The Divergence That Demands Attention
The data is unambiguous. Bitcoin's seven-day average hashrate has fallen 20.6% from its all-time high of 1,151 EH/s in August 2025 to approximately 914 EH/s by late September. This is not a routine fluctuation. This is the second-largest hashrate drawdown in Bitcoin's history, and it is happening while the price has climbed 34.9% between June and August. Price up. Hashrate down. That divergence has occurred only once before in Bitcoin's existence.
The last time this happened was 2012. The market structure then was fundamentally different. Today, we are witnessing something unprecedented: miners are not capitulating because of falling Bitcoin prices. They are leaving because artificial intelligence companies are offering them better economics for their most valuable assets—power, infrastructure, and operational expertise.
I have been tracking miner behavior since the 2017 ICO era, and I can tell you this: the current dynamic is not a typical miner capitulation cycle. The self-healing mechanism that has historically rescued Bitcoin's security budget—difficulty adjustment leading to improved profitability leading to hashrate return—is partially broken. The reason is locked-in AI contracts that make the economics of returning to Bitcoin mining unattractive for years.
The Structural Shift in Miner Economics
Let me break down the numbers with precision. The Puell Multiple currently sits at 0.73, placing it in the 16th percentile of historical readings. This metric, which measures the ratio of daily Bitcoin issuance value to its one-year moving average, indicates that miners' dollar-denominated revenue conditions remain at historically depressed levels. The Hashprice—the expected dollar value of 1 PH/s of hashrate per day—stands at $39.36, which is actually above its 30-day average. This short-term improvement suggests that the difficulty adjustment mechanism is functioning as designed.
But here is the critical insight that most market observers miss: despite the improvement in both price and difficulty-adjusted profitability, hashrate has not recovered. The block production time is running at approximately 9 minutes 56 seconds, slightly faster than the 10-minute target, confirming that the network is operating efficiently. The difficulty adjustment algorithm is working. The problem is that the miners who left are not coming back.
The reason is straightforward. When a miner signs a 20-year power purchase agreement with an AI hyperscaler like Anthropic—as Riot Platforms did—that electricity is contractually committed. It cannot be redirected to Bitcoin mining without breaching contractual obligations. The "lock-in effect" of these AI deals means that even when Bitcoin mining becomes more profitable on a marginal basis, the resources required to resume operations are simply unavailable.
The AI Conversion: A Strategic Technology Stack Transfer
This is not a story about ASIC hardware becoming obsolete. The mining rigs themselves remain functional. What we are witnessing is a strategic reallocation of resources at the corporate level. Companies like IREN and TeraWulf have dramatically reduced their Bitcoin mining capacity while pivoting toward AI and high-performance computing services. IREN, for instance, has cut its mining deployment while simultaneously building out GPU clusters for AI inference workloads.
The economics driving this transition are compelling. AI compute services command premium pricing with long-term contracts that provide revenue visibility—something Bitcoin mining, with its inherent price volatility, cannot offer. For publicly traded mining companies under pressure from shareholders to deliver consistent earnings, the AI pivot represents a rational risk management strategy. It is not an abandonment of Bitcoin; it is a hedge against Bitcoin price volatility.
Based on my experience auditing mining operations during the 2020 DeFi yield farming era, I can confirm that the operational expertise required for Bitcoin mining—power procurement, cooling systems, data center management, grid connectivity—transfers almost perfectly to AI infrastructure. The miners possess exactly what AI developers need: access to cheap power, existing industrial sites, and the technical capability to operate large-scale computing facilities.
The Divergence Within the Mining Sector
The mining industry is not monolithic in its response. We are seeing a clear bifurcation in strategy. MARA Holdings, Bitdeer, and Riot Platforms continue to expand their Bitcoin mining capacity, albeit with varying degrees of enthusiasm. Meanwhile, IREN and TeraWulf have pivoted decisively toward AI/HPC services. This strategic divergence reflects genuine disagreement among industry leaders about the long-term trajectory of both Bitcoin and AI compute markets.
The companies doubling down on Bitcoin are making a bet that the current hashrate drawdown is temporary and that the network's security budget will recover as price appreciation continues. The companies pivoting to AI are making a different calculation: that the risk-adjusted returns on AI infrastructure over the next five years will exceed those of Bitcoin mining, even accounting for potential Bitcoin price appreciation.
