Cango Plans Shift From Bitcoin Mining to AI Infrastructure, Citing Global Power and Data Center Demand
Dear Shareholders,
The year 2025 marked a pivotal chapter in our strategic transformation, representing a new phase of entrepreneurship for Cango. As we enter 2026, we believe it is appropriate to reflect on our progress and sharpen our focus for the opportunities ahead.
Over the past year, we have executed a disciplined entry into the industry, balancing speed with operational prudence to steadily build our position as a leading Bitcoin miner with a global footprint across four key regions.Importantly, we have delivered on several key commitments: acquiring and enhancing hashrate efficiency of 50 EH/s of on-rack machines, initially adopting a strategic treasury approach, divesting legacy operations, securing an initial 50 MW of energy infrastructure, and completing our transition to a direct NYSE listing. These steps have set a foundation for the future transition from hosted hashpower toward a global distributed inference compute grid.
In response to recent market conditions, we have made a treasury adjustment to strengthen balance sheet and reduce financial leverage, which provides increased capacity to fund our strategic expansion into AI compute infrastructure.
The Strategic Logic behind Our Proposed Pivot
As previously communicated, our global mining operations, operational experience, and infrastructure provide a practical pathway toward our AI compute objectives. The growing AI era continues to face a "Power Gap"—a disconnect between rising compute demand and existing grid capacity. By leveraging our globally accessed, grid-connected infrastructure, Cango is well-positioned to participate in this opportunity, and our business model will revolve around leveraging and connecting these diversified power sources to flexibly supply high-performance compute capacity to meet long-tail inference demand through a scalable business model.
This transition is expected to follow a disciplined three-phase roadmap:
Near Term: Standardization and efficient deployment. We plan to enter the market with modular, containerized GPU compute nodes designed as plug-and-play solutions for rapid deployment across our globally accessed infrastructure. Our focus will be on offering on-demand compute capacity by drawing on power from our current operations. Conversion to AI-ready infrastructure is designed to require limited upgrades, with sites expected to be operational in a relatively short timeframe, creating the potential for progressively developing additional revenue streams. While hyperscalers may encounter extended grid-connection timelines and substantial capital needs, our approach is relatively asset light and allows us to efficiently address the 'long-tail' of AI demand.
Medium Term: Software-defined orchestration. To complement our physical nodes, we intend to deploy a proprietary software platform to manage and integrate our distributed compute capacity. This architecture is expected to help us evolve into an ecosystem enabler, offering our global footprint as an integrated, enterprise-grade network without the typical infrastructure complexity.
Long Term: Global scaling and platform resilience. Our longer-term vision is to grow this network into a mature global AI infrastructure platform. By activating underutilized power across our mining ecosystem and eventually beyond, we aim to establish recurring revenue streams from platform services and compute agreements designed to be durable across market cycles.
We believe our infrastructure, partnerships, and operational resources offer a reasoned basis for this approach. As part of accelerating this transition, we have established EcoHash Technology LLC, a wholly-owned subsidiary based in Dallas, Texas, to focus on advancing our AI compute initiatives. Under the leadership of our newly appointed AI CTO, we are assembling a dedicated team to guide technical execution.
Execution of our roadmap is being conducted with clear financial and technical thresholds. Our initial rollout is intended to leverage our existing network to deliver scalable AI inference capacity—meeting the substantial 'long-tail' AI demand from SMEs at competitive price points.
Empowering the Ecosystem: From Miner to Enabler
Cango's transformation is also aimed at creating broader ecosystem value. We view ourselves as an "Ecosystem Enabler," recognizing that the Bitcoin mining industry holds significant underutilized energy infrastructure suitable for AI era needs. We observe that many in the mining community share an interest in this opportunity but may lack a clear technical path forward.
By advancing in this direction, Cango hopes to support the wider mining ecosystem's evolution. Our architecture is designed to offer a flexible growth path—enabling accessible, on-demand compute with manageable upfront commitment. It also provides a practical route for operators to adapt existing facilities for AI operations with relatively short lead times.
A Vision for Value Creation in the Age of AI
We are steadily evolving toward a technology-driven infrastructure platform. This shift is expected to open new recurring revenue streams—, building a business model that can endure across cycles while complementing our mining operations.
We remain mindful that this is a long-term journey. The transition from mining to AI compute will continue to develop over multiple years, and we are still in the early stages. While the path ahead requires sustained effort, our roadmap is clear, and we are committed to prudent, step-wise execution.
Thank you for your continued trust and support. We are focused on the opportunities ahead and look forward to updating you on our progress.
Paul Yu
Chief Executive Officer, Cango Inc.
About Cango Inc.