Cango’s EcoHash Starts Commercial GPU Compute at Georgia Site

Cango said on September 3 that its EcoHash subsidiary had started commercial GPU computing at the Bitcoin miner’s Georgia facility. According to the company’s announcement, EcoHash completed modifications for a dedicated artificial-intelligence compute section and powered on an initial batch of servers. The update marks an operating launch, although Cango did not identify the first customer or disclose revenue from the service.

Georgia Site Adds a 3-Megawatt AI Section

The dedicated section sits within Cango’s owned 50-megawatt facility in Georgia and currently supports up to 3 megawatts of AI-compute capacity. Cango said the layout leaves room for expansion, but it did not provide a timetable or committed capacity for a later phase.

EcoHash installed high-density compute containers supported by an industrial transformer setup, precision cooling and fiber connectivity. The company described each container as a standardized module designed to sustain dense GPU workloads. That architecture is intended to make deployments repeatable across suitable sites, though any expansion beyond Georgia remains a company plan rather than completed capacity.

Commercial GPU Services Begin With Initial Servers

EcoHash has begun providing paid GPU-compute services using part of the first server batch, while additional machines are still being deployed. Cango said the initial customer and prospective customers include GPU-cloud platforms and AI-focused cloud-infrastructure providers, without naming the organizations or stating the number of servers operating.

The distinction matters because the announcement confirms that commercial service has begun, but does not establish its scale or profitability. Cango also did not report utilization, contract length, pricing or expected contribution to group revenue. Those omissions leave the financial significance of the launch to be tested in later operating results.

Cango Reuses Bitcoin-Mining Infrastructure

Cango said placing the modules beside existing mining equipment reduced the new construction and retrofit work needed to reach power-on. The project therefore provides a concrete example of a Bitcoin miner repurposing part of its energy and site infrastructure for AI workloads rather than building a separate data center from the ground up.

The launch follows Cango’s broader shift toward energy and computing services. BlockchainReporter’s coverage of Cango’s second-quarter update examined the miner’s reduced active hash rate and its planned AI expansion. The Georgia deployment moves that strategy from preparation into initial commercial operation, but Bitcoin mining remains part of the company’s global business.

Replication Is Still a Forward-Looking Goal

Cango called the Georgia installation a proof-of-concept hub for a model it wants to replicate at future sites. That goal is not the same as an announced rollout: the release named no additional location, customer commitment or capital budget. For now, the dated development is limited to completed modifications, the first powered GPU servers and the start of commercial service at the Georgia facility.