At the beginning of 2026, a decision by Bitcoin mining company Cango drew widespread attention in the market. By the end of 2025, this mining company held over 7,528 bitcoins. In early February, it sold 4,451 BTC in one transaction, cashing out approximately $305 million to repay debt and support its strategic pivot to artificial intelligence (AI) computing infrastructure. This move wasn’t an isolated event—it reflected a collective shift across the Bitcoin mining sector in response to current market conditions. As mining costs surpass the price of Bitcoin, what was once considered a core asset—Bitcoin reserves—is now being redefined as a strategic resource to be deployed as needed.
Why Are Mining Companies Reducing Their Bitcoin Holdings During a Bear Market?
Cango’s decision to reduce its Bitcoin holdings directly addresses a harsh market reality: the economics of mining have fundamentally reversed. According to industry data, as of March 2026, the all-in cost to mine one Bitcoin was around $87,000, while the market price hovered near $67,000. This means each Bitcoin produced resulted in a net loss of $20,000. For Cango, its average total mining cost (including depreciation) in Q3 2025 was as high as $99,000 per coin, significantly above current market prices.
Against this backdrop, the logic of holding Bitcoin as a "store of value" no longer holds. In its official announcement, Cango made it clear that the sale aimed to reduce financial leverage, strengthen the balance sheet, and provide capital for strategic expansion into AI computing infrastructure. As of February 28, 2026, Cango’s Bitcoin holdings had dropped to 3,313.4 coins, while its deployed computing power remained steady at 50 EH/s. This demonstrates the company is rebalancing its asset structure—shifting version from "holding digital assets" to "controlling physical computing power."
What Drives the Mining Industry’s Shift to AI Computing?
There’s a natural physical connection between mining and AI computing: electricity and infrastructure. Cango’s strategic roadmap makes this mechanism clear: it leverages its globally connected, grid-integrated infrastructure to provide distributed computing power to the AI industry.
The core of this transformation lies in the repricing of computing resources. Bitcoin mining revenue faces triple pressures from price volatility, mining difficulty adjustments, and hardware depreciation. In contrast, AI data centers offer long-term contracts of 10 to 15 years, investment-grade enterprise clients (such as Microsoft and Meta), and stable, predictable dollar cash flows. Cango’s plan unfolds in three stages: in the short term, it deploys containerized GPU nodes at existing sites to serve small and medium-sized business needs; in the medium term, it develops a software orchestration platform to integrate distributed resources; and in the long term, it aims to become a mature AI infrastructure platform. To accelerate this process, the company appointed former Zoom technical expert Jack Jin as Chief Technology Officer for AI, leveraging his experience in GPU cluster deployment to support the new strategy.
What Are the Trade-Offs and Costs of This Structural Transformation?
Transformation comes with costs. In its February 2026 operational update, Cango disclosed that its average operating hash rate for the month was 34.55 EH/s, below its deployed 50 EH/s, due to "temporary downtime related to fleet optimization and relocation." This highlights the inevitable adjustment pains during the transition from ASIC miners to GPU computing. About 31% of its computing power went offline for upgrades, resulting in short-term revenue loss.
A deeper trade-off lies in the shift in corporate positioning. At one point, Cango was the world’s second-largest publicly listed Bitcoin miner. Its "HODL + mining accumulation" model was self-reinforcing during bull markets: rising Bitcoin prices boosted net asset value, which in turn supported further expansion of computing power. But the 2026 market environment forced the company to reassess this approach. Selling Bitcoin reserves means giving up potential gains from future price rebounds in exchange for immediate financial stability and the cash flow needed for transformation. This is a strategic choice between time preference and risk exposure—trading future uncertainty for present structural survival.
What Does the Industry-Wide Shift from Bitcoin to AI Mean for the Crypto Market?
From a market structure perspective, the collective transformation of mining companies could have a profound impact on Bitcoin’s supply and demand dynamics. For years, miners have been the largest "structural sellers" in the Bitcoin market—they regularly sell mined coins to cover electricity and operational costs. Once mining companies pivot to AI service providers, their stable dollar revenues come from AI hosting contracts, eliminating the need for routine Bitcoin sell-offs and potentially even turning them into buyers.
