Genesis Digital Assets Shifts From Mining to AI
SWI Group is turning an FTX-linked mining business toward AI and data centers. See what the $900 million raise means—and what still needs to change.

From FTX-Linked Bitcoin Mining to AI: What SWI Group's Deal Means
A crypto mining operation can look a lot like a data center from the outside: both run on heavy electricity use, specialized hardware, and uptime-focused infrastructure. But that resemblance doesn't mean they're interchangeable. Bitcoin mining rigs aren't a plug-and-play replacement for the systems that modern AI services rely on.
That distinction matters now that SWI Group is trying to reposition Genesis Digital Assets, a mining business tied to FTX, as part of a wider AI and data center push. SWI says it has raised about €800 million - roughly $900 million - to expand computing capacity in Europe and the United States. It also reported that it held a 70% stake in Genesis as of August, with the mining operation to be renamed SWI Digital.
The pitch brings together two industries competing for the same scarce inputs - power and infrastructure - while also linking the plan to one of the more consequential corporate failures in crypto. Here's what's been reported, what the shift could actually involve, and what investors should verify before assuming "mining capacity" can just become "AI-ready infrastructure."
What SWI Group has acquired
SWI Group describes itself as an investment company focused on scaling AI and data center capacity. In its announcement, it said it raised €800 million (about $900 million) to expand in Europe and the United States. The company frames Genesis as part of a broader computing infrastructure strategy, not just a standalone bet on crypto mining.
The transaction details have moved in stages. In June, SWI said it acquired a significant shareholding in Genesis for $500 million. At that time, it reported that Genesis ran 15 facilities with 1.3 gigawatts of capacity. Later, SWI said its stake had reached 70% as of August and that the mining business would be renamed SWI Digital.
Those numbers cover different parts of the story, so they shouldn't be merged into one simple claim. The $500 million figure relates to the June acquisition announcement. The €800 million raise is a separate financing step for expansion. And the 1.3 GW number is capacity SWI attributed to Genesis's facilities - it doesn't, by itself, tell you how much power is usable today, how much is contracted, or what portion could realistically support AI workloads.
Why the ownership percentages matter
A 70% stake generally suggests majority control, based on SWI's reported position. But majority control doesn't automatically mean every asset is owned outright or that all legal and operational questions are settled. Minority investors can still retain economic rights, and facilities can be tied up in separate contracts, financing structures, or local regulatory approvals.
If you're trying to evaluate the deal, the questions that actually drive outcomes are practical: which sites are included, what the capital raise will pay for, how much capacity is running now, and what additional investment is required before any of these sites can serve AI customers. Until those details are clear, the headline figures are just that - starting points, not proof that a complete conversion has already happened.
Why mining sites appeal to AI operators
There's a straightforward reason AI-focused infrastructure investors look at crypto mining assets. Mining operators spend years securing locations, arranging power supply, and building facilities designed to keep computing equipment running continuously. In a world where electricity access and construction timelines can slow down data center growth, those capabilities can matter.
AI demand is also rising, and that can make a prepared site look attractive. If a location has the right grid connection, available land, workable cooling, and a smoother path through local permitting, it can start closer to "ready" than building a new site from scratch. Utilities relationships and operating experience around the clock may help too - even if the facility still needs substantial upgrades.
That's the basic infrastructure logic behind a potential transition: reuse what can be reused, upgrade what needs to change, and direct capital toward a market that may work on different customer economics than mining. SWI's stated expansion across Europe and the United States suggests it wants to build a wider network, not just make one-off changes. Still, SWI's capital raise doesn't confirm that every Genesis location will be converted - or that any conversion is already complete.
The underlying business model is changing
Bitcoin mining revenue depends on the economics of the Bitcoin network. Operators compete to process blocks, earn mining rewards, and collect transaction fees. Returns depend on factors like Bitcoin's price, mining difficulty, hardware efficiency, and electricity costs. Those drivers can shift quickly, which is why cheap power and efficient operations are central to mining.
AI data centers typically sell or lease computing capacity to customers running demanding workloads. Revenue depends more on equipment uptime and availability, contract terms, utilization, service quality, and the cost of power and cooling. The customers and the commercial relationships are different from mining.
The overlap isn't the business model itself - it's the need to run large amounts of computing equipment with reliable power. Mining sites can be strategically useful, but only if the physical setup and commercial requirements of AI workloads can be met.
Why a mining facility is not automatically an AI data center
"Repurposing mining capacity for AI" can sound simple, like the work is mainly a matter of rebranding. In reality, it helps to separate site reuse from computing equipment reuse. A mining facility's building and infrastructure might provide a foundation, but the machines inside it are a different story.
Bitcoin mining typically uses specialized ASIC hardware built for a narrow job. AI services often depend on accelerators such as GPUs, along with the surrounding systems for networking, storage, and workload management. Mining hardware generally can't do the same work as an AI GPU cluster. So, an operator might reuse a site - and some supporting infrastructure - without keeping the exact equipment that made mining profitable.
Even the building requirements can diverge. High-performance AI hardware can impose tight demands for power density, heat removal, network performance, and redundancy. A facility designed around one pattern of power use and cooling might still need engineering work to support another. The cost and timeline will vary by site - there isn't one universal "conversion budget" that fits all 15 facilities.
