Providers & Leases

Learn how to choose providers and manage leases on Akash Network.

Providers are the backbone of Akash Network - they’re the data centers that host your applications. Understanding how providers work, how to evaluate them, and how to manage leases is essential for successful deployments.

What is a Provider?

A provider is a data center or infrastructure operator that offers compute resources on Akash Network.

Providers offer:

• CPU and GPU compute
• Memory (RAM)
• Storage (persistent and ephemeral)
• Network connectivity
• Optional features (IP leases)

Providers compete on:

• Price
• Performance
• Reliability
• Location
• Features

How Providers Work

Provider Registration

Providers register on Akash blockchain:

1. Publish their resource capacity
2. Set pricing models
3. Define attributes (location, features, certifications)
4. Run provider software to monitor orders

Note: Providers lock up AKT only for the duration of active bids and leases, not as permanent collateral.

Audited Providers

Some providers have undergone security audits and are marked as “audited” providers. You can restrict your deployments to only audited providers using the signedBy field in your SDL for enhanced security and compliance requirements.

Bidding Process

When you create a deployment:

1. Order posted - Your resource requirements published on blockchain
2. Providers evaluate - Each provider checks if they can fulfill the order
3. Automated bidding - Qualifying providers submit bids automatically
4. You choose - Review bids and select preferred provider
5. Lease created - Agreement finalized on blockchain

Bidding happens fast: Most bids arrive within 30-120 seconds

Provider Attributes

Providers publish attributes to help you make informed decisions:

Location

region: us-west
datacenter: equinix-sv15
city: San Jose
country: US

Why it matters:

• Latency to your users
• Data sovereignty requirements
• Disaster recovery planning

Capabilities

Providers advertise capabilities through attributes:

feat-persistent-storage: true
feat-endpoint-ip: true

Common feature attributes:

• feat-persistent-storage - Can provide persistent volumes
• feat-endpoint-ip - Offers dedicated IP addresses
• GPU capabilities are advertised via capabilities/gpu/vendor/<vendor>/model/<model> attributes

See the Provider Attributes guide for complete list.

Hardware

cpu: AMD EPYC 7643
gpu: NVIDIA RTX 4090
network: 10Gbps

Specifications:

• CPU model and generation
• GPU models available
• Network capacity
• Storage type (HDD/SSD/NVMe)

Certifications

audited: true
tier: premium
uptime-guarantee: 99.9%

Trust indicators:

• Security audits
• Tier classification
• SLA commitments
• Community reputation

Choosing a Provider

Evaluation Criteria

1. Price

Check:

• Bid price per block
• Total estimated monthly cost
• IP lease costs (if needed)
• Persistent storage costs (if needed)

Strategy:

• Lowest isn’t always best
• Balance cost with reliability
• Consider total cost of ownership

2. Reputation

Check:

• Provider uptime history
• Community feedback in Discord
• Active lease count
• Time in operation

Red flags:

• Brand new provider (no history)
• Frequent downtime reports
• Negative community feedback
• Suspiciously low prices

3. Location

Consider:

• Proximity to your users
• Compliance requirements (GDPR, data residency)
• Latency requirements
• Backup/DR location diversity

Example:

• US users US-based provider
• EU users with GDPR EU-based provider
• Global app Multiple providers in different regions

4. Features

Match your needs:

• Need persistent storage? Check feat-persistent-storage: true
• Need dedicated IP? Check feat-endpoint-ip: true
• Need GPU? Check capabilities/gpu/vendor/<vendor>/model/<model> attributes
• Need high bandwidth? Check network capacity attributes

5. Performance

Consider:

• CPU generation and model
• Storage type (NVMe > SSD > HDD)
• Network speed
• GPU models (for GPU workloads)

Understanding Bids

Bid Components

When you receive a bid, it includes:

Provider: akash1abc...xyz
Price: 8,500 uakt/block
Location: US-West
Attributes:
  - feat-persistent-storage: true
  - location-region: us-west
  - tier: community

