AO FUND. — Pitch Deck | Decentralized AI Infrastructure

×

τ

Bittensor

Wrong password

Introduction

What Is Bittensor?

The open, permissionless decentralized network for intelligence.

Centralized AI

Alphabet, OpenAI, Anthropic —
one company controls everything.
Single point of failure.

Decentralized AI — Bittensor

Bittensor is the base network.
Subnets (usecases) are built on top.
No single point of failure.

Open Access

No gatekeepers

Anyone can contribute AI models and earn rewards. No approval process — just performance.

Global Scale

GPUs worldwide, one network

Bittensor connects compute globally. Scale is limited only by participation, not by any single budget.

AI Marketplace

Best models earn TAO

Subnets compete across text, image, code & more. The best models are rewarded with TAO.

Incorruptible

No one can tamper with it

All results are verified on-chain. No single entity can manipulate rankings, censor models, or alter rewards.

The Protocol Analogy

Bittensor Network Is a Protocol — Just Like the Internet

The internet is an open protocol on which websites are built. Bittensor is an open protocol on which AI is built. Both are open-source infrastructure that anyone can build on.

The Internet

Built on TCP/IP Since 1983

Every website, app, and service runs on the same open protocol. No single company owns it.

Bittensor

Built on Bittensor Base Protocol Since 2021

Every AI model, subnet, and application runs on the same open protocol. No single company owns it.

Why It Matters

Open Protocols Win Long-Term

TCP/IP outlasted CompuServe & AOL. Bittensor is positioned to outlast centralized AI monopolies.

How Bittensor Works — The Flywheel

How Does Value Flow Through the Network?

Think of it like

Uber for AI Models

Just like Uber matches riders with drivers, Bittensor matches anyone who needs AI with the best models worldwide — and rewards the creators.

Why it works

Only the Best Get Paid

There is no quota or subscription. Models are ranked by performance. The best earn TAO rewards, the rest earn nothing. This drives constant improvement.

Why TAO gains value

More Use = More TAO Recycled

Transaction fees and neuron-registration costs are recycled into unissued supply, taking TAO out of circulation, and the Dec 2025 halving cut new issuance in half. More demand, slower net supply growth.

Why Bittensor?

Jensen Huang Explains the Bittensor Model

In this clip, NVIDIA CEO Jensen Huang describes the exact decentralized AI model that Bittensor is building — an open network where thousands of contributors compete to deliver the best intelligence, rewarded by performance rather than controlled by a single corporation.

The Growth Gap — Why Now?

Decentralized Training Is Scaling 4x Faster Than the Frontier

Key figures from Epoch AI’s decentralized-training analysis (Dec 2025), highlighted by Jack Clark (Anthropic co-founder) in Import AI #439.

Centralized AI (5x growth p.a.)

Decentralized training (20x growth p.a.)

2026 — Inflection Point

Critical Mass Reached

Decentralized compute power (10 GW+) crosses the threshold where network effects become exponential.

Efficiency Gain

1.5x More Efficient

Decentralized training leverages global distribution, achieving 1.5x efficiency gains over isolated data center clusters.

The Window

Trend Parity ~2031

At 20x vs 5x a year, the ~1,000x compute gap shrinks ~4x annually; Epoch estimates it would close in ~5.5 years if both trends held — we are buying tomorrow's infrastructure at today's prices.

Traction & Institutional Validation

From Theory to Global Infrastructure

Scaling

Network Expansion

128

128 Subnets diversifying across text, image, bioinformatics & robotics.

Text Image Bioinformatics Robotics

Market

Market Cap

~$3B

Growing institutional interest in the leading decentralized AI network. Early-stage valuation with significant upside potential as adoption accelerates.

Early Stage Institutional High Upside

Tokenomics

Disinflationary Supply

21M

TAO is hard-capped at 21 million tokens with Bitcoin-style halvings. The Dec 2025 halving cut new issuance in half, to ~3,600 TAO/day (~11% a year); the next halving (est. late 2029–30) halves it again.

Hard Cap 21M Disinflationary Bitcoin Model

Scenario Analysis - Bittensor Market Cap 2031

Four Scenarios for 2031

Bear Case

Niche Project

$1.5B

~0.5x from $3B

Bittensor remains a niche project driven by tech enthusiasts and never achieves mainstream adoption. Limited commercial traction, no institutional interest — the network never takes off.

Probability: ~25%

Base Case
AI Infrastructure Standard
$30B
~10x from $3B
Bittensor establishes itself as the go-to decentralized AI layer. Enterprise adoption accelerates, subnet revenue grows exponentially, and TAO captures a meaningful share of the AI compute market.
Probability: ~35%

Bull Case
Decentralized Anthropic
$300B
~100x from $3B
Bittensor approaches Anthropic’s Feb 2026 valuation ($380B; latest $965B, May 2026). As the dominant open AI network, institutional capital flows in via ETFs, and a broad base of high-revenue subnets drives massive TAO demand.
Probability: ~30%

Super Bull Case
The Decentralized OpenAI
$500B+
~167x from $3B
TAO becomes the reserve asset for decentralized intelligence — the “Bitcoin of AI.” Bittensor approaches the scale of today’s leading AI labs (OpenAI $852B, Anthropic $965B) as the global open-source alternative. Indispensable AI infrastructure.
Probability: ~10%

