Something shifted quietly in the American energy and crypto landscape this summer, and it feels bigger than just another company trimming operations. A firm that once poured serious capital into Bitcoin mining has now pulled the plug on every single United States site it operated. The decision came after a painful quarterly loss that climbed near sixty-five million dollars. Instead of waiting for better mining economics, management decided to repurpose the same power infrastructure for artificial intelligence and high-performance computing. I’ve been watching these pivots for a while now, and this one stands out because the miners are already offline while the new business has yet to produce a single dollar of data-center revenue.
Why Keel Walked Away From US Bitcoin Mining
The numbers told a blunt story. Second-quarter revenue dropped by half compared with the same period a year earlier, landing around thirty million dollars. Bitcoin mining revenue alone accounted for most of that decline. Lower average Bitcoin prices, rising network difficulty, and the deliberate shutdown of American operations all hit at once. Operating results swung from a modest profit the previous year to a loss exceeding one hundred forty million dollars. Non-cash charges related to retiring mining equipment added further weight.
What makes the move especially striking is the speed. Mining at the Washington State location stopped at the end of April. The remaining Pennsylvania sites—Panther Creek, Scrubgrass, and Sharon—went dark by the final day of June. Those facilities had once been core to the company’s North American footprint. Now the same electrical capacity is being redirected toward something far more capital intensive and, management hopes, far more profitable over the long term.
In my view, the decision reflects a broader reality that many listed miners have been quietly acknowledging. Pure Bitcoin mining has become a tougher business when hash rates keep climbing and price volatility refuses to cooperate. The power contracts and substations that once supported ASICs can, in theory, support denser and higher-value compute loads. Whether that theory turns into contracted revenue remains the open question.
The Scale of the US Shutdown
Washington’s site began converting into an eighteen-megawatt high-performance computing facility almost as soon as the miners stopped. In Pennsylvania the picture is larger. Panther Creek and Scrubgrass retained roughly sixty and sixty-three megawatts of energized capacity that was no longer under mining contracts. Sharon is being prepared for a planned one-hundred-ten-megawatt data center. Those are not small numbers. They represent meaningful chunks of grid capacity that the company now wants to sell under long-term AI and HPC agreements rather than short-term mining economics.
Interestingly, some of that power continued to be sold into the wholesale market after the machines were switched off. That temporary bridge keeps a modest cash flow alive while the longer conversion work continues. Still, wholesale power sales are a far cry from the contracted, multi-year revenue streams that data-center operators prefer.
Corporate Identity Shift and Capital Moves
The operational change arrived alongside a deeper corporate redesign. The company completed its redomiciliation from Canada to the United States earlier in the year, becoming a Delaware corporation and a domestic issuer. Its shares now trade under a new ticker that reflects the rebranded identity. That legal and structural shift appears designed to align the firm more closely with American capital markets and potential US data-center customers.
On the balance-sheet side, management has been actively reducing its Bitcoin holdings. Between early April and early August the company sold more than one thousand Bitcoin, generating roughly seventy-five million dollars in proceeds. The remaining treasury stood at just under nineteen hundred coins, valued around one hundred twenty-one million dollars for liquidity purposes. Leadership has stated a clear intention to liquidate the entire Bitcoin position before the end of the year. That plan is not yet finished, so the final proceeds still depend on future market conditions and execution timing.
Liquidity overall looks solid. Unrestricted cash plus the remaining unencumbered Bitcoin brought total liquidity near eight hundred nineteen million dollars by early August. That figure represented a meaningful increase from the previous quarter, helped in part by a large convertible-note issuance. The company raised four hundred fifty-eight million dollars through notes carrying a low coupon and due in the early 2030s. Management has indicated that a portion of those proceeds will support additional power capacity and site preparation at the Pennsylvania locations.
Financial Reality of the Transition
The second-quarter results make the cost of the pivot impossible to ignore. Revenue from continuing operations fell sharply. The United States contribution dropped from more than half of total revenue a year earlier to just over one-third. That geographic shift alone removed nearly twenty million dollars of revenue. Higher general and administrative expenses also appeared, driven by stock-based compensation, professional fees tied to the redomiciliation, and new hiring needed for data-center development work.
