Pool Token Overview

The Silicon pool token ($silGPUa) is an ERC-20 token representing a fractional ownership claim on GPU servers and their associated cash flows derived from AI compute workloads. The pool contract holds NFTs representing ownership and cash flow rights to individual GPU servers, as well as USDC received as operational revenue. Token holders realize returns through appreciation in the net asset value of the pool as USDC revenue accumulates and the underlying GPU assets generate cash flows. No staking is required.

GPU NFT Structure

The pool token extends Silicon's existing NFT tokenization framework to enable fractional ownership of GPU infrastructure with minimum investments as low as one dollar. The pool holds a portfolio of NFTs, each representing a specific physical server deployed in a datacenter with a unique serial number and discrete monthly revenue stream.

NFT-based tokenization provides several operational advantages:

Accurate valuation: Net asset value can be tracked on a per-asset basis
Performance monitoring: Individual server performance can be assessed and reported independently
Clean liquidation mechanism: NFTs can be burned from the pool in exchange for USDC equal to or exceeding the minimum liquidation value assigned at initial deposit

Revenue Distribution

Revenue flows through a structured waterfall from cloud platforms to the pool contract:

Revenue Flow

Cloud platforms such as Runpod disburse monthly revenue in USD to datacenter operators who maintain physical infrastructure and ensure server uptime
Datacenter operators remit USDC to the pool contract (1:1 conversion against USD), net of operational fees
Silicon retains a portion of revenue from the pool contract to cover US tax obligations on 1099-reportable income from platforms like Runpod

Revenue Allocation

The following allocation applies to gross revenue generated by pool assets:

51.4% to pool token contract
20.0% to Runpod (cloud marketplace provider)
13.2% to datacenter for electricity and operational expenses
5.0% to Silicon for protocol management
10.4% for taxes (net of depreciation and other deductions)

Demand Sources

AI compute demand primarily originates from Runpod, a leading GPU cloud platform specializing in on-demand inference and training workloads. Runpod operates multiple service tiers with differentiated pricing structures. Silicon's datacenter partners maintain the operational standards required for Runpod Secure Cloud certification, which ensures servers in the silGPUa pool command premium hourly rates relative to alternative platforms and relative to Runpod community cloud.

Silicon and its datacenter partners plan to diversify revenue sources to include direct customer agreements, long-term fixed-price contracts, and additional on-demand cloud platforms offering competitive pricing.

Net Asset Value (NAV)

Calculation Methodology

The pool token's Net Asset Value (NAV) represents the liquidation value of all pool holdings, including both GPU server assets and USDC reserves:

USDC: Valued at 1:1 with US dollars
GPU Servers: Valued at acquisition cost less accumulated depreciation

NAV increases as USDC revenue accrues to the pool. GPU server values decline according to a linear depreciation schedule over the term of each service agreement.

Depreciation Schedule

Depreciation is calculated on a straight-line basis over the service agreement lifecycle. Each server has a guaranteed minimum liquidation value at the end of its service agreement. These liquidation values are negotiated at the time of the original purchase and generally reflect a “floor price” for the value of the asset at the end of its service agreement. It’s possible assets could be liquidated at a premium (resulting in an additional gain to the pool contract).

Example: A server valued at $100,000 with a 36-month service agreement and a guaranteed end-of-life liquidation value of$ 20,000 would depreciate at $2,222 per month ($ 80,000 ÷ 36 months).

Redemption Mechanisms

USDC Redemption

Following the pool token launch, token holders may realize gains through two methods:

Swap on AMM: Sell tokens via automated market makers (AMMs)
Direct redemption: Redeem tokens for USDC from the pool contract at NAV less a withdrawal tax

Direct redemptions are executed at the current NAV per token. For example, if NAV per token is $1.20, redeeming one pool token yields 1.2 USDC, less applicable redemption fees.

Withdrawal Tax (Bonding Curve)

A dynamic withdrawal tax protects USDC reserves when the ratio of USDC to total pool value is low. The tax formula discourages rapid outflows based on the pool's USDC composition:

Pool State	Approximate Tax
Pool is 80% USDC	~1-2%
Pool is 50% USDC	~5-8%
Pool is 20% USDC	~15-20%
Pool is nearly empty	Up to 25%

Mathematically, the tax paid on a given withdrawal is:

\bar{t}(u_0, w) = T_{\max} \cdot \beta^2 \cdot \frac{1 - u_0}{w} \cdot \left( \frac{1}{\beta + \frac{u_0 - w}{1 - w}} - \frac{1}{\beta + u_0} \right)

where:

$u_0$ is the fraction of the Pool NAV that is USDC
$w$ is the fraction of the pool the user wishes to withdraw
$T_{\max}$ is the maximum tax rate
$\beta$ is a curve parameter

This protects remaining holders from being left with illiquid NFTs while others exit with all the cash.