Baseline: $3B market cap (rounded down from ~$3.5B, 30 Sep 2026) | OpenAI: $14B (2021) → $852B (Mar 2026), ~60x in 5 years | Anthropic: $4.1B (2023) → $965B (May 2026), ~235x in 3 years | 21M hard cap | Multiples on market cap, before emission dilution

The 3-Pillar Fund Strategy

Risk Mitigation Through Protocol Exposure + Alpha Generation + Complementary Infrastructure

Core Position

55–75%

The Reserve Asset

Strategic accumulation of TAO as core holding
Focus on long-term value appreciation
Positioning TAO as the “Bitcoin of AI”
Downside protection through systematic rebalancing & hedging instruments
Technical risk management (Multi-Sig Wallet)

Alpha Generation

20–35%

Subnet Alpha

Active management & VC-style investments in early-stage subnets
SN64 / Serverless Compute
SN100 / Agentic Coding
Early access to the strongest emerging sub-ecosystems
Technical risk management (Multi-Sig Wallet)

Complementary Infrastructure

0–15%

Decentralized AI & Compute

Opportunistic satellite — built only as opportunities arise
Decentralized AI & compute beyond Bittensor
Compute, Inference, Training & Data Networks
Complements the Bittensor core position
Technical risk management (Multi-Sig Wallet)

Our Edge

Deep technical expertise combined with direct ecosystem access. Our team evaluates subnet architecture, tokenomics & mechanism design before every capital deployment — ensuring conviction-driven allocation.

Staking — The Hidden Return

How Staking Rewards Cover All Fund Costs — And Then Some

The Bittensor network rewards participants who secure and validate the network — similar to earning interest for providing capital to a bank. By staking our holdings, we lock them into the network and in return receive a continuous yield (~10% APY, blended nominal*) on top of any price appreciation.

Gross Staking Yield (~10% blended nominal*, compounded)

+10.0%

Y1

+21.0%

Y2

+33.1%

Y3

+46.4%

Y4

+61.1%

Y5

+77.2%

Y6

Staking Covers Everything

Both Bittensor pillars are actively staked — root staking on the base protocol and subnet staking on alpha positions generate a blended nominal ~10% APY*.

Of that, ~3% funds all management & operations — meaning 0% management fee for LPs. A 20% performance fee is applied to the staking yield (~2%), with the remaining ~5% compounding over the 6-year fund lifecycle.

Gross Staking

+77.2%

Mgmt & Ops

−25.9%

Perf Fee (20%)

−17.3%

Net for LPs

+34.0%

This +34.0% net staking yield flows directly into the portfolio.

*Blended nominal yield: root staking pays ~5% p.a. in TAO; subnet (alpha) staking pays more, in subnet tokens counted at nominal value (Sep 2026). Rewards are newly issued tokens (TAO supply grows ~11% p.a.), so they offset dilution only in part (root ~5% vs ~11% supply growth), and alpha rewards carry subnet-token price risk. Not a guaranteed return.

Fund Terms & Team

Your Partner for the AI Infrastructure Decade.

The Base Layer Investment

55–75% Bittensor (Core) Exposure

High-Conviction Use Cases

20–35% Subnet (Alpha) Exposure

Complementary Infrastructure

0–15% Decentralized AI & Compute

Deep Ecosystem Access

Technical Due Diligence

Constant on-chain analytics and deep technical protocol-level monitoring. Every position is technically vetted.

Downside Protection

Active Risk Management

Dynamic rebalancing between FIAT, Bittensor, subnet and complementary infrastructure positions based on on-chain signals and constant technical monitoring.

Tax-Neutral Rebalancing

Optimize Without Tax Events

Our structure (GmbH & Co. KG) allows rebalancing between Bittensor and Subnets without triggering capital gains for our LPs.

Staking Yield

Passive Income from Day 1

Bittensor staking generates min. 3% APY — covering all fund operations. Zero management fee for investors.

Parameter	Details
Structure	GmbH & Co. KG
Tax Optimization	Fund: ~25% CGT \| LP Distribution: ~1.5% eff.*
Management Fee	0%
Carry / Super Carry	20% \| 30% at 20x
High-Water Mark	✓ Yes
Fund Admin	Tokenstreet GmbH
*~1.5% applies to corporate LPs meeting the participation thresholds (§8b KStG: ≥10% for corporate tax, ≥15% for trade tax). Other LPs, incl. private individuals, are taxed at their own rates (e.g. 26.375% flat tax). Indicative, not tax advice.

Johannes de Waal

General Partner

Investor since 2015 with deep understanding of game theory, decentralized systems & cryptographic networks. Transitioned from private investor to Solo GP. Strong network of researchers & engineers providing direct ecosystem access.

To be announced

Senior Software Engineer & Advisor

Fr. Senior Software Engineer @ Greenfield Capital, Europe’s largest Crypto VC | Fr. Senior Software Engineer @ Coinbase | Fr. Head of Engineer @ TokenAnalyst | Researcher in the Computational Intelligence Group | Master of Science in Software Engineering