On-chain data is already reflecting this shift. In early 2026, corporate Bitcoin treasuries saw three consecutive weeks of reductions, with Cango alone cutting its holdings by over 54% in two weeks. While this wave of selling put short-term pressure on prices, if the trend continues, the market’s largest "natural short sellers" are systematically exiting. This is a substantial long-term positive for Bitcoin’s supply structure. The Hash Ribbon indicator shows that the miner capitulation period from late November 2025 to now is one of the longest on record—such capacity shakeouts typically signal a market bottom is near.
How Will Mining and AI Computing Evolve Together in the Future?
Looking ahead, the relationship between mining and AI computing may develop into a dynamic balancing mechanism. MARA’s hybrid model offers a blueprint: using the same power infrastructure to flexibly switch between Bitcoin mining and AI computing. When electricity prices are low, computing power is allocated to Bitcoin mining; during peaks in AI demand, resources shift to GPU services. In this model, Bitcoin mining is "downgraded" from a core business to a flexible load balancer—filling electricity costs when AI demand is low and yielding to higher returns when AI demand surges.
Cango, on the other hand, is aiming for a complete transformation. The company has explicitly stated its goal to become a "global distributed inference computing grid." Its 40 global sites and grid-connected infrastructure form the physical foundation for this vision. The "grid edge" resources accumulated from Bitcoin mining—sites near cheap electricity but far from traditional data centers—are ideally suited for distributed AI inference deployments. The future of mining may no longer be a simple "hash rate race," but rather operating as distributed computing infrastructure providers.
What Are the Potential Risks and Limitations of This Transformation Path?
The road to transformation is far from smooth. First, there are significant technical challenges: Bitcoin mining relies on ASIC chips, while AI computing requires GPU clusters and sophisticated orchestration software. While Cango has brought in technical leadership, bridging the gap from toward mining machine operations to AI infrastructure management will take time.
Second, capital markets have limited patience. After selling its Bitcoin, Cango’s balance sheet improved, but the company still faces cash flow pressures. Analyst data shows its levered free cash flow stands at negative $252 million. Building AI infrastructure is capital-intensive with long payback periods. If financing conditions tighten, the company could face liquidity risks.
Finally, market competition is intensifying. Other miners like Core Scientific and Bitdeer are also pivoting to AI. As more companies enter the field, competition for prime power resources, GPU supply, and client contracts will become increasingly fierce. Whether a mining company can establish differentiated competitive advantages before "compute oversupply" sets in is a key question for every player making this transition.
Conclusion
Cango’s strategic shift from Bitcoin mining to AI computing power is a microcosm of the crypto industry’s cyclical adjustment and structural transformation. When the narrative of "holding equals belief" meets the reality of economic pressure, miners are redefining their core assets and business boundaries. While this transformation brings short-term selling pressure, it could fundamentally reshape Bitcoin’s supply-demand dynamics in the long run. As miners evolve from "passive sellers" to "infrastructure operators," the crypto market will see more mature participants.
FAQ
Q: Does Cango’s Bitcoin sale signal a bearish outlook on Bitcoin?
A: Not necessarily. Cango’s reduction in holdings is primarily a financial restructuring—reducing leverage and freeing up liquidity to invest in AI infrastructure. The company stated in its announcement that it remains committed to mining operations and continues to optimize mining economics. This is a shift in asset allocation strategy, not a judgment on the asset’s intrinsic value.
Q: How does the mining industry’s shift to AI affect Bitcoin network security?
A: In the short term, some hash rate going offline may slow the network’s overall hash rate growth. In the long term, however, this is a healthy capacity shakeout. Inefficient miners who can’t handle high costs exit, leaving more efficient and professional operators. As a result, the overall security of the Bitcoin network actually improves.
Q: Can mining companies succeed in their AI transformation?
A: Success depends on several factors: the ability to migrate technology, the strength of capital support, and the competitive landscape. Cango’s strengths lie in its grid-connected global infrastructure and phased implementation roadmap. However, this transformation is a long-term endeavor, and continued monitoring of its technical execution and client acquisition progress is needed.
