'We can't meet the demand fast enough,' says Iren co-CEO Dan Roberts

Key Concepts

• AI Factories: Data centers providing compute power to hyperscalers, AI labs, and enterprises for AI-related tasks.
• Hyperscalers: Large cloud service providers (e.g., AWS, Azure, Google Cloud).
• AI Cloud Compute: Providing computing resources for AI workloads as a service.
• Vertically Integrated: Owning and controlling multiple stages of the value chain (e.g., data centers, power, land).
• Renewable Energy: Energy from sources that are naturally replenished (e.g., wind, solar, hydro).
• Co-location: Providing space and infrastructure for customers to house their own servers and equipment.
• Asics: Application-Specific Integrated Circuits, specialized hardware for Bitcoin mining.
• GPUs: Graphics Processing Units, processors well-suited for AI and machine learning tasks.
• Powered Land: Land secured with access to significant power capacity.

1. AI Growth and Demand

• Rapid Growth: The AI compute demand is growing "super quickly," and the company cannot meet the demand fast enough.
• AI Factories: The company is focused on building "AI factories" to deliver compute to hyperscalers, AI labs, and enterprises for inference and training.
• Competition: The company is in a similar space to companies like CoreWeave and Oracle, providing AI cloud compute to end customers.
• Advantage: The company's key advantage is owning the data centers, power, and land.
• GPU Installations: The company is actively installing GPUs to meet the current demand.

2. Cyclical Concerns and Vertical Integration

• Cyclical Nature: There are concerns about the potential cyclical nature of the AI compute market, with possible swings from boom to bust.
• Vertically Integrated Advantage: The company is vertically integrated, owning data centers that were initially built for Bitcoin mining but are now being adapted for AI.
• Multi-Purpose Data Centers: The data centers were built for multi-purpose use, allowing for flexibility in adapting to different workloads.

3. Power and Renewable Energy

• Powered Land: The company has accumulated three gigawatts of powered land over the past seven years.
• Renewable Energy: The company has been 100% renewables since inception, using wind, solar, and hydro power.
• No Generation Risk: The company does not take generation risk on renewables; it connects to high-voltage transmission networks that guarantee reliable renewable energy.

4. Market Dynamics and Strategy

• Co-location vs. AI Cloud: While the company considered co-location, it is now focused on AI cloud to capture more of the economics by providing compute and building relationships directly with end customers like AWS.
• Industry Capacity: The industry cannot get enough capacity at the moment.
• Hyperscaler Strategy: Hyperscalers are open to both co-location and AI cloud.

5. Bitcoin Mining and AI Focus

• Bitcoin's Role: Bitcoin mining bootstrapped the business and currently delivers about $1 billion of annualized revenue.
• Future is AI: The future focus is on AI, with less than 2% of the company's power capacity capable of generating $500 million of annualized revenue.
• Growth Outlook: The outlook for growth in the AI service market is "super exciting."

6. Infrastructure and Expansion

• GPU Capacity: The company has contracted for 23,000 GPUs, which will deliver around $500 million in annualized revenue.
• Data Center Capacity: This represents less than 2% of the company's overall footprint.
• Future Shortfall: One US investment bank forecasts a shortfall of 44GW over the next three years.
• McKinsey Forecast: McKinsey forecasts an additional 100GW of data center capacity by 2030.
• Site Expansion: The company has a multi-gigawatt pipeline for future sites.

7. Transitioning Facilities

• Rapid Transition: The company can quickly transition facilities from Bitcoin mining to AI by swapping out racks of ASICs for Nvidia GPU servers.
• Speed of Installation: The speed of installation depends on the availability of labor and chips.

8. Data and Statistics

• 3 Gigawatts: Secured power capacity.
• 23,000 GPUs: Contracted GPU capacity.
• $500 Million: Annualized revenue from contracted GPUs.
• 2%: Percentage of overall footprint used for contracted GPUs.
• $1 Billion: Annualized revenue from Bitcoin mining.
• 44GW: Forecasted shortfall in data center capacity over the next three years (US investment bank).
• 100GW: Forecasted additional data center capacity needed by 2030 (McKinsey).

9. Notable Quotes

• "We can't meet the demand fast enough." - Dan Roberts, on the demand for AI compute.
• "It's the same data center that was built for multi purpose." - Dan Roberts, on the flexibility of their data centers.
• "Bitcoin was great. It bootstrapped the business." - Dan Roberts, on the role of Bitcoin mining.
• "The future is AI." - Dan Roberts, emphasizing the company's focus.

10. Synthesis/Conclusion

The company is strategically pivoting from Bitcoin mining to AI cloud compute, leveraging its existing infrastructure and renewable energy sources. The demand for AI compute is rapidly growing, and the company is focused on expanding its GPU capacity and securing more powered land to meet this demand. While Bitcoin mining remains a significant revenue source, the company sees AI as the primary growth driver for the future. The company's vertically integrated model and access to renewable energy provide a competitive advantage in the AI cloud market.