1Q26 results were above our revenue expectation, while headline losses mainly reflected market-driven accounting impacts. Cango reported 1Q26 revenue of US$102.0 million, above our US$72 million estimate, primarily driven by US$98.4 million from Bitcoin mining. Net loss from continuing operations was US$261.1 million, wider than expected, mainly due to non-cash and accounting-related items, including impairment charges on mining machines, and fair value changes related to Bitcoin collateral amid lower Bitcoin prices during the quarter. While the reported loss was significant, we view it largely in the context of Bitcoin price volatility and fleet optimization.

Balance sheet risk was materially reduced, while cost efficiency continued to improve. Cango reduced long-term debt to US$30.6 million from US$557.6 million at the end of 2025, substantially improving its financial flexibility and reducing balance sheet pressure. We forecast interest expense in FY26 will reduce to US$0.9 Million (-84% YoY). As of quarter-end, the company held approximately 1,026 BTC in digital asset reserves. Average cash production cost per Bitcoin declined 9.0% QoQ to US$76,928, reflecting disciplined cost management and the ongoing optimization of its mining fleet. In addition, the US$65 million investment from the chairman and a board director through affiliated entities strengthen capital structure and signaled continued insider confidence in the company’s long-term transition.

Mining operations are shifting from scale expansion toward margin resilience. As of March 31, 2026, total operational hashrate reached 37.01 EH/s, including 27.98 EH/s of self-mining capacity and 9.02 EH/s of leased or hosted hashrate. The company is systematically phasing out less efficient S19 mining rigs and selectively replacing them with higher-efficiency S21 series machines. Cango has also converted certain mining sites into revenue-sharing hosting arrangements, under which counterparties bear direct power, maintenance, and operating expenses while Cango retains revenue participation. This structure reduces direct exposure to site-level operating costs and gives the company greater flexibility during the fleet upgrade cycle.

EcoHash provides a potential second growth curve, though commercialization remains at an early stage. During the quarter, Cango officially launched EcoHash, its new commercial computing platform, marking an important step in its expansion from Bitcoin mining infrastructure into AI computing infrastructure. The company is advancing pilot deployment of modular, containerized high-density compute units, including testing at its Georgia facility, to assess potential cost and efficiency advantages versus traditional data center infrastructure. Its phased approach starts with GPU compute leasing, with longer-term plans to evaluate ecosystem integration through the EcoLink management platform and build toward a broader AI compute network. The strategic collaboration with DL Holdings Group, including a US$10 million convertible note and a memorandum of understanding, may provide additional ecosystem support for this transition.

Valuation remains undemanding, but execution visibility is still developing. Cango’s shares are currently trading at US$0.32 per share, implying a market capitalization of approximately US$130 million. We forecast CY2026 revenue of US$258 million, implying a 0.5x P/S multiple, below the peer group average of 9.2x. The valuation discount likely reflects investor concerns over Bitcoin price volatility, the recent reported loss, and the early-stage nature of the AI compute transition. However, with a materially deleveraged balance sheet, improving mining cost structure, insider capital support, and a clearer EcoHash roadmap, Cango is gradually repositioning itself from a pure Bitcoin mining operator toward a more diversified energy and AI compute infrastructure platform. Key catalysts ahead include further reductions in mining cash cost, evidence of sustainable mining cash flow, progress in EcoHash pilot commercialization, and clearer visibility on the economics of AI compute leasing.