Oysterworld

Web 4.0

Physical Intelligence

The open sensor network that gives AI agents eyes in the physical world. 25K nodes deployed. $4M revenue. Zero funding.

25K
Nodes
$4M
Revenue
10K
Pre-Orders
140K+
Community
$0
Funding Raised
Scroll

The Problem

AI Hit a Data Wall

AI consumed all internet data. The next frontier is the physical world — but there’s no infrastructure to collect it.

100%

Internet Data Exhausted

AI Ate the Internet

GPT, Claude, Gemini trained on all public text, images, and video. The next performance leap requires real-world sensor data — which barely exists.

$10B+

Wasted on Proprietary HW

Everyone Building Their Own

Tesla spent $10B+ on custom hardware just for driving data. Boston Dynamics, Figure, Waymo — each building proprietary sensors. Fragmented and unscalable.

0

Unified Data Platforms

No Data Infrastructure

The digital world has AWS. The physical world has nothing. No common platform to collect, verify, and distribute real-world sensor data at scale.

The internet evolved three times. Each time, the prize went to whoever built the new infrastructure first.

The Evolution

From Reading the Web to
Sensing the World

Web 1.0

Read

Static pages. One-way broadcast.

The internet was a library. You could read, but not write. Information flowed in one direction.

Web 2.0

Write

Social platforms. User-generated content.

Everyone became a creator. But the platforms owned your data and monetized your attention.

Web 3.0

Own

Intelligent. Autonomous.

AI promised transformation. But it remained trapped in digital abstraction — AI assistants. Generative models.

Web 4.0

Sense

Physical intelligence. AI meets the real world.

AI agents need to perceive, verify, and act in the physical world. The internet extends beyond screens into reality.

We Are Here

Why Now

The Physical AI Moment:
2026 is the Window

$6B+

Physical AI VC in H1 2025

The biggest capital wave in AI history is flowing into embodied intelligence. Robotics, spatial computing, and physical AI are the new frontier.

110%

YoY AI Glasses Growth

5.1M AI smart glasses shipped in H1 2025. Predicted to hit 10M+ in 2026. Apple entering in 2027. The hardware wave is here.

$50T

Physical AI TAM (NVIDIA)

Jensen Huang declared Physical AI a $50 trillion opportunity at GTC 2025. NVIDIA is investing in Pi, Figure, World Labs, and Wayve.

$39B

Figure AI Valuation

Figure builds bodies. Pi builds brains. World Labs builds simulations. They all need real-world ground truth data — that’s the upstream layer we own.

Figure builds the body. Pi builds the brain. World Labs builds the simulation.
We supply them all with verified real-world data. The upstream layer.

The question isn't whether Physical AI is happening. It's who captures the value.

B2A: Business to Agent

The Customer is
Not Human.
The Customer is AI.

In 2025, AI agents crossed a threshold: they can autonomously browse, purchase, deploy, and operate tools without human intervention. The next wave of infrastructure won't be built for people — it will be built for agents that need to perceive the physical world.

Content agents need to understand what's happening IRL
Robotics agents need verified spatial data for navigation
Healthcare agents need real-time biometric ground truth
Autonomous systems need physics-verified sensor feeds

“We don't sell glasses. We sell eyes to AI.”

Agent Capability Stack

Layer What Agents Need Oysterworld Product
Eyes Physical perception ClawGlasses
Body Physical presence ClawPhones (25K)
Health Biological signals Puffy Health (10K)
Context Ambient awareness AI Pendant
Economy Payment & incentives Oyster AI Platform

Building for AI agents requires a different architecture than building for humans.

How It Works

From Raw Sensor Data to
Physics-Verified Intelligence

1

Capture

Distributed Sensing

Distributed devices collect multi-modal sensory data in real-time: biometrics, spatial signals, environmental conditions, motion patterns.

2

Verify

Physics Filtering

PINNs encode physical and biological laws into the model. Data that violates pharmacokinetic curves or physical constraints is automatically rejected.

