Bitdeer's latest operational disclosures reveal a critical divergence in the Bitcoin miner-to-AI infrastructure transition, characterized by a sharp increase in production volume coupled with a drastic reduction in on-chain reserves. The company generated 921 BTC during May, a significant leap from the 196 BTC produced in the same period the previous year, yet its total Bitcoin balance contracted from 1,351 BTC to just 171 BTC. This inverse relationship between output and retention highlights a fundamental liquidity strategy where mined assets are rapidly converted to fund operational expansion rather than accumulated as a treasury reserve. Data compiled by Woofun AI indicates that while production metrics surged, the retained coin stack remained modest relative to the scale of operations, suggesting that the firm is prioritizing immediate capital deployment over long-term asset hoarding. The discrepancy is particularly stark given that the May 2026 figure encompasses both self-mined and co-mined output, whereas the prior year's data reflected self-mining only, yet the magnitude of the holdings gap remains undeniable regardless of the accounting distinction.

The financial mechanics behind this shift are evident in the company's first-quarter performance, which underscores the capital intensity of the dual-strategy model. Bitdeer reported $188.9 million in revenue for the quarter alongside positive adjusted EBITDA of $14.4 million, yet the balance sheet reflects a heavy reliance on debt and cash burn to sustain growth. The firm utilized $346.9 million in net cash for operating activities and deployed $93.7 million in capital expenditures specifically for data center infrastructure, GPU procurement, tariffs, and the delivery of mining rigs. With $1.9 billion in total borrowing and $297.7 million in cash, cash equivalents, and restricted cash on hand, the company is executing a large-scale capital program that interconnects Bitcoin mining, debt financing, and infrastructure investment. This approach treats mined Bitcoin not merely as a store of value but as a liquid asset class convertible into the capital required for aggressive expansion into the AI sector.

The AI Cloud segment presents the primary counter-narrative to the depletion of Bitcoin reserves, offering a potential pathway to stabilize cash flows without relying on volatile coin sales. In May, Bitdeer announced that its AI Cloud Annual Recurring Revenue (ARR) stabilized near $69 million, supported by 90% GPU utilization across 4,248 deployed units, with 3,305 GPUs under external subscription. The company further expanded its technical capabilities by launching two NVIDIA GB300 NVL72 clusters and integrating support for the NVIDIA Nemotron 3 model through its model studio. Woofun AI notes that while the $69 million ARR figure represents a robust run-rate metric, the actual recognized revenue in the first quarter was only $3.7 million, creating a significant gap between potential future visibility and current income statement realization. This distinction is critical because the annualized figure must ultimately translate into tangible cash flow capable of covering power bills, interest costs, and ongoing capital expenditures amidst the inherent volatility of the mining sector.

Strategic logic dictates that the transition to AI infrastructure offers a mechanism to smooth cash generation and reduce the necessity of selling Bitcoin during periods of weak mining economics. Contracted compute revenue can theoretically decouple the company's financial health from hashprice fluctuations, difficulty adjustments, and energy cost variances, thereby making power assets valuable beyond their utility for mining.

However, the execution of this pivot introduces a new set of operational risks, including tenant quality, delivery milestones, GPU supply chain constraints, construction timing, and contract term negotiations. Woofun AI analysis suggests that while these risks may be more predictable than market-driven mining variables, they shift the exposure from price volatility to execution discipline and capital market access. The company's ability to convert owned power sites into long-duration contracted revenue streams, as exemplified by advanced negotiations with a potential colocation tenant at the Tydal site, remains the central proof point for this strategy's viability.

The ultimate investment thesis hinges on whether the AI business can generate sufficient durable cash flow to act as a buffer against routine Bitcoin sales. If the bridge from AI run-rate to actual cash flow materializes, the company could maintain a larger Bitcoin treasury while funding growth through operational earnings. Conversely, if this transition fails to materialize, the pivot may simply alter the form of exposure, resulting in fewer coins on the balance sheet and a heightened dependence on contracted compute revenue and capital markets. The May update leaves this question unresolved, presenting a scenario where the company mined 921 BTC but the more critical metric is the 171 BTC remaining at month-end. This residual balance serves as a barometer for the success of the strategy, indicating whether the firm is successfully leveraging AI revenue to preserve its Bitcoin holdings or if it remains dependent on continuous asset liquidation to sustain its capital-intensive buildout.