Interhash: Mining Shift to AI Data Centers Faces Limits

The rapid growth of artificial intelligence is reshaping demand for computing infrastructure. At the same time, the Bitcoin mining industry faces rising network difficulty and higher electricity costs. Market cycles also continue to pressure mining profitability.

These trends have fueled speculation about converting mining farms into AI data centers. However, Interhash CEO Alexander Lozben believes this assumption oversimplifies reality.

According to Lozben, the transition from Bitcoin mining to AI infrastructure faces economic and infrastructure barriers. These constraints are particularly visible in Russia.

Today, competition is not between Bitcoin mining and artificial intelligence. Instead, companies compete for access to electricity and grid infrastructure.

Energy Access Becomes the Key Constraint

Both Bitcoin mining and artificial intelligence require significant energy consumption. Large industrial mining farms typically use between 20 and 100 megawatts of power. Similar levels are required for large AI training clusters.

Lozben notes that only a limited number of Russian regions have surplus electricity capacity. Obtaining new grid connections can also take a long time. As a result, the key factor is not energy price but energy availability.

Globally, the situation looks different:

• electricity demand from data centers in the U.S. grows more than 15% yearly. By 2030, data centers could become a major driver of grid demand.
• Middle Eastern countries offer electricity below $0.04 per kWh.

In Russia, however, the main limitation remains access to available power capacity.

Mining and AI Require Different Infrastructure

Despite similar energy needs, mining farms and AI data centers are built for different tasks. Bitcoin mining uses specialized ASIC hardware. These machines are optimized for density and efficiency.

AI infrastructure relies on GPUs and accelerators. These systems require higher reliability standards. Typical AI data center requirements include:

• uptime above 99.99%;
• redundant power supply lines;
• backup diesel generators;
• multiple telecom providers;
• liquid cooling systems.

Because of these requirements, most mining facilities cannot support AI workloads.

Capital investment also differs significantly. Mining infrastructure typically costs about $300,000–$600,000 per megawatt. AI data centers require roughly $3–6 million per megawatt.

In practice, this means building entirely new infrastructure.

Different Business Models

Mining and AI data centers operate under different economic models. Bitcoin mining functions within protocol economics. Revenue depends on block rewards, transaction fees, and network difficulty. There is no counterparty risk because the blockchain protocol acts as the customer.

According to Lozben, this mechanism makes the mining industry resilient. When miners leave the network, mining difficulty adjusts automatically.

AI data centers operate differently. Their revenue depends on long-term contracts with enterprise clients. Most global demand comes from large technology companies: Microsoft, Google, Amazon, and Meta. In Russia, the potential customer base remains significantly smaller.

Global Industry Examples

Several companies are often cited as examples of a transition from mining to AI infrastructure. However, the real situation is more complex. CoreWeave originally started as a mining operation. Later it exited mining and built GPU-focused infrastructure.

Similar strategies have been pursued by Northern Data, Hut 8, and IREN. In most cases, companies reallocate energy capacity rather than convert mining facilities directly.

According to Lozben, investor interest in AI infrastructure is driven by valuation differences. In the U.S., AI data centers often receive much higher valuations than mining facilities.

AI Inside Mining Operations

Artificial intelligence is already used within mining operations. However, its role remains operational rather than strategic. Common applications include:

• predictive equipment maintenance;
• infrastructure monitoring;
• logistics optimization;
• asset management.

These tools improve efficiency but do not change mining economics.

Conclusion

According to Interhash, a large-scale transformation of mining farms into AI data centers remains unlikely. Both industries rely on energy infrastructure. However, their economics, infrastructure requirements, and risk structures differ significantly.

In Russia, additional constraints include:

• limited electricity capacity;
• high capital investment requirements;
• a narrow domestic AI customer market;
• regulatory factors.

Hybrid projects may emerge, but a full industry transition is unlikely. Bitcoin mining therefore remains an independent digital infrastructure sector.

Read also: Bitcoin Miners Pivot to AI as Nvidia Expands Rubin