Primis Protocol

Routing on-chain capital into AI compute

A decentralized system that uses yield to reduce the cost of AI infrastructure.

Capital Yield Compute Revenue

The Problem

AI is expensive for one reason:
compute pricing is static.

The Insight

Compute doesn't need to be cheap.
It needs to be subsidized.

No leverage. No emissions. Just value reallocation.

The Solution

Primis is an inverse-subsidization
engine for AI compute.

Capital pooled
Base yield generated
Yield subsidizes compute
Usage fees return

Primis is not a cloud.
It's a pricing layer on top of compute.

The Flywheel

Why this compounds

Capital Yield Compute Revenue
  • Capital wants higher, safe yield
  • AI builders want lower, predictable costs
  • Each side funds the other
  • Demand ↑ → Revenue ↑ → Yield ↑ → Capital ↑

Neither side needs to understand the other.

Who It's For

Two users, one system

Capital Providers

Stake capital

Earn base yield + compute usage income

Exposure to AI demand without running GPUs

AI Builders

Buy compute at lower, predictable prices

No contracts, no minimums

Pay per GPU-hour / batch job

Real users. Real incentives.

Revenue Model

Primis monetizes
compute usage.

Why Now

This wins today

Primis exists because this inefficiency is now visible.

What We Built

Capital Provider View

primis.protocol/portfolio
Connected
Total Earned
+0.0 SOL
Yield from staking +0.0 SOL
Revenue from compute +0.0 SOL
Effective APY
Staked
0 SOL
Network Util.
63%

AI Builder View

terminal
$ primis jobs submit --type fine-tune --model llama-3-8b

Allocating GPUs...
Subsidy applied

{
"job_id": "job_8f7d3k2l",
"status": "running",
"estimated_cost": "$12.40"
}

Job running.
Job Status
GPU allocation 4x A100
Est. completion ~2h 15m
Subsidy rate Applied

Why We Win

Primis is different

GPU marketplace Cloud alternative Yield farm

Primis is an economic system.

Primis turns yield into infrastructure —
and infrastructure into yield.

We're building the pricing layer for AI compute.

to navigate · F fullscreen