Investor Guide
Understanding Returns
How Returns Are Generated#
Revenue Sources#
1. AI Model Training#
Clients pay hourly rates for GPU access to train:
- Large language models (LLMs)
- Computer vision models
- Recommendation systems
- Reinforcement learning agents
Typical rate: $2–8 per GPU hour
2. AI Inference Serving#
Real-time model serving for:
- API endpoints
- Production applications
- High-frequency inference
- Batch processing
Typical rate: $1.50–5 per GPU hour
3. 3D Rendering & Graphics#
Studios and creators use cluster capacity for:
- Movie VFX and animation
- Architectural visualization
- Product rendering
- Real-time graphics
Typical rate: $1–4 per GPU hour
4. Scientific Computing#
Research institutions run workloads such as:
- Molecular dynamics simulations
- Climate modeling
- Genomic analysis
- Financial modeling
Typical rate: $2–6 per GPU hour
5. Blockchain & Web3#
Computational tasks may include:
- Zero-knowledge proof generation
- Consensus validation
- Cryptographic operations
- Decentralized network support
Typical rate: $1.50–4 per GPU hour
Return Calculation Example#
Scenario: Enterprise Package#
- Your investment: $1,299
- Cluster: H100 ×4 configuration
- Rental period: 65 days
Daily Revenue Breakdown#
- Cluster charges B2B clients: $4/hour average
- Cluster operates: 20 hours/day average (83% utilization)
- Daily B2B revenue: $320 ($4 × 4 GPUs × 20 hours)
- Platform operational costs: 30% ($96)
- Net distributable revenue: $224
- Your share: $22.40 as 1 of 10 node investors
- Your daily return: 1.72% ($22.40 / $1,299)
Over 65 Days#
- Total accumulated: $1,456 ($22.40 × 65)
- Profit: $157 ($1,456 - $1,299)
- ROI: 12.1% over 65 days
- Annualized: approximately 68%, if similar performance is maintained
Important Notes#
- Actual returns vary based on real cluster utilization
- Some days may be higher or lower than the advertised range
- Weekends and holidays may produce lower demand
- Peak seasons, such as Q4 for rendering and year-round AI demand, may produce higher utilization
Factors Affecting Returns#
Positive Factors (Higher Returns)#
- High B2B client demand
- Premium client contracts
- Extended job durations
- Peak-season workloads
- New cluster deployments
- Geographic demand surges
Negative Factors (Lower Returns)#
- Reduced B2B bookings
- Seasonal slowdowns
- Technical maintenance windows
- Market competition
- Hardware issues, even when covered by SLA processes
Return Stability#
Historical performance data shows:
- 80–90% of days fall within the advertised range
- 5–10% of days exceed the high end
- 5–10% of days fall below the low end
- Average utilization is 85% across all clusters