Cango Inc. (NYSE: CANG) reported a second-quarter net loss from continuing operations of $81.6 million, a significant improvement from the $261.1 million loss recorded in the prior quarter. The company also marked a strategic milestone by signing its first customer for its new high-performance computing (HPC) business.

The Georgia-based facility, converted from an existing bitcoin mining site, is now operational with 3 MW of computing capacity. CEO Paul Yu stated that Cango will run its bitcoin mining and AI operations as parallel businesses, with expectations to generate initial revenue from the AI segment in the third quarter.

Financial Performance

Revenue from continuing operations fell roughly by half to $50.8 million in the second quarter, down from $102 million in the first quarter. This decline was driven by the phasing out of older, less efficient mining machines and a strategic shift from self-mining to a hosted leasing model.

Despite the revenue contraction, profitability metrics improved sharply due to cost discipline and operational efficiency gains.

Long-term debt remained low at $31.2 million at the end of June, slightly up from $30.6 million at the end of March but significantly reduced from $557.6 million three months prior to that. The company used cash from bitcoin sales to pay down debt earlier in the year.

Bitcoin Mining Operations

Cango mined 656 bitcoins during the quarter, averaging about 219 per month, down from a monthly average of 422 bitcoins in the first quarter. Total mining capacity decreased to 27.58 EH/s as of June 30, from 37 EH/s at the end of March.

The average cost to mine each bitcoin stood at $73,313, down 5% sequentially. This unit cost is below the current market price of approximately $77,500. However, the all-in mining cost was $98,405 per bitcoin, which remains above the market price.

The company’s bitcoin treasury held 1,056 bitcoins at the end of June, similar to the 1,026 held at the end of March, but down dramatically from nearly 7,500 in January before the company began selling holdings. CFO Simon Tang noted that the company has started executing a hedging strategy for risk management purposes only.

What the Numbers Show

A critical divergence exists between Cango’s unit economics and its overall profitability. While the direct mining cost of $73,313 per bitcoin is below the market price of $77,500, the all-in cost of $98,405 exceeds it. This indicates that while operational efficiency has improved, fixed costs and other overheads still pressure the bottom line, resulting in a net loss despite positive contribution margins on individual coins mined.

AI Business Expansion

The Georgia facility serves as a proof of concept for hosting heavy-duty computing power for AI applications. The site has infrastructure supporting up to 3 MW of computing power, with room for future expansion. Hardware installation is proceeding in batches to support a phased ramp-up.

Outside Georgia, Cango has begun operating AI test nodes in Texas and on the U.S. West Coast to serve customers with proximity-based deployment needs. The company continues to evaluate potential new sites and the possibility of building new HPC facilities.

Cango Inc. (NYSE: CANG) announced that its high-performance computing subsidiary, EcoHash Technology LLC, has completed key operational and commercial milestones for its AI compute business. The company activated an initial batch of GPU servers at its owned facility in Georgia.

Infrastructure Deployment

EcoHash finished infrastructure modifications and retrofits for the dedicated AI compute section within Cango's 50MW Georgia facility. This specific section supports up to 3 megawatts of capacity, with scope for future expansion. The initial deployment involved delivering, installing, and testing high-density compute containers. These containers have commenced their operational rollout through the power-on and activation of the first batch of GPU servers.

The containerized data center solution serves as a modular building block for AI compute. It enables modular deployment and high-density GPU capacity in a compact footprint. By standardizing power, cooling, networking, and operations into a repeatable unit, Cango aims to scale its AI infrastructure faster and more cost-effectively across suitable sites.

Technical Specifications

The site's modular design relies on standardized, high-density AI compute modules engineered for sustained high-density GPU loads. Key technical components include:

• Power delivery through an industrial-grade transformer setup
• Precision cooling system supporting the site's compute density
• Fiber infrastructure providing network connectivity for AI compute workloads

Because these compute modules were deployed alongside Cango's existing mining infrastructure at the same Georgia site, EcoHash minimized new construction and site retrofit work. This approach accelerated the path from module delivery to power-on.

What the Numbers Show

The data reveals a strategic leverage of existing assets to reduce capital intensity in the AI transition. EcoHash utilized only 3 megawatts of the available 50 megawatts capacity at the Georgia facility for this initial AI compute phase. This indicates that the vast majority of the site's power capacity remains allocated to Bitcoin mining or is reserved for future expansion, allowing Cango to diversify its revenue streams without requiring immediate greenfield construction.