Cango Inc. ($CANG)

Hello, and welcome to the Cango Inc. First Quarter 2026 Earnings Conference Call. [Operator Instructions] Please note, this event is being recorded. I would now like to turn the conference over to Mr. Paul Yu, Chief Executive Officer. Please go ahead.

Good morning, everyone, and thank you for joining Cango's First Quarter 2026 Earnings Call. First, I will summarize our key financials and operational performance for the quarter. The first quarter of 2026 was characterized by industry-wide adjustments and our results reflect these macro headwinds alongside our ongoing efforts to manage our strategic transition. During Q1, we generated total revenue of approximately $102 million, primarily driven by revenue from our Bitcoin mining business. We reported a net loss from continuing operations of $261.1 million primarily due to noncash impairment charges on Bitcoin mining machines and loss from changes in fair value of receivable for Bitcoin collateral, both resulting from the decline in Bitcoin market price. By the end of the quarter, we held 1,025.7 Bitcoin, and we reduced our long-term debt to $30.6 million. As of March 31, 2026, Cango's total operational hash rate was 37.01 exahash per second, comprising 27.98 exahash per second of self-mining capacity and 9.02 exahash per second of hosted hash rate. This operational model prioritizes margin resilience over scale. In Q1, we mined 1,266 Bitcoin. Through disciplined cost management, our average cash cost per Bitcoin mined was $76,928 showing a 9% decrease from Q4 2025. These figures reflect our continued focus on profitability and operational efficiency as our business model evolves. Following this brief quarterly review, I'd like to provide an update on our operational activities during April and May, which offer additional context regarding our strategic direction. Regarding our mining business, our immediate priority is to streamline operations and carefully manage our resources at location. In April, we maintained our focus on cost optimization measures and operational efficiency. Our self-mining operations produced 230.04 Bitcoin in for the month when the average cash cost per core further decreased. This result stems primarily from our ongoing fleet upgrade beginning in March, we have been selling less efficient older generation S19 miners and selectively replacing them with more energy-efficient S21 series machines. As of the end of May, within our self-mining hash rate composition, the contribution ratio between S19 and S21 models is approximately 8:2. This operational mix supports our efforts to enhance our overall cost structure. Our objective is to manage our Mining segment toward an operational baseline capable of supporting improved cash flow resilience. Currently, some sites have transitioned to a revenue-sharing hosting arrangement, while this arrangement introduces depreciation expenses on our financial statements from a cash perspective, the hosting structure requires the counterparty to cover direct power costs and maintenance and operation expenses, allowing us to participate in revenue sharing while reducing our direct exposure to site level operating expenses. This structure helps mitigate operating risk and provides an operational buffer as we optimize our fleet. As our fleet adjustments proceed and stabilize, our strategic intent is to focus our operations primarily on disciplined self-mining while managing an orderly exit from less efficient hardware or higher-cost sites as of April 30, through a diversified footprint across 26 active mining sites globally. We operated a total hash rate of 31.58 exahashes per second, comprising 20.43 exahashes per second in self-mining capacity and 11.15 xahashes per second in hosted capacity. This current hash rate structure helps mitigate operational risk, supporting our ability to manage market volatility and execute our fleet upgrade strategy. Next, turning to our AI infrastructure initiatives. The objective of EcoHash is to leverage Cango's power access and mining operational expertise to develop a standardized compute solutions. We are continuing to advance our milestones. Pilot evaluation, site retrofitting and hardware installation at our Georgia location have progressed significantly and testing for modular high-density compute units is underway. Our objective with this modular design is to evaluate whether modular development can reduce cost and improve operational efficiency relative to traditional data center infrastructure. Operational model. This framework is intended to allow us to utilize existing operational assets to address market demand aiming to serve small and medium-sized enterprise efficiently. Based approach, our multistate strategy begins with an entry to GPU compute capacity leasing. Over the long term we plan to evaluate ecosystem integration through Ecolink management platform with the objective of developing an AI compute network. We have taken a disciplined approach to improve our capital structure and balance sheet position. Through active treasury and debt management, we have reduced our Bitcoin-backed loan balance to approximately at $30.6 million. Concurrently, our remaining Bitcoin reserves stands at 1,057.46 Bitcoin as of April 20, reflecting our strategic priority to lower leverage and reserve balance sheet stability. Our strategic alignment and partnerships support our ongoing operational focus. In Q1, our Chairman and our Board Director made an investment of $65 million in the company through entities they control. Furthermore, we established a strategic collaboration with DL group a Hong Kong listed company, which includes a $10 million convertible note and a strategic operation MoU which complements our commitment to AI infrastructure opportunities. As we look to the remainder of 2026, we have closely monitoring the evolving dynamics between global AI compute demand and power infrastructure capacity. Within this market environment, our operational priorities are twofold. First, to continue optimization of cost efficiency of our mining business; and second, to methodically advance the evaluation of EcoHash and continuous technical testing of our pilot project. We will continue to approach our strategy with a focus on capital discipline, aiming to leverage our existing infrastructure assets to support long-term stability and shareholder value. That concludes my remarks. I will now turn the call over to our CFO, Simon for a detailed financial review. Thank you.