From a portfolio construction perspective, this divergence creates an interesting dynamic. Publicly traded mining companies are no longer pure-play Bitcoin proxies. Some are becoming diversified compute infrastructure providers. This decoupling from Bitcoin's price means that investors seeking Bitcoin exposure through mining equities must now conduct additional due diligence on each company's strategic direction.
The Security Budget Question
The most significant long-term concern is what this means for Bitcoin's security budget. At 914 EH/s, the absolute hashrate remains substantial. The cost of mounting a 51% attack is still prohibitive for any rational actor. However, the marginal decline in security margin is real, and if the trend continues, it could eventually raise questions about the network's resilience.
I have audited the security assumptions of multiple PoW networks, and the pattern is consistent: hashrate concentration and decline are lagging indicators of miner confidence. When miners with access to cheap power choose to allocate that power elsewhere, they are signaling that the expected value of Bitcoin mining, adjusted for risk, is lower than the expected value of AI services. This is a market signal that should not be ignored.
The difficulty adjustment mechanism provides a floor for miner profitability, but it cannot address the opportunity cost problem. If AI services continue to offer superior risk-adjusted returns, the equilibrium hashrate for Bitcoin will settle at a lower level than historical trends would suggest. This is not necessarily bearish for price—lower hashrate means lower sell pressure from miners—but it does represent a structural change in how the network's security is provisioned.
The Hidden Risk: Double Exposure
The contrarian angle that most analysts are missing is the double-exposure risk facing mining companies that pivot to AI. If the AI compute market experiences a correction—and history suggests that all infrastructure booms eventually correct—these companies will find themselves in a precarious position. Their Bitcoin mining operations will have atrophied, their AI contracts may become less profitable as competition intensifies, and their capital expenditure on GPU infrastructure may not generate the expected returns.
This is the "two chairs" problem. A company that tries to sit on both the Bitcoin mining and AI compute chairs simultaneously may find itself falling between them. The operational requirements of both businesses are similar, but the customer bases, contract structures, and competitive dynamics are fundamentally different. Execution risk is substantial.
I have seen this pattern before in the 2022 Terra collapse, where companies that tried to hedge across multiple strategies found themselves exposed to correlated failures. The lesson from that experience is clear: when you are operating in a high-volatility environment, focus and discipline matter more than diversification across superficially similar but fundamentally different businesses.
The Path Forward: What to Watch
The key metric to monitor is whether hashrate stabilizes above 900 EH/s or continues its decline. If we see a sustained drop below 800 EH/s, the market will begin pricing in a permanent reduction in Bitcoin's security budget. This would have implications for the "digital gold" narrative that underpins much of Bitcoin's institutional adoption thesis.
The second signal to track is the revenue mix of publicly traded mining companies. When Q3 earnings are reported, pay close attention to the percentage of revenue derived from AI services versus Bitcoin mining. A significant shift toward AI revenue will confirm that the transition is structural, not cyclical.
The third factor is the AI compute pricing environment. If GPU cloud service prices decline due to oversupply—which is a real possibility given the massive buildout currently underway—the economics of the AI pivot will deteriorate, and we may see miners return to Bitcoin. This would create a rapid hashrate recovery and potentially a positive price shock.
The Bottom Line
Bitcoin's hashrate decline is not a technical failure. The protocol is functioning exactly as designed. The difficulty adjustment mechanism is working, block production is on schedule, and the network remains secure. What we are witnessing is an economic reallocation of resources driven by market forces. Miners are rational actors, and they are responding to incentives.
The question that matters is whether this represents a temporary rebalancing or a permanent structural shift. My analysis suggests it is closer to the latter. The lock-in effects of long-term AI contracts, the strategic pivot of major mining companies, and the fundamental economics of AI compute versus Bitcoin mining all point toward a new equilibrium with lower Bitcoin hashrate than historical trends would predict.
This is not necessarily bearish. Lower hashrate means less miner sell pressure. It means the remaining miners are more efficient and more committed. It means the network's security is provisioned by true believers rather than mercenaries. But it also means that the narrative of Bitcoin as an ever-strengthening network with ever-increasing security must be revised.
The market has not fully priced this in. The price-hashrate divergence is a warning signal that the relationship between Bitcoin's fundamental security and its market value is being renegotiated. Smart money is watching. The question is whether the market will recognize this structural shift before it becomes obvious to everyone.
I audit the code, not the charisma. The code is fine. The economics are changing. Strategy beats speculation every time. Volatility is the price of entry. Yields are calculated, not guaranteed. Diversification is the only safety net. Verify the source, trust no one. Liquidity dries up faster than hope. Smart contracts don't lie, but their operators can.