📰 Key Takeaways

Before FTX’s bankruptcy, Alameda Research spent over $1 billion to acquire a roughly 20% stake in Genesis Digital Assets, and now these crypto mining assets have been repurposed for AI compute. Investment firm SWI Group announced it completed roughly 800 million euros (about $900 million) in funding to expand its data center and AI compute capacity in Europe and the US, which includes the Genesis Digital Assets mining business once tied to FTX. SWI Group had already acquired a “significant stake” in Genesis for $500 million this past June, when Genesis owned 15 facilities with a combined 1.3GW of compute capacity. By August, SWI’s stake had grown to 70%, leading it to rename the mining company SWI Digital and fold it into its data center compute network. The Genesis acquisition took place while FTX’s bankruptcy estate was actively pursuing asset recovery — former FTX CEO Sam Bankman-Fried reportedly directed sister company Alameda Research to spend over $1 billion acquiring the 20% stake in Genesis between 2021 and 2022. FTX filed for bankruptcy in November 2022, leading to criminal charges against several executives for misappropriating customer funds. SWI co-founder and CEO Max-Hervé George said all legal disputes related to the FTX stake have now been fully resolved.

💬 JudyAI Lab Take

The crypto mining assets left over from FTX’s bankruptcy liquidation have now transformed into AI compute hubs — and that handoff itself is a signal worth watching in the industry.

SWI Group just closed nearly 800 million euros in funding to expand its data center and AI compute capacity across Europe and the US, folding in the Genesis Digital Assets mining assets that FTX affiliates once acquired through Alameda Research for over $1 billion. In just a few short years, 15 facilities originally built for crypto mining — with a combined 1.3GW of compute capacity — have turned into AI compute bases. That reflects a clear trend: compute infrastructure’s purpose is being redefined. As long as the power and cooling are in place, mining hardware can quickly be repurposed for AI workloads instead of being written off as a sunk cost. For AI builders, this is a reminder that infrastructure value isn’t locked to its original design purpose — it’s about whether it can pivot when demand shifts.

Worth considering: when evaluating the compute costs of an AI project, it might be worth keeping an eye on this kind of “repurposed” data center and whether it can offer a more cost-effective compute source than building new facilities from scratch.

📅 Source Info

🔗 Further Reading