Red Flags

APLD — Applied Digital Corporation

Applied Digital Corporation designs, develops, and operates digital infrastructure solutions to high-performance computing (HPC) and artificial intelligence industries in North America. It operates through: Data Center Hosting, and HPC Hosting Business. The company provides energized infrastructure services to crypto mining customers; and cloud services to customers, such as AI and machine learning developers. It also engages in the designing, constructing, and operating of data centers to provide computing power and support HPC applications. The company was formerly known as Applied Blockchain, Inc. and changed its name to Applied Digital Corporation in November 2022. Applied Digital Corporation is based in Dallas, Texas.

**THE CORE CONCERN**

Applied Digital’s Beneish M-Score of 0.82 breaches the −1.78 manipulation threshold on a scale where values closer to zero indicate higher statistical likelihood of earnings manipulation. The same model produced comparable readings for Enron in fiscal 1997–1998. The score is driven primarily by a DSRI of 3.08—accounts receivable expanding 208 percent faster than revenue—and an SGI of 2.67. These ratios coincide with revenue rising from $0.1 billion to $0.6 billion over the measured period, a 122.6 percent compound annual growth rate. While the company reports no cumulative net-income-to-operating-cash-flow divergence and zero years of positive earnings paired with negative cash flow, the M-Score still flags the pattern of accrual growth relative to the rapid top-line expansion.

**HISTORICAL PRECEDENT**

The combination of an M-Score near zero, sharply elevated DSRI, and outsized revenue growth mirrors the profile observed at certain high-growth technology and infrastructure companies that later faced restatement or regulatory scrutiny. In those cases, the initial forensic signal was not negative operating cash flow but rather the disproportionate buildup of receivables and contract assets that proved difficult to collect once growth normalized or financing conditions tightened.

**WHAT TO VERIFY**

Investigators should examine three items in the most recent 10-K and 10-Q. First, the aging schedule and concentration of accounts receivable, specifically whether any customer or related-party balance exceeds standard payment terms by more than 90 days. Second, the precise revenue-recognition criteria for data-center and hosting contracts, including the percentage of revenue booked on a percentage-of-completion or milestone basis versus upon delivery. Third, the reconciliation between reported revenue growth and cash collections from customers, with particular attention to any non-cash consideration or vendor financing arrangements disclosed in the footnotes.

**COUNTERARGUMENTS**

Rapid revenue expansion itself generates several of the observed ratios. High-growth companies in information-technology services routinely exhibit DSRI values above 2.0 and SGI above 2.5 during periods of customer acquisition and longer enterprise contracts; these metrics do not, by themselves, demonstrate manipulation. The company remains pre-profitability, a stage at which negative cash flow and elevated accruals can reflect legitimate investment in capacity rather than accounting distortion. Altman Z-Scores below 1.81 are also common among development-stage technology firms that have raised substantial equity or debt to fund infrastructure build-out. Short interest at 22.1 percent of float may simply reflect market awareness of execution risk in a capital-intensive sector rather than confirmed accounting concerns. Distinguishing between these explanations requires direct verification of the receivable aging and contract terms noted above.

CORZ — Core Scientific, Inc.

Core Scientific, Inc. provides infrastructure for high-density colocation services and digital asset mining in the United States. It operates through three segments: Colocation, Digital Asset Self-Mining, and Digital Asset Hosted Mining. The company offers space, power, cooling, facilities operations, security, and other services to third-party customers to support workloads for machine learning and artificial intelligence; deployment, monitoring, troubleshooting, optimization, and maintenance of its customers' digital asset mining equipment; and electrical power, repair, and other infrastructure services for its customers to operate, maintain, and mine digital assets. It is also involved in the deployment and operation of its fleet of miners within its digital infrastructure as part of a

1. THE CORE CONCERN

The most direct quantitative signal is the Beneish M-Score of 62.68, which breaches the −1.78 manipulation threshold by an extreme margin. On the negative scale used by the model, values closer to zero indicate higher statistical likelihood of earnings manipulation; a positive reading of this magnitude places the company far into the zone the model historically associated with Enron in FY1997–1998. This occurs alongside an Altman Z-Score of 0.16 (well below the 1.81 distress threshold) and a DSRI component of 72.52, indicating accounts receivable expanded 7,152% faster than revenue. Revenue has contracted at a −8.9% CAGR from $0.6B to $0.4B over four years, yet the M-Score flags remain elevated despite zero years of positive net income paired with negative operating cash flow and a cumulative NI–OCF divergence of $0.0B.

2. HISTORICAL PRECEDENT

The combination of an extreme M-Score breach, sharply elevated DSRI, and a Z-Score deep in the distress zone mirrors the pattern the Beneish model identified in Enron three years before its 2001 collapse. In that case, the model flagged aggressive revenue recognition and accruals management well before cash-flow shortfalls became visible in reported operating results. Here, the absence of current NI–OCF divergence does not eliminate the precedent; Enron also showed periods where reported earnings appeared supported by cash flow before the underlying receivables and contract accounting issues surfaced in subsequent filings.

3. WHAT TO VERIFY

Investigators should examine three items in the 10-K or 10-Q. First, the precise composition and aging of the accounts receivable balance that produced the DSRI of 72.52, including any concentration with related parties or cryptocurrency counterparties. Second, the footnotes on revenue recognition for colocation and hosting contracts, specifically the criteria used to determine when performance obligations are satisfied and whether any portion is recognized on a percentage-of-completion or milestone basis. Third, the reconciliation of the $5.3B market capitalization to tangible asset values and debt covenants, given the Z-Score of 0.16 and 22.2% short interest.

4. COUNTERARGUMENTS

The signals could reflect non-manipulative factors. Core Scientific operates in a capital-intensive, volatile sector where large equipment financings and customer prepayments can distort receivable ratios without implying improper recognition. The revenue decline itself may stem from Bitcoin price cycles and data-center utilization shifts rather than channel stuffing. A high beta of 5.48 is consistent with the industry’s sensitivity to energy costs and crypto economics, which can produce statistical outliers in accrual and distress models calibrated on more stable sectors. Finally, the zero NI–OCF divergence and clean accruals ratio of 0.000 provide some evidence that current-period earnings are not being inflated through working-capital accruals. These factors warrant examination but do not by themselves confirm manipulation.