# What are the expected IRRs?
> Understanding projected internal rates of return for GPU investments.
Source: https://help.computelabs.ai/en/docs/deals-performance-liquidity/expected-irr
Language: en

Returns vary by GPU model, operator, and market conditions. Historical deals have delivered double-digit yields, though future performance may differ. Compute Labs has an internal mandate that the offer must project or estimate a 20% IRR before it is offered to investors.

## IRR Framework

### Minimum Standards
- **Internal Mandate**: All deals must project at least 20% IRR to be offered
- **Conservative Modeling**: We use conservative assumptions for utilization and pricing
- **Stress Testing**: Models are tested against various market scenarios

### Factors Affecting IRR

#### Hardware Factors
- **GPU Model**: Latest generation hardware typically commands higher rates
- **Performance Efficiency**: More efficient chips generate better margins
- **Hardware Costs**: Lower acquisition costs improve returns

#### Market Factors
- **Demand Cycles**: AI training and inference demand fluctuations
- **Competitive Pricing**: Market rates for GPU compute services
- **Energy Costs**: Impact on operating expenses and net margins

#### Operational Factors
- **Utilization Rates**: Percentage of time GPUs are actively generating revenue
- **Operator Efficiency**: Quality of data center operations and sales
- **Geographic Location**: Regional pricing and cost variations

## Historical Performance

### Track Record
- Double-digit yields achieved across multiple vault deployments
- Consistent performance despite market volatility
- Strong utilization rates from enterprise AI demand

### Performance Drivers
- Growing demand for AI compute resources
- Limited GPU supply creating pricing power
- Professional operator relationships ensuring consistent utilization

## IRR Components

### Revenue Generation
- **Base Compute**: Standard AI training and inference workloads
- **Premium Services**: Specialized high-performance computing
- **Long-term Contracts**: Stable revenue from enterprise partnerships

### Cost Structure
- **Operating Expenses**: Power, cooling, maintenance, facility costs
- **Management Fee**: 10% of net proceeds to Compute Labs
- **Depreciation**: Tax advantages from hardware depreciation schedules

### Value Creation
- **Cash Flow**: Monthly USDC distributions from operations
- **Asset Appreciation**: Potential upside from GPU value retention
- **Tax Benefits**: Depreciation shields for applicable investors

## Risk Considerations

While we target 20%+ IRRs, actual returns depend on:
- Market demand for AI compute services
- Hardware performance and reliability
- Operator execution and utilization rates
- Broader economic and technology cycles

Past performance does not guarantee future results. All investments carry risk of loss.

## Due Diligence Process

Before offering any deal, we conduct:
- Comprehensive financial modeling
- Operator due diligence and verification
- Market analysis and competitive positioning
- Stress testing under various scenarios

This rigorous approach helps ensure that projected returns are achievable while maintaining appropriate risk management standards.