The number of companies building physical infrastructure for AI has increased by 187% over the past 12 months. It reflects a genuine land grab happening at the intersection of two industries that, until recently, seemed to exist in parallel universes: artificial intelligence and cryptocurrency mining.

Bitcoin miners find their second act

The pivot from mining Bitcoin to hosting AI workloads has been nothing short of dramatic for several publicly traded crypto miners. Hut 8, one of the more recognizable names in North American Bitcoin mining, saw its stock climb roughly 211% over the past year. Several peers in the space have posted gains ranging from triple digits to as high as 800%.

TeraWulf may have made the splashiest move of all, signing a $19 billion AI data center contract with Anthropic, the company behind the Claude AI model. Cipher Mining, trading under the ticker CIFR, and Hut 8 (HUT) have both leaned hard into the AI infrastructure narrative. The physical infrastructure is remarkably transferable: Bitcoin mining requires massive amounts of electricity, industrial-scale cooling systems, and facilities designed to run 24/7 without interruption. AI model training and inference require exactly the same things.

The numbers behind the gold rush

Analysts at Compass Point have flagged something interesting about the valuation gap in this space. Several AI infrastructure stocks, including TeraWulf, may actually be trading below the implied value of the AI contracts they’ve already signed.

Some stocks in the sector have surged by over 493% in 12 months. Lumentum, which makes optical networking components critical to data center connectivity, has seen its stock increase by over 1,000% on the back of AI-driven demand.

Hyperscalers — the Microsofts, Amazons, Googles, and Metas of the world — are projected to invest roughly $700 billion in AI-related infrastructure by 2026. That figure represents a cumulative estimate across major players, and it creates a massive downstream demand for exactly the kind of physical capacity these infrastructure companies provide.

Why crypto investors should pay attention

Bitcoin mining has always been a brutally cyclical business. Margins expand during bull runs and compress violently during downturns, especially after halving events that cut block rewards in half. AI data center contracts, by contrast, tend to be long-term agreements with predictable revenue streams. For companies like TeraWulf, a $19 billion contract with Anthropic provides the kind of earnings visibility that Bitcoin mining simply cannot.

Many are running dual operations, maintaining their mining rigs while simultaneously building out AI capacity. The risk is execution. Converting mining facilities to AI-grade data centers isn’t trivial. AI workloads demand different networking configurations, higher-density power delivery, and more sophisticated cooling solutions than Bitcoin mining.

There’s also the question of competition. As the 187% increase in AI infrastructure companies suggests, the field is getting crowded fast. Traditional data center operators like Equinix and Digital Realty have decades of operational expertise and established relationships with enterprise customers. Bitcoin miners entering the space are essentially arguing that their cost advantages on power and real estate can offset their relative inexperience in enterprise data center operations.

The Compass Point analysis offers a useful framework: if AI infrastructure stocks are genuinely trading below the value of their signed contracts, that gap represents either a buying opportunity or a market signal that execution risk is being priced in more heavily than the headline numbers suggest. An investor who bought Hut 8 a year ago for Bitcoin exposure has, perhaps inadvertently, become an AI infrastructure investor.