Redemption Purpose

The redemption mechanism serves as a liquidity backstop in two scenarios:

Pool tokens trade below NAV on secondary markets
Large redemptions exceed the absorption capacity of secondary market liquidity

NFT Redemption

Token holders may redeem individual GPU server NFTs directly from the pool by burning the requisite quantity of pool tokens based on the NFT's current NAV.

Capital Deployment

Idle USDC reserves may be deployed periodically to acquire additional GPU server assets at Silicon's discretion. Future governance mechanisms may enable token holder participation in asset acquisition decisions. Initially, Silicon oversees capital allocation to ensure only high-quality servers with strong return profiles are added to the pool.

Future Pool Token Issuance (Inflation)

Issuance Cadence and Purpose

Silicon will periodically conduct additional pre-deposit campaigns to raise capital for acquiring new GPU servers. These campaigns are essential to sustain pool returns over time, as existing servers have finite operational lifecycles ranging from 12 to 36 months. Each campaign finances a discrete server cluster and mints new pool tokens in exchange for deposited USDC.

NAV Pricing Mechanism

Initially new pool tokens will be issued at the prevailing NAV per token at the time of issuance. This pricing floor ensures that new capital does not dilute existing token holders.

Example: If the NAV of $silGPUa is$ 1.20 per token at the time of a Cluster 02 campaign, participants must deposit 1.2 USDC to receive one pool token.

Auction Mechanism

As the pool token achieves sufficient liquidity and price discovery, Silicon may transition from fixed NAV pricing to an auction-based issuance model. In this model, participants bid for newly minted tokens with NAV per token serving as the reserve price (minimum price the auction will accept). This approach captures incremental value when market demand exceeds supply while maintaining protection for existing token holders.

The transition to auction-based issuance will occur gradually as the pool reaches minimum thresholds for total value locked (e.g. $50M+) and spot AMM market trading activity (e.g.$ 20K depth at +/- 1%).

Asset Diversification Strategy

Future issuance campaigns will diversify pool holdings across multiple GPU architectures to balance risk and optimize returns. The pool will acquire both established models with proven economics and latest-generation hardware offering premium performance capabilities.

Model Selection Criteria

Silicon evaluates GPU models across three primary dimensions:

Utilization rates: Historical and projected demand for specific architectures across cloud platforms and direct customer workloads
Price-to-performance ratio: Acquisition cost relative to hourly revenue potential and operational efficiency
Depreciation profile: Expected residual value at end-of-life based on secondary market liquidity and technological obsolescence risk

Portfolio Composition

The pool will maintain exposure across GPU generations:

Mature architectures (e.g., H100): Lower acquisition costs, established demand patterns, and predictable depreciation curves. These assets typically offer stable cash yields with moderate capital appreciation risk.
Current-generation hardware (e.g., B200): Premium hourly rates driven by cutting-edge performance characteristics. Higher acquisition costs are offset by superior utilization rates and extended demand visibility as AI workloads scale in compute intensity.

Diversification across model types reduces concentration risk while capturing returns from both high-yield mature assets and high-growth frontier hardware. Silicon will adjust portfolio composition dynamically based on prevailing market conditions, hardware availability, and demand trends across compute platforms.

Service Agreements

GPU servers in the silGPUa pool operate under standardized service agreements executed between Silicon, GPU owners, and datacenter providers such as FarmGPU. These agreements establish terms for hosting, operations, revenue sharing, and asset management throughout each server's operational lifecycle.

Key Terms

Service Duration

Each GPU NFT includes an end-of-life (EOL) date in its metadata, typically 36 months from deployment. Service agreements remain effective until the EOL date unless terminated early by either party with appropriate notice.

Uptime Commitment

Datacenter providers maintain a minimum 95% uptime guarantee. Equipment must remain powered, network-connected, and available for compute workloads. Scheduled maintenance and certain qualifying hardware failures are excluded from uptime calculations.

Physical Custody

Datacenter providers hold physical custody of equipment solely as bailees. Legal and economic ownership remains with GPU owners as represented by GPU NFTs at all times. Providers may not pledge, transfer, sell, or otherwise encumber the equipment.

Operational Responsibilities

Providers are responsible for:

Installation, configuration, and initial deployment
Ongoing maintenance and firmware updates
Active monetization through cloud platforms and direct customer relationships
Continuous monitoring with real-time performance reporting

Insurance and Risk

Providers maintain commercial insurance covering full replacement costs and bear risk of loss during the period equipment is under their control, excluding uninsured Force Majeure events. In the event of hardware failure, providers must replace equipment with equivalent units within 10 business days.

End-of-Life Process

At the EOL date, GPU owners may elect either:

Physical redemption: Reclaim physical equipment with 60 days advance notice
Liquidation: Equipment is sold at fair market value with proceeds distributed net of liquidation costs and provider fees (typically 10% of liquidation value)

Transfer Rights

GPU NFTs and all associated service agreement rights transfer automatically with on-chain NFT transfers. The current wallet holder is recognized as the rightful owner for all purposes, including revenue payments, redemptions, and official notices.

Mission & Vision Points