Will APLD achieve FY2027 adjusted EBITDA above $250M?

The math strongly supports $250M+ if Building 2 delivers on or near its July 1, 2026 RFS target. Building 1 contributes ~$176M at its proven $441K/MW/quarter run-rate for all 4 quarters of FY2027. Building 2 at 150MW needs to contribute only ~$74M, which requires roughly 1.1 quarters at full utilization — achievable even with a modest 2-month delay and 3-month ramp. The context analysis shows that even a September RFS with 3-month ramp produces ~$275M total, still above $250M. The primary risk is a delay beyond October 2026, which I estimate at roughly 30-35% probability given that APLD has publicly committed to the July timeline and CoreWeave's lease structure creates mutual incentives for timely delivery. Base case: ~58% probability of exceeding $250M.

This is a compound probability question that the market may be underweighting the execution risk for. APLD has exactly one data point of successful data center delivery and ramp (Building 1). Construction delays are endemic in the data center industry — typical ramp from RFS to stabilized operations is 1-3 quarters per context. Building 1 itself had a partial Q2 before full Q3 contribution, suggesting the ramp period is at least one full quarter. If Building 2 follows the same pattern with a July RFS, full contribution wouldn't begin until October-November 2026, giving only ~2 quarters of full EBITDA — roughly $132M from B2 + $176M from B1 = $308M. But any delay to the July RFS pushes this below $250M quickly. Additionally, operating costs at 250MW are unproven — shared SG&A assumptions may not hold at 2.5x scale. The $2.7B debt burden and $182M annual interest expense are real constraints on operating leverage.

The CoreWeave 15-year lease with make-whole provisions creates strong bilateral incentives for on-time delivery. Both parties lose revenue during any delay. The $2.15B refinancing at 6.75% (saving $53.5M annually) improves the EBITDA math and signals institutional confidence in APLD's execution timeline. Analyst consensus is unanimously bullish (9/9 Strong Buy, $54.25 avg target vs $27.79 current), suggesting the market broadly expects execution on the current timeline. The Q3 EBITDA of $44.1M from 100MW is a clean data point — adjusted EBITDA from real base rent, not fit-out revenue (per Fugazi Filter). The unit economics are proven at the per-MW level. The primary question is timing, and the math gives reasonable margin: even a 2-month delay scenario produces $275M. I weight this at 55% probability, reflecting genuine construction risk balanced against strong incentive alignment and proven unit economics.

Breaking this into component probabilities: (1) Building 2 RFS by July-August 2026: ~60-65% probability given public commitment and CoreWeave incentives, but construction delays are common; (2) Ramp to near-full utilization within 3 months of RFS: ~70% probability based on Building 1 precedent; (3) Building 1 maintaining Q3 run-rate: ~90% probability given contractual nature. If B2 on time with fast ramp: ~$341M total (well above). If B2 on time with slow ramp: ~$280M (above). If B2 delayed 2 months with fast ramp: ~$275M (above). If B2 delayed 2 months with slow ramp: ~$240M (below). If B2 delayed 5+ months: ~$226M (below). Weighting these scenarios, probability of exceeding $250M is approximately 52%.

Data center construction delays are the industry norm, not the exception. APLD has a single successful delivery to date, which is insufficient to estimate reliability. The 16-month horizon to resolution introduces additional macro risks — potential recession, AI capex slowdown, or CoreWeave financial stress could all impact the timeline. The $2.7B debt burden at 6.75% creates $182M in annual interest expense, meaning APLD's financial flexibility to manage construction overruns is limited. Furthermore, the $250M threshold assumes per-MW EBITDA holds at $441K when operating at 2.5x current scale — this is extrapolation from a single quarter. SBC at $39.3M/quarter is excluded from adjusted EBITDA but represents real dilution that may eventually force accounting changes. The narrative-reality gap (DIVERGING) appropriately captures the distance between current operations and management's vision.

This is genuinely a coin-flip question. The bull case: proven unit economics ($441K/MW/quarter), contractual revenue certainty (CoreWeave 15-year lease), management's public July 1 commitment, and the math showing even moderate delays still clear $250M. The bear case: single delivery track record, compound execution risk (delivery AND ramp), unproven operating costs at scale, and industry-standard construction delays. The context analysis helpfully brackets the scenarios: on-time = $341M (easy clear), 2-month delay = $275M (modest clear), 5-month delay = $226M (miss). The question resolves to whether a 5+ month delay is likely, and I estimate that at roughly 25-30%. But ramp speed is the second variable — if Building 2 takes 2 quarters to reach full utilization instead of 1, each scenario loses ~$66M. This tips several scenarios below $250M.

Building 1 proves $441K/MW/quarter EBITDA at 100MW. Building 2 at 150MW with July RFS and 3-month ramp contributes ~$165M in FY2027. Combined with Building 1's $176M = $341M, well above $250M. Even with 2-month delay the math works ($275M). Only a significant delay (5+ months) causes a miss. CoreWeave lease incentivizes timely delivery. Lean slightly above 50%.

Only one successful delivery in APLD's history. Construction delays are standard in data center industry (1-3 quarter ramp typical). Building 1 had partial Q2 suggesting full ramp takes a quarter minimum. At 250MW, operating costs may scale worse than expected. $2.7B debt creates financial pressure. The 16-month horizon introduces additional uncertainty. Below coin-flip probability.

The math provides a reasonable margin — $250M is not the absolute best case but rather a threshold that accommodates moderate delays. Building 1's $176M annual contribution is contractually secure. Building 2 needs only $74M, achievable in ~1.1 quarters at full utilization. However, compound risk of delay plus slow ramp is real. True coin-flip given the opposing forces.