Four checks that determine whether a site can make the jump
Before calling a mining facility AI-ready, operators (and investors) should be able to answer a few basic questions:
- Power availability: Is the electricity supply reliable and sufficient for the planned load? A reported capacity number isn't the same as usable, contracted power at each site.
- Cooling and building design: Can the facility handle the heat from the intended hardware, and what upgrades are required? This affects both capital costs and day-to-day operating efficiency.
- Network and equipment: Can the site support the accelerators, connectivity, storage, and monitoring needed for AI workloads? Mining infrastructure doesn't automatically include these capabilities.
- Customers and economics: Is there a believable route to contracted demand, and do expected revenues cover conversion and operating costs? Building capacity without secured demand can leave expensive assets underused.
These checks aren't meant to kill the idea. They're how you tell whether this is a real infrastructure opportunity - or just a headline.
The trade-off: speed versus fit
Turning an existing site into an AI facility can be faster than starting from scratch. But speed only helps if the end result actually fits customer requirements. Keeping mining running while developing AI capacity can also preserve flexibility. The trade-off is that the company has to operate two different businesses, each with its own equipment cycles, customers, and operating risks.
A careful rollout would evaluate facilities one by one, rather than assuming a single plan works across the portfolio. Some sites might fit AI workloads well. Others may stay closer to mining or require too much investment to convert. The decision will come down to power costs, grid constraints, local rules, equipment costs, and demand - not just the fact that miners previously operated there.
The FTX connection and the legal context
Genesis Digital Assets is part of the FTX story because the failed exchange had a stake in the company. One report says Sam Bankman-Fried, FTX's former CEO, directed Alameda Research to spend more than $1 billion to acquire a 20% stake in Genesis between 2021 and 2022. FTX filed for bankruptcy in November 2022, and the use of customer funds became central to criminal cases involving former executives.
Because of that background, the subsequent sale and ownership changes come with more than typical deal risk. FTX's bankruptcy estate and creditors have had strong incentives to review assets and transactions connected to the former exchange, including whether value can be recovered for creditors. If you're buying assets linked to a major bankruptcy, you want clarity on title, outstanding claims, and the status of relevant disputes.
SWI co-founder and CEO Max-Hervé George reportedly said legal disputes tied to FTX's stake in Genesis had been resolved. That's a helpful signal from management, but it's still just a reported position - not a substitute for reading court filings, deal documents, and any applicable legal rulings. The report doesn't lay out the resolution in detail or describe the terms.
Why this history still matters
The FTX connection doesn't automatically mean SWI's strategy will fail. It also doesn't erase questions about how the earlier stake was funded. The key point is that asset value and legal certainty are separate issues. A facility can have strong power access and still face problems in ownership, contracts, or claims that affect how - and whether - it can operate.
For customers and investors, due diligence should cover both layers:
- Operational diligence: Which sites, equipment, power arrangements, and upgrade plans are actually included?
- Legal diligence: What claims, transfer conditions, or unresolved proceedings could affect ownership or day-to-day use?
This matters even more when a transaction involves assets tied to a high-profile bankruptcy. A confident public statement can be useful context, but investors will still want the underlying documentation before concluding that residual risk is gone.
What to watch next
The core question isn't whether AI demand is increasing or whether mining companies have large facilities. The question is whether SWI can turn specific assets into reliable, contracted computing capacity at a cost that makes commercial sense.
A few steps would make the plan easier to evaluate. First, more detail on how the €800 million raise will be spent would help distinguish upgrades from hardware purchases, acquisitions, or other expansion costs. Second, site-level information would clarify how much capacity could be made available for AI and on what timeline. Third, customer contracts or other proof of demand would show whether this is heading toward utilization - or just pitching a market opportunity.
Investors should also look for operational milestones, not only announcements. Examples include completed facility upgrades, installed AI hardware, confirmed power availability, and disclosed customer deployments. Each milestone reduces a different kind of uncertainty. A financing announcement signals access to capital, but it doesn't prove the facilities are converted, customers are signed, or returns will work.
A practical way to assess the announcement
When you're evaluating a pivot like this, it helps to follow a simple sequence:
- Separate reported facts from plans. The reported stake, facility count, and financing don't equal completed AI capacity.
- Confirm what's actually being reused. A site and power connection may transfer more easily than mining machines.
- Demand cost and timeline detail. Conversion needs vary by location, so portfolio-level capacity figures aren't enough.
- Look for demand evidence from customers. AI infrastructure needs utilization and reliability, not just available floor space.
- Track legal clarity. Any reported resolution of FTX-related disputes should be checked against primary documentation.
This approach applies beyond Genesis. As miners explore AI and data centers, the same questions will decide whether the pivot becomes a durable business change - or just a new label for old assets.
Conclusion
SWI Group's move makes sense if you're looking purely at infrastructure. Mining operators can control sites, power access, and operational know-how that would take time to build elsewhere. But a Bitcoin mining facility isn't automatically an AI data center, and a large capital raise isn't proof that conversion is already done.
The right way to evaluate it is asset by asset. Focus on power, upgrades, equipment, customers, timelines, and legal status. If SWI can convert suitable facilities into reliable, utilized AI capacity, Genesis's mining footprint could become the base for a different business. Until then, it's better to treat the deal as a major infrastructure plan - not a finished transformation.
For more context, read Cointelegraph's report on the FTX-linked mining venture and compare future announcements against the operational milestones outlined above.