#Gate广场五月交易分享 #加密矿企加速布局AIDC From mining to AI computing power, this migration at least reveals five levels of market conditions: 1. Bitcoin mining has entered the "capital efficiency trap"

After the 2024 halving, the block reward is only 3.125 BTC, and the total network computing power once surpassed 1 ZH/s in 2025, with difficulty peaking at 155.9 trillion. Even if difficulty drops to about 135 trillion in early 2026 and mining profitability recovers somewhat, fundamentally—the capital efficiency of mining has been structurally compressed: investing the same electricity and funds, the returns from mining are volatile and unpredictable (depending on BTC price), whereas AI/HPC returns are locked in long-term contracts and predictable.
Capital always chases certainty; this is the underlying logic of the transformation.
2. AI computing power is experiencing an "infrastructure bottleneck"
GPU data center construction cycles are 3–5 years, data center GPU delivery wait times are 36–52 weeks, and HBM and advanced packaging capacities are severely insufficient. Five major tech companies are expected to spend over $600 billion on GPU/data center capital expenditures by 2026. The bottleneck is not in the chips themselves but in power, land, and cooling—precisely the core assets accumulated by mining companies over ten years. Their power contracts, existing data centers, and operations teams make them the only entities capable of "skipping the 3-year construction period" in AI expansion.
3. The valuation system in capital markets has undergone a "paradigm shift"
CoinShares predicts that the share of mining revenue from transformed mining companies will drop from 85% to less than 20% by the end of 2026. IREN rose from a yearly low of $31.62 to about $52; Starboard Value estimates Riot’s AI/HPC transformation valuation could reach $21 billion, far exceeding pure mining valuations. Morgan Stanley has given buy ratings to CIFR and WULF, while downgrading MARA (slower to transform) to underperform. The market is clearly signaling through prices: the pricing power of AI infrastructure far exceeds that of crypto mining.
4. Two "power-intensive infrastructure industries" are competing for resources
The core competitive elements of Bitcoin mining and AI computing power are exactly the same—who can acquire the largest scale of electricity and infrastructure at the lowest cost. When AI’s electricity demand growth far outpaces mining, power resources naturally flow toward higher marginal returns. This is not simply "riding the AI wave," but a reallocation of two trillion-dollar tracks for the same scarce resource. As Crucible Capital partner Meltem Demirors said: "Bitcoin mining has created a blueprint for the prosperity of AI computing power and modern data centers."
5. The Bitcoin network security model faces latent risks
Large-scale exits of mining companies mean a decline in total network hash rate. By early 2026, hash rate has fallen from its peak to about 900–948 EH/s. If more mining companies shift their mining hardware’s power to AI businesses, the security margin of the BTC network will be compressed. This is a long-term contradiction—AI’s competition for electricity may continue to squeeze mining hash rate supply, while Bitcoin’s security relies precisely on the scale of hash rate.
This migration reflects a core market condition: the crypto industry is entering a phase of low returns, the AI industry faces infrastructure bottlenecks, and capital has made a clear priority choice—power and infrastructure are shifting from serving BTC hash rate to serving GPU floating-point computations. This is not a short-term trend change but a structural reallocation of two trillion-dollar tracks for the same scarce resource.