The data shows a 50% decline in Q2 revenue for miner Keel, followed by the closure of its US Bitcoin mining operations. That is not a footnote. It is a ledger entry in the industry's structural shift. The ledger never lies, only the interpreter does. And the interpreter here sees a clear pattern: the post-halving hash price crunch is forcing a binary choice for miners—either shut down or pivot to AI/HPC infrastructure. Keel chose the latter, but the data reveals more than just a press release.
Context: The Halving Math That Broke the Model
To understand Keel's move, we need to audit the basics. Bitcoin's fourth halving in April 2024 cut block rewards from 6.25 BTC to 3.125 BTC. For a miner operating at the industry average efficiency of 30 J/TH, the hash price—the daily revenue per TH/s—dropped from ~$0.08 before the halving to ~$0.04 by Q2 2024, according to Hashrate Index data. At the same time, the cost of electricity for US miners in states like Texas and New York ranges from $0.04 to $0.08 per kWh. Simple math: for many miners, the marginal cost of mining exceeded the marginal revenue. Keel's Q2 revenue decline of 50% is not an outlier; it is the mean.
Keel's announcement of completing the shutdown of US Bitcoin mining and pivoting to AI/HPC infrastructure is a strategic response to this arithmetic. But the data on the transition is sparse. The company did not disclose GPU models, cluster sizes, or cooling solutions. This is a red flag. In my 2020 DeFi yield farming quantification project, I learned that the absence of technical details often signals a gap between narrative and capability. When I analyzed Liquity’s stability pool, the data showed a clear path to solvency only if certain token ratios were maintained. Keel’s lack of specificity suggests they are still in the early stages of planning, not execution.
Core: The On-Chain Evidence Chain of Miner Distress
Let’s follow the data. The first signal is the hash price. Post-halving, the hash price has been oscillating between $0.04 and $0.05 per TH/s, as tracked by CoinWarz. Meanwhile, the total network hashrate has remained above 600 EH/s, indicating that only the most efficient miners are surviving. The second signal is the secondary market for ASIC miners. Prices for the Bitmain S19 series have dropped from $20/TH to below $10/TH in 2024, per mining hardware brokers. This is a classic sign of a supply glut—miners are offloading equipment at a loss. The third signal is the migration of miners to AI. Core Scientific, for example, secured a 35-year contract with CoreWeave worth $3.5 billion to host AI computing. That is a real revenue stream, not a pivot in name only.
Keel’s situation is different. The company is small, likely with less than 5 EH/s of hashrate. Its US shutdown means it is abandoning its primary asset: the ASIC fleet. But the value of its power capacity contracts—likely in the range of 50-100 MW—remains. That is the real asset for AI. In the 2024 ETF approval flow analysis, I quantified how institutional investors value power capacity over mining hardware. The data shows that power contracts with fixed pricing and long duration are the new scarcity. Keel is betting that its power capacity can be repurposed for AI, but it faces competition from established data center operators like Equinix and hyperscalers like AWS.
Contrarian: The Pivot Is a Necessity, Not a Strategy
The market narrative is that miners are “pivoting to AI” as a visionary move. The data says otherwise. Correlation is not causation. The pivot is a survival mechanism, not a strategic choice. The real driver is the collapse of mining revenue, not a sudden epiphany about AI. Consider the evidence: the number of public mining companies pivoting to AI has increased from 3 in 2023 to over 15 in 2025, but the success rate in securing long-term AI contracts is low. Only Core Scientific and Hut 8 have announced multi-year agreements. The rest are in the “exploring” phase, which is code for “we have not signed anything.”
Furthermore, the AI infrastructure market is not a blank check. AI workloads require specialized cooling, high-density power, and robust networking. Miners have the power, but they lack the technical expertise. The data from my 2025 AI-Agent On-Chain Interaction project showed that the transaction patterns of AI agents are distinct, but the hardware requirements are even more distinct. A mining facility designed for 30 kW per rack for ASICs cannot handle the 100 kW per rack needed for Nvidia H100 clusters without major retrofitting. Keel’s announcement did not mention any capital expenditure plan for this. That is a risk.
Takeaway: The Signal to Watch Next Week
The next signal is the hash price. If it stays below $0.05/TH/s for another three months, expect more miners to follow Keel’s path. The secondary market for ASICs will crash further, and the Bitcoin network’s security will see a temporary dip. But the bigger question is whether Keel’s pivot will succeed. The data suggests that the company’s power contracts are the only asset with real value. Without a binding AI client contract, the pivot is just a story. Volatility is the tax on uncertainty. Investors should watch for any announcement of a partnership or a capital raise. Until then, the ledger shows a miner in distress, not a rebirth.
Every transaction leaves a shadow in the block. Keel’s shadow is one of a forced exit, not a strategic pivot. The data is clear: the mining industry is undergoing a structural shift, and not all survivors will be miners. The next chapter belongs to those who can quantify the chaos and reveal the pattern. And the pattern says: power capacity is the new hashrate, but only if you can make it speak to AI clients.