Bitcoin miners are becoming AI landlords, and Wall Street just validated the pivot with a price target that implies 76% upside.
The Summary
- Bernstein analysts set a $100 price target for IREN following the company's $3.4 billion AI infrastructure deal with NVIDIA, representing 76% implied upside from current levels.
- IREN's shift from crypto mining to AI infrastructure demonstrates how energy-intensive mining operations can transition into compute providers for the agent economy.
- The deal reshapes market perceptions of former crypto miners, positioning them as critical infrastructure for AI workloads rather than speculative energy consumers.
The Signal
IREN just closed a $3.4 billion cloud infrastructure contract with NVIDIA, and Bernstein's analysts see enough signal to call $100 a share. That's not hype. That's a Wall Street firm putting a number on what happens when you own power infrastructure, cooling systems, and data center space at the exact moment AI companies need all three at scale.
The company's shares surged on the announcement, and the market is starting to understand what crypto miners figured out years ago: compute is scarce, energy is scarce, and if you control both, you're not in the mining business anymore. You're in the infrastructure business. NVIDIA doesn't cut $3.4 billion checks to keep GPUs warm. They're buying access to the physical layer that makes training runs possible.
"The lucrative potential for former crypto miners reshapes market perceptions and investment strategies."
Here's the math that matters:
- Bernstein's $100 target implies 76% upside from current trading levels
- A $3.4 billion contract represents multi-year revenue visibility in a sector where most deals are quarterly commitments
- IREN already owns the hard infrastructure: buildings, power contracts, cooling systems, connectivity
The NVIDIA deal validates a thesis that's been floating around the edges of crypto Twitter for two years: Bitcoin miners were always in the business of converting electricity into computation at profitable margins. When Bitcoin's hash rate made that equation harder, the smart play wasn't to shut down. It was to pivot to the workload with better unit economics. AI training and inference fit that profile perfectly.
This pivot highlights how mining companies are repositioning as essential AI infrastructure providers rather than speculative players tied to Bitcoin's price volatility. IREN isn't alone. CoreWeave started the same way. So did several others now billing themselves as "AI cloud providers." The difference is IREN just got Bernstein to put a price target on the transformation.
Key infrastructure advantages:
- Existing power purchase agreements negotiated during mining boom
- Built-out data centers with industrial cooling already in place
- Experience managing compute at scale in energy-constrained environments
The timing matters too. Hyperscalers like AWS and Azure are capacity-constrained. Building new data centers takes 18-24 months. IREN's facilities exist now, wired for high-density compute, in locations where power is cheap and available. NVIDIA needs that capacity to fulfill GPU orders. Customers need it to train models. The $3.4 billion is NVIDIA saying "we can't wait for new builds."
This isn't a one-off. It's a template. Every Bitcoin miner sitting on stranded energy and empty racks is looking at this deal and doing the same math. Some will make the jump. Most won't have the capital or the relationships. But the ones that do will stop being "crypto companies" in the eyes of institutional investors and start being "AI infrastructure plays." That's a multiple expansion story, and Bernstein just wrote the first chapter.
The Implication
Watch the other miners. If IREN can pull this off, expect Riot, Marathon, and others to announce similar pivots. The compute shortage isn't easing. The power shortage isn't easing. Companies that own both become strategic assets, not speculative bets. For investors, this is a signal to reassess any position that looks like "old crypto" but sits on real infrastructure. For AI companies, it's a reminder that the bottleneck isn't algorithms or models. It's power and racks.