Price Calculation

Bid price is per block (~6 seconds):

Bid: 10,000 uakt/block

Hourly cost: 10,000 × 600 blocks = 6,000,000 uakt = 0.006 AKT
Daily cost: 0.006 × 24 = 0.144 AKT
Monthly cost: 0.144 × 30 = 4.32 AKT

At $2.50/AKT: ~$10.80/month

Comparing Bids

Example bids for same deployment:

Decision factors:

• Provider A: Good price, all features, good location
• Provider B: Slightly more, no IP lease
• Provider C: Cheapest, but no persistent storage (dealbreaker if you need it)
• Provider D: Most expensive, but might be worth it for premium tier/reputation

Lease Management

Creating a Lease

Process:

1. Review available bids
2. Select preferred provider
3. Submit lease creation transaction
4. Wait for lease confirmation (~6 seconds)
5. Upload manifest to provider

Console: Click “Accept” on your chosen bid CLI:

provider-services tx market lease create \
  --dseq <deployment-id> \
  --provider <provider-address> \
  --from <wallet>

Active Lease

Once lease is active:

Your responsibilities:

• Monitor escrow balance
• Pay gas fees for updates
• Close lease when done

Provider responsibilities:

• Run your containers
• Maintain uptime
• Process manifest updates
• Accept escrow payments

Lease States

active - Running normally

• Containers active
• Escrow being paid
• Services accessible

insufficient_funds - Escrow depleted

• Containers stopped
• Lease still exists but inactive
• Cannot restart without new deployment

closed - Terminated

• You closed it
• Provider closed it
• Permanent end state

Provider Communication

Manifest Upload

After lease creation, you send the manifest to the provider:

What it includes:

• Full service specifications
• Environment variables
• Port configurations
• Image pull credentials (if private)

How it’s sent:

• Direct HTTPS to provider endpoint
• Authenticated with JWT or mTLS certificate
• Not stored on blockchain

Provider endpoint:

https://provider.akash.network:8443/

Service Endpoints

Provider gives you URLs to access services:

https://5g8qj7kl3m-8080.provider.akash.network

Format:

• Unique subdomain per service
• Provider’s domain
• Port number
• HTTPS by default

Logs and Status

Access logs:

• Console: View logs in UI
• CLI: Query provider directly
• SDK: Programmatic log access

Check status:

• Service health
• Container state
• Resource usage
• Recent events

Provider Selection Strategies

Strategy 1: Lowest Price

When to use:

• Cost-sensitive workloads
• Non-critical applications
• Development/testing

Approach:

1. Sort bids by price
2. Accept lowest bid
3. Monitor performance

Risk: May sacrifice reliability for cost

Strategy 2: Best Reputation

When to use:

• Production applications
• Business-critical workloads
• Long-running services

Approach:

1. Research providers in Discord
2. Check uptime history
3. Prioritize established providers
4. Accept higher cost for reliability

Benefit: Higher reliability, better support

Strategy 3: Geographic Optimization

When to use:

• Latency-sensitive applications
• Compliance requirements
• Multi-region deployments

Approach:

1. Filter by location
2. Choose closest to users
3. Consider data residency laws

Example: EU app with GDPR EU-based provider only

Strategy 4: Feature Requirements

When to use:

• Specific technical needs
• GPU workloads
• Persistent storage requirements

Approach:

1. Filter by required attributes
2. Eliminate providers without needed features
3. Choose among qualified providers

Example: Need NVIDIA A100 GPU Filter for that specific model

Multi-Provider Deployments

For high availability, deploy to multiple providers:

Load Balancing

       ┌──────────────┐
       │ Load Balancer│
       │ (External)   │
       └───────┬──────┘
               │
       ┌───────┴────────┐
       │                │
   ┌───▼───┐       ┌───▼───┐
   │Prov. A│       │Prov. B│
   │US-West│       │US-East│
   └───────┘       └───────┘

