Cango Inc. sold 4,451 BTC for roughly $305 million in USDT on February 9, 2026, using the liquidity to partially repay a Bitcoin-collateralized loan and reinforce its balance sheet. In the same breath, the miner signaled a strategic pivot: it plans to redeploy capital to convert parts of its mining footprint into GPU compute capacity for AI workloads.
Cango said the sale reduced its reported crypto holdings to about 3,000 BTC, and it framed the move as a deliberate reset as mining economics tighten after the halving and rising network difficulty. The company is effectively trading some BTC exposure for more balance-sheet flexibility while it retools its revenue model.
Deleveraging first, then reinvestment
Cango executed the BTC sale into a market already under downward pressure, receiving approximately $305 million in USDT. The immediate priority was deleveraging, with proceeds earmarked to partially repay a Bitcoin-backed loan and add operational liquidity as mining margins compressed. The end-state from this step is clear in the company’s own framing: a smaller BTC treasury position and a stronger liquidity buffer intended to extend runway.
The divestment also functions as a portfolio decision, not just a financing move. By shrinking holdings to roughly 3,000 BTC, Cango reduces direct mark-to-market sensitivity and shifts risk from price exposure toward execution on a new business line. That trade can stabilize near-term balance-sheet optics, but it also raises the bar on delivery for whatever comes next.
Cango’s “next” is an AI compute buildout that leans on existing infrastructure. The company plans to repurpose grid-connected mining sites into modular, containerized GPU compute nodes aimed at AI inference capacity, initially targeting small and medium enterprises. This is a classic adjacency play: reuse power, cooling, and site operations while changing the workload from hashpower to compute services. It also appointed Jack Jin as CTO for its AI business line, citing his experience managing multi-node GPU clusters at Zoom Communications.
Execution risk becomes the headline variable
Cango has not positioned this as a hard exit from mining. The company says it will keep mining operations, but shift emphasis toward operational efficiency and a blended hashrate-plus-compute model. In practical terms, that’s an attempt to diversify revenue sources while maintaining optionality if mining conditions improve.
The pivot comes with a candid constraint: execution will decide whether GPU compute becomes a durable cash-flow engine. KBW summarized the central risk bluntly, warning that “the path to monetization is fraught with execution challenges,” because technical deployment, client acquisition, and contract terms will determine whether the strategy produces stable recurring revenue. The financial backdrop adds urgency to that point, with the company described as trading near a 52-week low alongside negative free cash flow and a current ratio near 1.2.
What to watch next is less about the headline sale and more about operational proof. Early signals will come from phased GPU deployments at existing sites, CTO-led cluster integration, and evidence of real utilization—contract wins, sustained demand, and predictable service performance. At the same time, ongoing balance-sheet management will remain a parallel track, since the pivot’s credibility will be measured not only by buildout speed but by whether leverage and liquidity stay controlled while the new compute line ramps.
