MARA Holdings CEO: AI data center revenue per unit power exceeds Bitcoin mining

CEO of MARA Holdings: AI data center revenue per unit of electricity exceeds Bitcoin mining

CEO of MARA Holdings pointed out that the revenue per unit of electricity in artificial intelligence data centers has exceeded the company's gains from Bitcoin mining, a comparison that highlights the central position of power economics in the infrastructure planning of the largest listed miners.

This term measures the relationship between monetization efficiency and electricity, the single resource on which two businesses rely most. Revenue per unit of electricity describes how much revenue can be generated for a given power capacity, usually in units per megawatt. When artificial intelligence or high-performance computing workloads generate more revenue than the mining hash for the same power consumption, the economic considerations for power distribution begin to shift.

Electricity is a common constraint for both activities. Bitcoin miners compete for access to cheap, reliable electricity and the efficiency of facilities that convert electricity to hash rates, while AI data centers have urgent needs for the same grid connections and megawatts of capacity. MARA directly pointed out this overlap in its comments on Bitcoin mining in the era of artificial intelligence, positioning energy infrastructure rather than pure mining as a core asset.

Importance of income density to miners

For mining, profit margins depend on the difference between the cost of electricity and the bitcoins obtained. If a use case generates more revenue for the same energy consumption, then existing sites will be redefined as flexible energy infrastructure rather than single-use mines. Because of this, investors are increasingly evaluating miners based on revenue density per megawatt rather than just hash rates. The comparison made by the CEO shows that on this metric, AI hosting can surpass mining at the same power capacity.

How does this term fit into MARA's strategy

MARA has shown signs of moving in this direction. The company recently laid off about 15% of its workforce, switched to artificial intelligence and digital infrastructure, and made balance sheet adjustments, such as the consent solicitation of Long Ridge notes related to asset acquisitions. Both indicate that the company views power procurement and construction as core capabilities that go beyond mining. MARA detailed the infrastructure and energy assets that support this positioning in an investor presentation in March 2026. The overlap between mining facilities and AI or high-performance computing hosting requirements is real, but the transformation carries execution risks, including renovation costs, different customer contracts, and operational requirements that the mine was not originally designed to assume.

What does this mean for other miners

This comparison raises a question for every listed miner with energy-intensive infrastructure: Can AI workloads deliver better returns on the same energy base? The companies weighing this trade-off are those that appear on the list of the largest public Bitcoin holders, so this decision affects both their operating models and their financial reserve strategies. It is unclear whether this is a structural shift or opportunistic diversification. The mining economy also changes with network conditions; the difficulty of Bitcoin mining has recently dropped by 5% to 127.17T, reminding people that the revenue side of the comparison is not fixed. MARA did not say that it would withdraw from mining. The CEO's statement only concerned relative economics rather than confirmation of a comprehensive transformation.

What does revenue per unit of electricity mean?

It refers to revenue generated by fixed power capacity, usually measured per megawatt, allowing operators to compare mining with other uses of the same power.

Is MARA withdrawing from Bitcoin mining?

There is no evidence of this. The CEO compared the relative economics of AI hosting and mining; the company has not announced its withdrawal from mining.

Will this affect other Bitcoin mining companies?

Possibly. Any miner with a large amount of power capacity faces the same question: whether AI workloads can use that power more efficiently to generate revenue.