Depreciation charges linked to the retirement of mining equipment added further pressure. When you retire specialized hardware early, the accounting hit can be substantial even if the cash impact is limited. The net result was a loss from continuing operations near sixty-four million dollars. These figures are not abstract. They show that the company is deliberately accepting short-term pain in exchange for what it hopes will be a more durable business model.
I’ve found that markets tend to give these pivots a grace period only if management can demonstrate tangible progress on the new front. Right now the new front is still mostly potential. As of the latest reporting date, none of the US sites had begun high-performance computing operations or recognized any related revenue. That fact keeps the story in development mode rather than operating mode.
What the AI and HPC Opportunity Actually Looks Like
The broader industry context helps explain the urgency. Demand for dense, reliable power near major fiber routes has surged as artificial-intelligence training and inference workloads scale. Companies that already control large power positions and have experience managing high-density electrical infrastructure sit in an interesting position. Converting a mining site is not as simple as swapping machines, of course. Cooling systems, power distribution, redundancy, and customer interconnection requirements all differ from the relatively straightforward setup of ASIC farms.
Keel has reported that commercial discussions are underway with multiple prospective tenants for each of its three priority sites. Those talks remain private. No named customers or signed long-term leases have been announced. Until contracts are executed, the revenue potential stays theoretical. Permitting also remains incomplete. Zoning and land-development approvals have been secured for certain Pennsylvania sites, yet environmental permits are still in process. One site has secured three hundred fifty megawatts of utility capacity, with management pointing to 2027 as the earliest realistic ready-for-service window for the larger build-out.
That timeline matters. Capital raised today needs to generate returns in a competitive environment where other former miners and traditional data-center developers are chasing the same AI demand. Speed of execution, quality of interconnection, and ability to offer competitive power pricing will separate the winners from the rest.
Canadian Operations Still Running
The US exit does not mean a complete departure from Bitcoin mining. Legacy Canadian assets continue to operate. The company is also exploring further conversions in Canada, including a proposed ninety-six-megawatt campus in Quebec. Local approvals for power transfer have been obtained, although provincial review of the shift toward high-performance computing is still required. Maintaining some mining cash flow while the larger US transformation unfolds provides a useful bridge.
This dual approach—shutting down the higher-cost or strategically more valuable US sites while keeping lower-cost Canadian mining alive—feels pragmatic. It reduces immediate cash burn from unprofitable US mining without abandoning the entire skill set and operational knowledge the company built over years.
Industry Parallel Moves and Competitive Pressure
Keel is far from alone. Across the listed mining sector, several operators have redirected capital and electrical capacity toward AI hosting as mining margins compressed during the past year. Some have signed early hosting agreements. Others remain in the marketing and permitting phase. The common thread is recognition that the same physical assets—substations, transformers, cooling infrastructure, and grid interconnections—can support workloads that command higher and more stable pricing than Bitcoin mining currently offers.
Yet the competitive landscape is intensifying. Traditional data-center operators with deeper balance sheets and established customer relationships are also expanding aggressively. Hyperscalers continue to secure their own power positions. Success for a former miner therefore depends on more than simply owning megawatts. It requires credible development timelines, proven operational reliability, and the ability to structure deals that satisfy sophisticated enterprise customers.
In my experience following these transitions, the companies that treat the pivot as a full operational and cultural change tend to fare better than those that simply rebrand the same mining culture under an AI label. Hiring experienced data-center talent, adjusting maintenance and reliability standards, and building genuine sales capabilities all take time and money. The increase in general and administrative expenses already visible in Keel’s results suggests management understands at least part of that requirement.
Key Risks That Still Hang Over the Story
Several risks deserve clear attention. First, construction and interconnection delays are common in this sector. Even with secured utility capacity, the path from approval to energized, customer-ready space can stretch longer than initial projections. Second, customer concentration risk remains high until a diversified set of leases is signed. A single large tenant can provide welcome revenue visibility, but it also creates dependency. Third, the remaining Bitcoin treasury, while still substantial, is intended for liquidation. That means future cash inflows from coin sales will depend on market prices at the time of sale.