Thanks, Paul. Hello, everyone, and welcome to our first quarter earnings call. Before I start to review our financials, please note that unless otherwise stated, all amounts discussed are in U.S. dollars. Total revenues in the first quarter was $102 million. Revenue during the quarter from the Bitcoin mining business was $98.4 million with a total of 1,200 and 66.1 Bitcoins mined during the period. The average cost to mine Bitcoin, excluding depreciation of mining machines, was $76,928 per Bitcoin with all-in cost of 99,747 per Bitcoin. Compared to the fourth quarter of 2025, total revenue decreased by approximately 43%. This decline primarily reflects our proactive reduction in operational hash rate as we began to phase out older and less-efficient S19 series mining machines and temporarily transitioned some capacity to a leasing model that Paul discussed just now. While this adjustment has reduced top line mining revenue, it has also contributed to lower operating costs and improved cash flow profile. And some of these efforts remain ongoing in the second quarter as we speak. Now let's move on to our cost and expenses. Cost of revenue, excluding depreciation in the first quarter was $99.6 million, down from $155.3 million in the fourth quarter, driven by lower electricity and hosting expenses following the hash rate reductions. Depreciation in the first quarter was $29.4 million. General and administrative expenses, including related parties totaled $7.2 million. There was an impairment loss from mining machines in the first quarter of $49 million and a loss on disposal of mining machines in the first quarter of $20.3 million. Loss from changes in fair value of receivable for Bitcoin collateral was $151.8 million compared to $171.4 million in the fourth quarter. This noncash loss was primarily driven by the decline in Bitcoin price during the quarter as we started off this quarter with over 7,500 Bitcoins. Operating loss for the quarter was $254.4 million with a net loss from continuing operations of $261.1 million. On a non-GAAP basis, adjusted EBITDA was a loss of $154.1 million, of which there was a $151.8 million impact from the loss from changes in fair value of receivable for Bitcoin collaterals. Moving on to our balance sheet. As of March 31, we had cash and cash equivalents of $7.2 million, down from $41.2 million at year-end, mainly due to debt repayments and operational activities. That said, our balance sheet also includes cryptocurrencies of $7.9 million as well as receivables for Bitcoin collaterals of $68.2 million. In terms of operational assets, we carried our mining machines at a net value of $130.8 million. On the liability side, we had $30.6 million in long-term debt, which is significantly lower than the $557.6 million recorded as of year-end. The substantial reduction in both the receivable for Bitcoin collaterals and the associated long-term debt reflects our proactive deleveraging efforts during the quarter. By selling a portion of our Bitcoin holdings and using the proceeds to repay related party loans, we have meaningfully strengthened the balance sheet and also reduced our interest expenses. This concludes our prepared remarks. Operator, we are now ready to take questions.

[Operator Instructions] Your first question comes from Pingyue Wu from Citic Securities.

I'm Pingyue from Citic Securities. And my first question is the company's cash cost per core declined in the first quarter compared with the first quarter of last year, and management also mentioned further optimization in April. What were the main drivers behind the cost reduction? Is there still room for further cost improvements going forward? And also my second question is our management team mentioned that the 2026 strategy is efficiency over scale. And in April, total operating hash rate was 31.55 exahash per second, including 11.50 exahash per second of leased hash rate. Will the hash rate continues to decline over the next few months? Could you explain in more detail how the leasing model works and a specific impact on the financial statements?

Regarding your first question, the cost reduction was mainly driven by 2 factors. First, we proactively phased out part of our higher energy consumption S19 series mining machines and gradually replaced them with small energy-efficient S21 series models. Second, we continued to migrate hash rate to regions with lower power costs including developing next-generation miners in locations such as Paraguay and Oman. At the same time, we temporarily adopted a revenue-sharing model at certain higher-cost mining sites, which effectively reduced power costs. Looking ahead, we intend to leverage our ongoing fleet upgrades and as some of our hosting contracts expire, we will strive to optimize our hosting arrangements to lower power costs.

