Yes, io.net offers reserved capacity plans for enterprise customers with 10-20% additional discounts on top of already competitive on-demand pricing. Unlike AWS/Azure reserved instances that require 1-3 year commitments with significant upfront payments, io.net's reserved plans start at just 3 months with monthly billing and no early termination penalties. Since io.net's on-demand pricing ($0.18-$2.20/hr) is already 50-70% cheaper than AWS on-demand, adding a 20% reserved discount makes io.net reserved capacity up to 76% cheaper than AWS reserved instances, delivering enterprise-grade availability guarantees at unbeatable prices.
Understanding Reserved Capacity on io.net
Reserved capacity guarantees GPU availability for mission-critical workloads:
How It Works:
1. Commit to minimum GPU usage (10+ GPUs for 3-12 months)
2. Receive volume discount (10-20% off on-demand pricing)
3. Get guaranteed availability (99.5-99.9% SLA with credits for downtime)
4. Access enterprise support (dedicated account manager, priority provisioning, 24/7 support)
Key Differences from AWS/Azure Reserved Instances:
- Flexible commitments: 3, 6, or 12 months (vs. AWS 1-3 years)
- No upfront payment: Pay monthly, not all upfront (vs. AWS requiring 50-100% upfront)
- No termination penalties: Cancel after minimum term with no fees (vs. AWS no refunds)
- Already low base pricing: Discounts apply to pricing that's 50-70% below AWS on-demand
Reserved Capacity Pricing Tiers
| Commitment Length | Discount | Availability SLA | Min Monthly Spend | Billing |
|---|---|---|---|---|
| On-Demand (No Commitment) | 0% | 99%+ (no SLA) | $0 | Per-second |
| 3-Month Reserved | 10% | 99.5% | $5,000/month | Monthly |
| 6-Month Reserved | 15% | 99.5% | $10,000/month | Monthly |
| 12-Month Reserved | 20% | 99.9% | $25,000/month | Monthly |
| Custom Enterprise | 20-25% | 99.95% | $50,000+/month | Custom |
Minimum GPU Requirements:
- 3-month: 10 GPUs minimum
- 6-month: 15 GPUs minimum
- 12-month: 20 GPUs minimum
- Custom: Varies by requirements
Total Savings: io.net Reserved vs. AWS Reserved
Real pricing comparison for enterprise workloads:
Scenario: 10x H100 SXM GPUs running 24/7 for 12 months
| Provider/Plan | Hourly Rate | Monthly Cost (720 hrs × 10 GPUs) | Annual Cost | vs AWS On-Demand | vs AWS Reserved |
|---|---|---|---|---|---|
| AWS On-Demand | $6.98/hr | $50,256 | $603,072 | Baseline | N/A |
| AWS 1-Year Reserved (No Upfront) | $4.89/hr | $35,208 | $422,496 | 30% savings | Baseline |
| AWS 1-Year Reserved (All Upfront) | $4.19/hr | $30,168 | $362,016 | 40% savings | 14% savings |
| io.net On-Demand | $2.20/hr | $15,840 | $190,080 | 68% savings | 55% cheaper |
| io.net 12-Month Reserved | $1.76/hr | $12,672 | $152,064 | 75% savings | 64% cheaper |
Key Insight: io.net's 12-month reserved capacity at $1.76/hr is 64% cheaper than AWS's best reserved instance pricing ($4.19/hr all upfront), while requiring zero upfront payment and offering flexible monthly billing.
Scenario: 20x A100 80GB GPUs running 12 hours/day for 6 months
| Provider/Plan | Hourly Rate | Monthly Cost (360 hrs × 20 GPUs) | 6-Month Cost |
|---|---|---|---|
| AWS On-Demand | $4.10/hr | $29,520 | $177,120 |
| AWS 6-Month Reserved | Not available (min 1 year) | N/A | N/A |
| io.net On-Demand | $1.49/hr | $10,728 | $64,368 |
| io.net 6-Month Reserved | $1.27/hr | $9,144 | $54,864 |
Savings: $122,256 (69%) vs. AWS on-demand
AWS doesn't offer 6-month reserved terms - you're locked into 1 or 3 years or forced to pay on-demand prices.
Reserved Instance Comparison: io.net vs. AWS vs. Azure
| Feature | io.net Reserved | AWS Reserved | Azure Reserved |
|---|---|---|---|
| Minimum term | 3 months | 12 months | 12 months |
| Payment options | Monthly billing | All upfront, partial upfront, no upfront | Monthly or upfront |
| Upfront cost | $0 | 0-100% of total cost | 0-100% of total cost |
| Cancellation | After minimum term, no penalty | No cancellation or refunds | No cancellation or refunds |
| Instance flexibility | Switch GPU types within commitment | Limited (same instance family only) | Limited (same instance family only) |
| Regional flexibility | Move between regions freely | Region-locked (or pay 10% premium) | Region-locked |
| SLA included | 99.5-99.9% | 99.99% (separate cost) | 99.95% (separate cost) |
| Support included | Enterprise support included | Extra 10-20% for support | Extra 10-20% for support |
| Discount vs on-demand | 10-20% additional (already 50-70% below AWS) | 30-40% | 30-40% |
Winner: io.net offers superior flexibility (shorter terms, no upfront costs, free region switching) while delivering deeper total savings.