Power-market dynamics also matter. Sites that can sell excess power into wholesale markets enjoy a temporary cushion, yet wholesale prices fluctuate. Long-term contracted revenue from data-center tenants offers greater predictability, which is why management is prioritizing those deals.
Perhaps the most interesting aspect is the valuation question. Markets will eventually need to decide whether to value the company as a residual Bitcoin miner with a side data-center project or as an emerging infrastructure provider that happens to still run some mining. That shift in perception usually requires visible contracted revenue and a credible pipeline of additional capacity.
What Success Would Actually Look Like
For the pivot to be judged successful, several milestones need to arrive in sequence. Environmental and remaining construction permits must clear. At least one meaningful customer lease should be announced and begin generating revenue. Construction progress at the priority sites needs to stay on a publicly credible schedule. And the company must demonstrate that its cost structure for developing and operating data-center space is competitive.
If those pieces fall into place, the former mining sites could become valuable AI infrastructure assets. The same electrical capacity that once supported Bitcoin hashing would instead support training clusters or inference workloads. The capital structure, already strengthened by the convertible issuance and Bitcoin sales, would then support further expansion rather than merely funding the transition.
If progress stalls, the company would face a less attractive set of options: restart mining at higher difficulty and potentially lower margins, sell power wholesale indefinitely, or seek partners who can accelerate the data-center build-out. None of those alternatives look as attractive as a successful AI conversion, which is why management appears focused on the harder path.
Broader Lessons for the Mining Sector
Watching this transition closely offers a few lessons that extend beyond one company. Power is the scarce resource. Firms that control large, well-located electrical capacity hold an asset that multiple industries now want. The ability to redeploy that capacity quickly and credibly is becoming a competitive advantage. Balance-sheet flexibility also matters. Selling Bitcoin at scale and raising long-term capital gave Keel room to absorb the short-term earnings hit without immediate distress.
Cultural and operational adaptability may prove equally important. Mining cultures tend to optimize for uptime of relatively standardized hardware and rapid deployment of new machines. Data-center cultures optimize for extreme reliability, complex customer requirements, and multi-year planning cycles. Bridging those mindsets is not automatic.
I’ve noticed that the most convincing transitions so far combine three elements: clear public milestones, transparent discussion of remaining risks, and a willingness to spend real money on the talent and processes required by the new business. Companies that simply announce a pivot without those supporting actions tend to face more skepticism.
Looking Ahead Through the Rest of the Year
The coming months will test whether the commercial talks turn into signed contracts and whether permitting progresses on schedule. Investors and industry observers will watch closely for any announcement of a first tenant or a firm ready-for-service date that moves earlier than the current 2027 expectation for larger capacity. Continued Bitcoin sales will also shape the cash position and reduce the company’s exposure to crypto-price swings.
Meanwhile the Canadian mining operations provide a residual earnings base and keep operational knowledge alive. That dual structure may prove useful if AI demand softens or if conversion timelines stretch. Flexibility remains valuable in a sector still finding its equilibrium between traditional mining and newer compute workloads.
The decision to shut down US mining was decisive. The harder work of turning those sites into reliable, contracted AI infrastructure is only beginning. How that next chapter unfolds will determine whether this particular pivot becomes a model for others or a cautionary tale about the gap between strategic intention and operational reality.
One thing feels certain. The era when large public miners could rely solely on Bitcoin price appreciation and relatively simple mining economics is fading. Those that control meaningful power positions now face a choice: keep mining through thinner margins or attempt the more complex, higher-upside conversion to AI and high-performance computing. Keel has chosen the second path for its American assets. The rest of the industry is watching to see whether the economics ultimately justify the disruption.
For anyone following the intersection of energy, crypto, and artificial intelligence, the story offers a live case study in real time. The miners are already off. The power is still there. The customers are still negotiating. And the calendar is moving. The next set of updates will reveal whether the transition can close the gap between current losses and future contracted revenue.
Until those contracts materialize and the first data-center megawatts start generating recognized revenue, the pivot remains a high-stakes development story rather than a completed transformation. That tension—between decisive operational change and still-uncertain commercial outcomes—is what makes the current moment worth following closely.
