Bitfarms Transitions Washington Site to AI Infrastructure, Targeting December 2026 Completion
Bitfarms (NASDAQ: BITF) announced plans to convert its 18 MW Bitcoin-mining facility in Washington State into a high-performance computing (HPC) and AI workload site, targeting full deployment by December 2026. The company has entered a fully funded, binding US $128 million agreement with a major U.S. infrastructure provider to supply all IT equipment and building materials necessary for the 18 MW conversion.
The upgraded facility will support advanced Nvidia GB300 GPUs with up to 190 kW per rack, feature modular infrastructure for phased deployment, and operate with an anticipated Power Usage Effectiveness (PUE) between 1.2 and 1.3—well-above industry benchmarks for efficient data centers.
Bitfarms’ CEO, Ben Gagnon, emphasized this move as part of the company’s “HPC/AI infrastructure development strategy,” stating:
“We continue executing on our HPC/AI infrastructure development strategy with a fully funded supply chain and plan to convert our Washington site to support Nvidia GB300s with state-of-the-art liquid cooling.”
He added that, despite the site representing “less than 1% of our total developable portfolio,” the conversion “could potentially produce more net operating income than we have ever generated with Bitcoin mining.”
From an investor perspective, this pivot marks a strategic shift: Bitfarms is repositioning from pure cryptocurrency mining toward AI-driven infrastructure—leveraging its energy assets and operational footprint to capture stable, long-duration cash flows from colocation and GPU-as-a-service models. For shareholders, key catalysts to watch will include the pace of modular build-out, sign-up of colocation contracts, leverage of Nvidia-class GPU inventory, and ramp-up of non-mining revenue streams.
Still, execution risk remains. Transitioning from mining to HPC/AI hosting involves new customer segments, service models, and competitive dynamics. Investors should monitor equipment delivery timelines, PUE performance validation, and the broader market for AI-data-centre real-estate and GPU infrastructure.