When to Choose Reserved Capacity
Reserved capacity makes sense if you answer "yes" to 2+ of these:
Ideal for Reserved Capacity:
- ✅ Predictable workloads running 8+ hours/day consistently
- ✅ Mission-critical production workloads requiring SLA guarantees
- ✅ Long-running training pipelines (multi-month model development)
- ✅ Production inference APIs with sustained traffic
- ✅ Monthly GPU spend exceeding $5,000
- ✅ Need for guaranteed GPU availability (no waiting for capacity)
- ✅ Enterprise compliance requirements (SOC 2, uptime SLAs)
Better Suited for On-Demand:
- ⚠️ Experimental workloads with uncertain duration
- ⚠️ Sporadic usage (2-4 hours/week)
- ⚠️ Rapidly changing GPU requirements (switching between GPU types frequently)
- ⚠️ Startup/small team with <$3,000/month GPU spend
- ⚠️ Development and testing (non-production)
Break-Even Analysis:
Reserved capacity pays off if your actual usage exceeds 70% of your commitment. For example:
- 3-month commitment: 10 GPUs × 720 hours/month × 3 months = 21,600 GPU-hours
- Need to use: 21,600 × 0.70 = 15,120 GPU-hours to break even
- That's ~7 GPUs running 24/7 for 3 months
If you use less than 70% of your commitment, on-demand is more cost-effective.
Flexible Reserved Capacity Features
io.net's reserved capacity is designed for real-world flexibility:
1. GPU Type Flexibility:
Switch between GPU types within your commitment:
- Reserved 10x H100 capacity, but only need 5x H100 this month?
- Use remaining capacity as 10x A100 or 50x RTX 4090 (equivalent value)
- No penalties, no approval needed
Example:
- Commitment: $15,000/month reserved capacity
- Month 1: 5x H100 ($7,920) + 30x RTX 4090 ($3,888) + 5x A100 ($4,320) = Total: $16,128 (within commitment + small overage)
2. Regional Flexibility:
Move GPUs between regions based on needs:
- Train models in US-West (lower latency to your team)
- Deploy inference in EU-West (closer to European customers)
- No region-locking or transfer fees
3. Burst Capacity:
Scale beyond your reserved capacity at on-demand rates:
- Reserved: 20 GPUs
- Need 30 GPUs for a week? Provision 10 more at on-demand pricing
- Pay reserved rates for first 20, on-demand for additional 10
4. Partial Utilization Credit:
If you use less than your commitment, credits roll over:
- Committed to $10,000/month, only used $8,000 in March?
- $2,000 credit applies to April usage
- Credits expire after 3 months (prevents gaming the system while providing flexibility)
Reserved Capacity Use Cases
Use Case 1: AI Startup - Production Inference API
- Workload: Serve LLM inference API for B2B SaaS product
- Requirements: 10x RTX 4090 running 24/7, 99.5% uptime SLA
- Plan: 6-month reserved capacity
- Cost: $0.18/hr × 0.85 (15% discount) = $0.153/hr per GPU
- Monthly: $0.153/hr × 720 hrs × 10 GPUs = $1,102/month
- vs. AWS on-demand equivalent (g5.xlarge): $5,040/month
- Annual savings: $47,256 (78%)
Use Case 2: Research Lab - Multi-Month Model Development
- Workload: Train and fine-tune multiple LLaMA 3 70B variants over 12 months
- Requirements: 8x H100 SXM running 12 hours/day on average, enterprise support
- Plan: 12-month reserved capacity
- Cost: $2.20/hr × 0.80 (20% discount) = $1.76/hr per GPU
- Monthly: $1.76/hr × 360 hrs × 8 GPUs = $5,069/month
- vs. AWS on-demand: $20,102/month
- Annual savings: $180,396 (75%)
Use Case 3: Game Studio - Rendering Farm
- Workload: 3D rendering and AI-assisted asset generation
- Requirements: 40x RTX 4090 running 16 hours/day during production sprints
- Plan: 3-month reserved capacity (aligned with production cycle)
- Cost: $0.18/hr × 0.90 (10% discount) = $0.162/hr per GPU
- Monthly: $0.162/hr × 480 hrs × 40 GPUs = $3,110/month
- vs. building on-premise render farm: $15,000-20,000 upfront + $2,000/month operating costs
- 3-month TCO savings: $12,000-15,000 (avoiding capital expenditure)
How to Get Started with Reserved Capacity
Step 1: Analyze Your Usage (Optional but Recommended)
Review past 3 months of GPU usage:
# Get usage report for last 90 days
io billing report --period 90d --format csv --output usage-analysis.csv
Step 2: Contact Enterprise Sales
Email [email protected] with:
- Estimated monthly GPU usage (GPU type and hours)
- Desired commitment length (3, 6, or 12 months)
- Any special requirements (SLA level, compliance needs, multi-region)
Step 3: Receive Custom Quote
io.net provides within 24 hours:
- Reserved capacity pricing based on your usage
- Commitment terms and SLA details
- Contract for review
Step 4: Activate Reserved Capacity
- Sign agreement (typically 1-2 pages, no legal complexity)
- Reserved capacity activates within 24 hours
- Start provisioning GPUs immediately with reserved rates
No credit checks, no lengthy procurement processes. Average time from inquiry to activation: 2-3 business days.