Benefits:

• Redundancy
• Geographic distribution
• Load distribution
• Zero-downtime updates

Challenges:

• More complex setup
• Higher costs
• Need external load balancer
• Shared state management

Provider Issues & Troubleshooting

Provider Downtime

Symptoms:

• Services inaccessible
• Logs show “connection refused”
• Manifest queries fail

What to do:

1. Check provider status in Discord
2. Wait 15-30 minutes (may be temporary)
3. If persistent, close and redeploy to different provider

Prevention: Multi-provider deployment

Performance Issues

Symptoms:

• Slow response times
• High latency
• Resource constraints

Diagnosis:

1. Check logs for resource limits
2. Monitor resource usage
3. Compare with provider’s advertised specs

Solutions:

• Increase resources (requires redeploy)
• Move to different provider
• Optimize application

Provider Not Responding

Symptoms:

• Cannot upload manifest
• Cannot query logs/status
• No response from provider API

What to do:

1. Verify provider endpoint
2. Check certificate/authentication
3. Try from different network
4. Contact provider in Discord
5. If no response within 24h, close lease and redeploy

Provider Economics

Provider Revenue Model

Providers earn from:

• Base compute charges (per block)
• IP lease fees
• Persistent storage fees
• Premium tier pricing

Provider Costs

Providers pay for:

• Hardware and infrastructure
• Electricity
• Maintenance
• Staff/Operations
• AKT token stake

Competitive Dynamics

Market forces:

• More providers = lower prices
• Unique features = premium pricing
• Reputation = price premium
• Location scarcity = higher prices

Example:

• GPU providers can charge 3-10x CPU-only pricing
• EU providers may charge premium for GDPR compliance
• Established providers command loyalty premium

Best Practices

Selection

DO:

• Research providers before accepting bids
• Check community feedback
• Verify required features are available
• Test with small deployment first
• Document your provider choices

DON’T:

• Always choose cheapest bid
• Ignore provider reputation
• Skip attribute verification
• Deploy production to unknown providers

Management

DO:

• Monitor lease health regularly
• Keep provider contact info
• Have backup provider list
• Plan migration procedures
• Test failover scenarios

DON’T:

• Ignore provider downtime warnings
• Let escrow run out unexpectedly
• Deploy without backup plan
• Assume all providers are equal

Communication

DO:

• Join provider Discord channels
• Report issues constructively
• Provide clear error information
• Be patient with provider responses

DON’T:

• Expect 24/7 instant support
• Be hostile or demanding
• Ignore provider announcements
• Skip troubleshooting steps

Provider Ecosystem

Provider Types

Tier 1 - Established

• Long operating history (6+ months)
• High uptime (>99%)
• Active community presence
• Professional support

Tier 2 - Growing

• Moderate history (1-6 months)
• Good uptime (>95%)
• Building reputation
• Responsive to issues

Tier 3 - New/Experimental

• Recently launched (<1 month)
• Unknown reliability
• Lower prices to attract users
• Higher risk

Finding Providers

Resources:

• Akash Stats - Provider list and metrics
• Discord - Provider announcements

Advanced Topics

Provider APIs

Providers expose REST/gRPC APIs:

• Manifest submission
• Status queries
• Log streaming
• Shell access (some providers)

Provider Attributes Deep Dive

Custom attributes for filtering:

placement:
  requirements:
    attributes:
      - key: location-region
        value: us-west
      - key: tier
        value: community
      - key: feat-persistent-storage
        value: true

Provider Economics for Deployers

Maximize value:

• Batch similar workloads to one provider
• Commit to longer leases (negotiate if possible)
• Use off-peak pricing (some providers)
• Bundle services for discounts

Related Topics

• Deployments & Lifecycle - Understanding deployments
• IP Leases - Dedicated IP addresses
• Persistent Storage - Data persistence

Questions about providers? Ask in Discord #providers channel!