#加密矿企加速布局AIDC #Gate广场五月交易分享 The collective shift of encrypted mining companies toward AI data centers (AIDC) appears on the surface to be an enterprise-level strategic transformation, but in reality, it reflects several deep market conditions:

1. Mining economics have entered the "diminishing marginal returns" stage
The total network hash rate of Bitcoin once surpassed 1 ZH/s (1,000 EH/s) in 2025, with difficulty rising to a historic high of 155.9 trillion. Although difficulty slightly decreased to about 135 trillion in early 2026, the overall competitive landscape continues to intensify— the cost to mine each BTC keeps rising, and the block reward will only be 3.125 BTC after the 2024 halving.
Mining companies' return on investment in hash power is being compressed, turning mining from a "high-profit race" into a "struggling to cover electricity costs" chore. Bitfarms CEO's statement is straightforward: it's not that mining is no longer profitable, but that under the same electricity input, AI/HPC can generate far higher and more certain returns, making continued investment in mining impossible to justify to shareholders.
2. AI computing power is experiencing structural scarcity on the supply side
Building GPU data centers typically takes 3–5 years, with delivery wait times for data center GPUs reaching 36–52 weeks. High-bandwidth memory (HBM) and advanced packaging capacities are severely insufficient. In 2026, the top five tech companies are expected to spend over $600 billion on GPUs and data centers, while physical infrastructure—electricity, land, cooling systems—becomes a more urgent bottleneck than the chips themselves.
Mining companies happen to hold these "most difficult-to-acquire slow-variable assets": signed power contracts, built data centers, operational cooling systems, and maintenance teams. This makes them the most practical "plug-and-play" partners during the AI expansion cycle.
3. The narrative power in capital markets has undergone a fundamental shift
The transformation of mining companies is not a gradual trial but a capital-driven sharp turn. IREN signed a $9.7 billion five-year contract with Microsoft, TeraWulf signed a $12.8 billion order with Fluidstack, and six mining companies have a combined AI/HPC contract scale of about $38.5 billion. Many mining firms' AI/HPC revenue share has risen from less than 15% to 40–60%. CoinShares predicts that by the end of 2026, the share of mining income for transformed mining companies will drop from 85% to less than 20%.
Stock performance also confirms this: IREN rose from a yearly low of $31.62 to about $52, and Starboard Value estimates Riot’s potential valuation after transitioning to AI could reach $21 billion—far exceeding its pure mining valuation. The market’s pricing of "AI infrastructure" far exceeds that of "cryptocurrency mining," and this valuation gap is the strongest driver of the transformation.
4. Deeper signals: the intersection of two "infrastructure industries"
This migration reveals an often overlooked fact—Bitcoin mining and AI computing power are fundamentally both "electricity-intensive infrastructure businesses."
Over the past decade, mining has validated key engineering issues for AIDC, such as large-scale power procurement, high-density cooling, and deployment in remote areas. As Crucible Capital partner Meltem Demirors said: "Bitcoin mining has created a blueprint for the prosperity of AI computing power and modern data centers." The core competitive elements of both industries are the same—who can acquire the maximum scale of electricity and infrastructure at the lowest cost, wins. When AI’s electricity demand growth far outpaces mining, resources naturally flow toward higher marginal returns.
5. Risks and concerns: the costs of transformation are not negligible
CoreWeave attempted to acquire Core Scientific for about $9 billion but was halted due to shareholder opposition; many mining companies sold large amounts of BTC holdings during their transition (Core Scientific sold 1,900 BTC worth about $175 million, Riot sold 1,080 BTC worth about $96 million); huge capital expenditures pose risks of equity dilution; transforming ASIC mining farms into GPU data centers also presents significant technical challenges.
More concerning is that if mining companies exit mining on a large scale, the security model of the Bitcoin network will face downward pressure on hash rate—by early 2026, the total network hash rate has fallen back to about 900–948 EH/s from its peak, partly because mining machines are being diverted to AI businesses.
In summary, this "shift from mining to computing power" reflects a market state: after the halving, the crypto industry has entered a low-return normal, while the AI industry faces infrastructure bottlenecks during its explosive growth, and capital has made a clear priority choice—electricity and infrastructure are shifting from serving BTC hash rate to serving GPU floating-point calculations. This is not simply "riding the trend," but a reallocation of two trillion-dollar tracks competing for the same scarce resources (electricity + infrastructure).