How DePINly estimates Akash provider earnings. Last verified 2026-05-22.
TL;DR
Akash is a decentralized compute marketplace: tenants submit deployment specs and providers (you) win leases via reverse auction. Earnings are demand-driven (you earn when your capacity is leased). The product's edge is a live AWS comparison badge — Akash typically prices 60-85% below AWS.
Earning mechanics
- You provide CPU/RAM/GPU; tenants pay an hourly lease.
- Protocol take rate by currency (Akash GitHub #147): AKT → provider keeps 96%; USDC → provider keeps 80%.
- Net = lease revenue × provider share − electricity.
Formula (sketch)
tenantPay = cores×AKASH_CPU/hr + GB×AKASH_RAM/hr + AKASH_GPU/hr(type)
gross = tenantPay × hours × utilization × providerShare(currency)
awsEquiv = cores×AWS_CPU/hr + GB×AWS_RAM/hr + AWS_GPU/hr(type) // comparison badge
Assumptions & confidence
| Assumption | Basis | Confidence |
|---|---|---|
| AWS reference pricing | aws.amazon.com on-demand, us-east-1 (2024 snapshot) | 🟢 |
| Provider share 96%/80% | Akash docs / GitHub #147 | 🟢 |
| Akash tenant pricing | Akash console + tenant reports | 🟡 ESTIMATE |
| Utilization | user input — demand-variable | 🟡 |
NOT modeled
Bid competition dynamics, provider attributes/audited tiers, downtime penalties, AKT price volatility. GPU leases are the main earner and most demand-sensitive.
Confidence rating
🟡 — Fee split + AWS prices are sourced; Akash tenant prices are market estimates and income depends on lease demand.
Sources (verified 2026-05-22)
Akash docs (provider economics, fee distribution) · AWS EC2 on-demand pricing · Lambda Labs / Vast.ai GPU reference · CoinGecko (akash-network).
Code
src/lib/calculators/akash-rewards.ts · src/components/calculators/akash/AkashCalculator.tsx · src/app/calculate/akash/page.tsx
These figures are point estimates inside a wide band and depend on local demand, hardware, uptime and token price. They may be wrong in either direction. Nothing here is financial advice — always do your own research.
Methodology updated 2026-05-22 · View source on GitHub →