And I'll take your second question with regards to the hash rate, we're not spending a hard hash rate target and instead, we're really focusing on margin and cash flow KPIs for the mining business for now. We do -- and we are continuing to retire older S19 series machines in certain higher power cost sites. So during this period, our total hash rate may experience modest fluctuations in the short term. And at the same time, we are selectively deploying more energy efficient S21 machines. So this process has helped us reduce cash cost per point and improve the resilience of our mining fleet in general. And as for your question regarding our leasing model, we reiterate that it is a temporary arrangement, especially with some of the higher cost sites where the arrangement instead of paying for power cost on a consumption basis. The Bitcoin mines will go to the site owner who will share mining revenue with us based on the agreed ratios. And thereby, the power cost and maintenance and operation fees are born by the site owner. From a cash flow perspective, this leasing model ensures that we do not mine at a loss purely as a result of the higher cost. And this is our core strategy to protect -- in line with our core strategy to protect cash flow. And currently, the lead hash rate is mainly deployed in certain parts of America, but this may change once the relevant -- once the respective mining hosting contract expires. So we will enter into new contracts or alternatively, we may move the machine to alternative sites.

[Operator Instructions] Your next question comes from Marco Zhang from Geelong Research.

This is Marco from Geelong Research. I have 3 questions here. My first question is regarding your Bitcoin business. You sold 2,000 Bitcoin in Q1 and currently hold approximately 1,057 Bitcoins. Will the company continue to sell Bitcoin going forward? Has the company's long term holding strategy changed?

Our BTC treasury strategy has shifted from mine and hold to a more dynamic balanced approach. Given the current level of market volatility, we placed greater emphasis on liquidity and balance sheet strength. The BTC sale in Q1 was mainly used to reduce BTC-backed loans and the outstanding loan balance has now declined to approximately $30.6 million as of the end of the first quarter. Going forward, we will address flexibly based on market price, operational needs and debt levels, while we maintain a positive long-term view on Bitcoin, our treasury decisions will align with our overall capital allocation strategy.

Got it. So we understand that the company's AI business will be carried out through EcoHash. Could you share an update on the [ LN ] pilot mentioned previously. Are there any specific commercialization milestones for 2026? And when could it start contributing revenue?

Sure. So the [ LN ] site is currently our only fully self-owned infrastructure assets with 50 megawatts of grid connected capacity, and the power contract is in place till 2029. And in terms of the progress of the construction and renovation, that in itself is now close to completion, and we have placed orders for standardized compute containers, which are arriving at phases and will be ready for installation and testing very soon. We plan to activate a portion of the power capacity at this site for this purpose. And at the same time, this site is expected to serve as a real-world production environment showroom. What that means is that the containers are of different specifications, and we expect to evaluate and showcase the different specifications. There are air cooled containers, liquid cooled as well as hybrid containers for different environmental conditions. And this allows us to assess the conversion, deployment and operating performance of the compute nodes in an actual set environment. So this project in itself is a proof-of-concept stepping stone towards scaled commercialization initiatives. And once this model is ready and proven, we'll evaluate opportunities to replicate this model at other suitable sites as well, whether it be sites for our wholesale partners. And from the perspective of the overall AI project build-out, we have not set any specific revenue target at this point, but revenue generation will start in the second half of this year. Our top priority at the moment is to complete the technical validation of this pilot, and we're in the process of ordering a small number of servers at the moment. If the validation results meet expectations, we will be begin to work with partners to deploy more compute nodes. AI compute services take time to move from pilot stage to scale, but we will update the market in a timely manner once there's substantial progress. Thank you.

Got it. And how about the CapEx or how much CapEx will be required for the EcoHash pilot and future expansion? And how do you plan to fund it?

We're actually doing in phases. So the thing about our business model on this side that is modularized. So in terms of the containers, we have the flexibility of doing it per container. So in terms of the CapEx, we're being very prudent at the moment. And in the first is the model validation phase, we will mainly use our own capital right now. So we've deployed our own capital for the site renovation. So the Georgia pilot leverages the existing site infrastructure and the power, right? And the retrofit cost is relatively limited. The bulk of the project CapEx itself will be for the purchases of the servers, which we're in the process of doing right now. In the future, we do hope that we'll be able to use other types of financing, whether it's GPU-backed financing or using a financial lease model rather than just purely rely on our own capital. And obviously, we are open to and hope to establish other strategic partnerships as well so that we can do it together with other partners.

There are no further questions at this time. I'll now hand the conference back to management for any closing remarks.

No. We don't have any closing remarks. Thanks a lot.

Thank you. That does conclude our conference for today. Thank you for attending today's presentation. You may now disconnect.