benefits NVDA and liquid large-cap platforms more directly, while APLD remains exposed to a narrower set of construction, power-delivery and tenant-concentration outcomes.

Key Decisions for Investors

• Maintain an underweight/short bias in APLD over the next 1-3 months, preferably sized modestly because high-short-interest infrastructure equities can gap on financing announcements. Cover or reduce if management demonstrates sequential capex normalization, fully funded construction through the next major delivery milestone, and no additional common-equity issuance.
• Use a relative-value expression: long IREN / short APLD for a 3-6 month horizon, subject to monitoring bitcoin exposure. The trade isolates APLD’s development-finance and dilution risk, but fails if APLD secures low-cost non-recourse financing or IREN’s crypto-linked earnings sensitivity dominates.
• For AI exposure, favor liquid beneficiaries NVDA and ORCL over APLD until operating capacity converts into recurring EBITDA. This avoids underwriting construction completion and refinancing risk while retaining hyperscaler capex exposure; reassess APLD only after delivered MW and customer billing visibly accelerate.
• Set event alerts for any ATM activity, convertible/preferred issuance, construction timetable revisions, and secured-debt pricing. A financing announcement at terms materially more expensive than existing secured debt is a short-add signal; a large asset-level financing with limited recourse to the parent is thesis-negative.