In a notable turn of events, Bitcoin’s network has experienced a significant dip in power this week, falling beneath the crucial one-zettahash threshold for the first time since September 2025. Current reports indicate that the seven-day average hashrate stands at approximately 993 EH/s, marking a clear departure from the highs of the previous year.
The Hunger for Power
The increasing demand for electricity by large AI data centers is at the forefront of this change. These facilities are securing long-term power contracts and are willing to pay premium prices for consistent, round-the-clock electricity. As a consequence, many Bitcoin miners are re-evaluating their operations, some choosing to cut back or relocate initiatives altogether. This fierce competition for power availability is radically altering the landscape for who gets to access the lowest electricity rates on the grid.
In response to these dynamics, publicly traded mining companies have begun securing deals to lease space to technology firms focused on AI, converting portions of their mining locations into hybrid AI and cryptocurrency data centers. One prominent miner recently finalized a multi-year lease agreement with a leading chip manufacturer, illustrating how companies are strategizing to offset unpredictable mining revenues by diversifying into AI workloads.
On January 19, 2026, Leon Lyu, CEO and founder of StandardHash, announced via X that the ongoing dip in Bitcoin’s hashrate is largely driven by shifts in electricity allocation from mining operations towards AI computing, which promises more attractive profit margins.
Why the Shift Matters Now
Electricity consumption represents the largest expenditure for Bitcoin mining operations. When data centers engage in bidding wars for electricity supply, miners are confronted with a stark choice: pay more to secure power, accept tighter margins, or repurpose their capacities for alternative uses.
The recent adjustments in hashrate have prompted a reduction in mining difficulty, which helps in maintaining relatively stable block times on the network. However, this mechanical adjustment does not address the pressing reality of who retains the most favorable power contracts.
PJM, the operator of the mid-Atlantic power grid, has proactively proposed new regulations to manage the escalating demand from AI users, insisting that these newly established power consumers either take responsibility for their own supply or agree to curtailment measures to avoid disruptions for essential services and residential users. These regulatory moves aim to alleviate the pressures that rapid AI expansion could impose on the energy infrastructure.
Bitcoin Versus AI: The Political Landscape
In a broader context, U.S. President Donald Trump, along with several state leaders, has been advocating for legislation that would require tech companies to pay more to secure electricity, including proposals for emergency auction systems to fund new power plants. These calls for action reflect concerns over rising costs and the potential risks that sprawling AI facilities could monopolize grid resources, limiting access for other users.
To adapt to the shifting landscape, many miners are not just shutting down operations during peak power pricing periods; instead, they are retrofitting their facilities to accommodate GPUs and related AI computing hardware. This pivot presents the potential for more stable revenue streams and longer contracts compared to traditional mining operations, marking a structural evolution where Bitcoin mining forms merely one component of a much larger computational enterprise.
While block rewards and protocol mechanisms continue to safeguard the Bitcoin network, prolonged reductions in hashrate may lead to new periods of centralization, especially in regions where the cost of power remains low. For everyday users, the Bitcoin system continues producing blocks; however, for miners, the ongoing contest for electricity is set to become the defining challenge of their operations.
Featured image from Unsplash, chart from TradingView