Reserved Capacity Contract Terms
What's Included:
- ✅ Guaranteed GPU availability (99.5-99.9% SLA)
- ✅ 10-20% discount on on-demand pricing
- ✅ Enterprise support (24/7 email/Discord, dedicated account manager)
- ✅ Priority provisioning during high-demand periods
- ✅ Monthly billing (no upfront payment)
- ✅ Custom monitoring dashboards
- ✅ Quarterly business reviews (12-month plans)
What's Not Included (But Available for Additional Cost):
- Egress beyond 1TB/month ($0.05/GB - same as on-demand)
- Storage beyond included amount ($0.05/GB/month)
- White-glove managed services (model optimization, DevOps support)
Termination Terms:
- After minimum commitment period, cancel with 30 days notice
- No early termination penalties after minimum period
- Early termination during commitment: Pay 50% of remaining commitment value (fair middle ground)
Enterprise Procurement & Compliance
io.net reserved capacity supports enterprise procurement requirements:
Billing & Payment:
- Monthly invoicing (NET-30 payment terms available)
- Annual invoicing option with 5% additional discount
- Wire transfer, ACH, credit card accepted
- Multi-year purchase orders accepted
Security & Compliance:
- SOC 2 Type II audit in progress (available Q3 2026)
- GDPR compliant
- HIPAA support via Confidential Compute (separate engagement)
- Custom data residency requirements (enterprise plans)
Vendor Onboarding:
- W-9 and insurance certificates provided
- Standard MSA (Master Service Agreement) available
- Custom contract terms for Fortune 500 customers
Related Questions
Can I convert my on-demand usage to reserved capacity mid-month?
Yes. If you've been using on-demand and want to switch to reserved capacity, contact [email protected]. We can retroactively apply reserved pricing to the current month's usage once your reserved capacity contract is active. This ensures you don't lose savings by starting mid-billing-cycle. Reserved capacity activation typically takes 24-48 hours.
What happens if I exceed my reserved capacity?
You simply pay on-demand rates for usage beyond your commitment. For example, if you have 20 GPUs reserved but provision 25, the first 20 are billed at reserved rates and the additional 5 at on-demand rates. There are no penalties or overage fees - on-demand rates are already competitive. Many teams reserve 70-80% of typical usage and handle spikes with on-demand provisioning.
Can I upgrade or downgrade my reserved capacity during the commitment period?
Yes, with some limitations. You can upgrade (increase commitment) anytime by contacting your account manager - the new discounted rate applies immediately. Downgrading (reducing commitment) is allowed once per contract term and requires 30 days notice. The goal is flexibility while preventing abuse of the reserved discount structure. Most teams start conservatively and upgrade as usage grows.
How does reserved capacity work with auto-scaling?
Auto-scaling works seamlessly with reserved capacity. Set your reserved capacity as the "base" cluster size, and auto-scale above that as needed. For example: reserve 10 GPUs for your baseline inference load, and auto-scale up to 30 GPUs during traffic spikes. The base 10 are billed at reserved rates, additional auto-scaled GPUs at on-demand rates. This optimizes cost while maintaining performance during peak demand.
Does io.net offer spot + reserved hybrid pricing like AWS?
io.net doesn't have spot instances (all instances are stable on-demand). However, the hybrid approach is similar: use reserved capacity for predictable baseline workloads, and on-demand for variable/burst workloads. Since io.net's on-demand is already 50-70% cheaper than AWS spot, you get spot-like economics without interruption risk. Reserved capacity adds another 10-20% savings on top of already-low on-demand pricing.
Lock In Enterprise-Grade Savings
Get guaranteed GPU availability with the industry's best pricing:
- 10-20% additional discounts on already low on-demand rates
- 3-month minimum terms - No 1-3 year lock-in like AWS/Azure
- $0 upfront payment - Monthly billing with NET-30 terms
Contact enterprise sales → or calculate your savings →
Last updated: April 2026 | Reserved capacity pricing and terms subject to minimum commitments and contract approval.