3

Harden

Verification-Grade Output

Cross-referencing heart rate, SpO2, and motion makes coherent forgery mathematically impossible. Data becomes enterprise-grade for regulated industries.

4

Serve

Agent-Callable API

Verified data available via API for AI agents, pharma trials, insurance underwriting, and autonomous systems. Open network, not a walled garden.

What We Actually Collect

The Data Intelligence Matrix

Each device captures a different layer of physical reality. Together, they form the most comprehensive real-world data network in existence.

6 Product Lines × 20+ Sensor Types × 25K Deployed Nodes

25K LIVE

AI Smartphone

ClawPhones

Sensors

Accelerometer, Gyroscope, Magnetometer, GPS, Barometer, Ambient Light, Microphone, Camera

Data Types

Motion patterns, Gait analysis, Location traces, Environment noise, Air pressure, Altitude

Who Buys This Data

Urban planners (mobility patterns), Insurance actuaries (activity risk scoring), Transportation analytics, Environmental monitoring agencies

Precedent: Apple uses iPhone accelerometers for Parkinson's research (Movement Disorder Society, 2022). Google uses phone GPS + barometer for traffic prediction. Same sensors, proven methodology.

10K PRE-ORDER

Smart Health Device

Puffy Health

Sensors

PPG (Photoplethysmography), SpO2, Skin Temperature, Accelerometer, Bioimpedance

Data Types

Heart rate, Blood oxygen, Body temperature, Sleep quality, Stress index, Activity levels

Who Buys This Data

Pharma companies (Real-World Evidence studies), Insurance health assessment, Remote patient monitoring, Clinical trial recruitment & validation

Precedent: Same PPG technology as Apple Watch. FDA accepts Real-World Evidence for drug approvals — a $2B+ market growing 15% YoY. PINNs cross-validation achieves 0.35 MAE (clinical threshold: 0.61).

BETA 2026

AI Glasses

ClawGlasses

Sensors

RGB Camera, IMU (6-axis), Ambient Light Sensor, Microphone Array, Proximity Sensor

Data Types

First-person visual, 3D spatial mapping, Head movement tracking, Environmental audio, Object recognition

Who Buys This Data

World Model training data (OpenAI Sora, Google Genie, DeepMind UniSim — all need massive first-person physical-world video), AR/VR spatial datasets, Retail analytics (foot traffic, attention mapping), Autonomous navigation, Robotics perception training

Precedent: World Models are the next frontier — AI that understands physical reality from video. OpenAI, Google, Meta all need first-person footage at scale but can’t collect it themselves. Meta spent $10B+ on Reality Labs. Tesla uses fleet vision for FSD. We crowdsource verified spatial data through 25K+ distributed users.

SHIPPING

Health & Wellness

Oral Health Spray

Product Type

Smart oral care product. Freshening spray with health-optimized formulation connected to user health profiles

Data Types

Usage frequency, Purchase cycles, Health behavior patterns, Compliance tracking, Consumer preference data

Strategic Value

Recurring consumer touchpoint that deepens ecosystem engagement. Health behavior data complements biometric sensor data from Puffy and ClawPhones.

Ecosystem play: Hardware captures biometrics, wellness products capture behavior. Combined: a 360° view of physical health that no single device can provide alone.

Why This Isn't "Just Phone Data"

Apple uses iPhone accelerometers for Parkinson's research. Google uses phone GPS for traffic prediction. These are the same sensors in our devices — with one critical addition: PINNs cross-validates every data point against physical laws, making forgery computationally impossible. No single sensor is the product. The verified, multi-modal network is the product.

The Technology

Hybrid Proof Engine

Two engines. One proves data is physically real. The other proves it at 1000x speed.

PINNs for precision. Neural Operators for speed. Together: real-time physics verification at scale.

🧠

PINNs Engine

Precision Verifier

Encodes physical laws directly into the neural network. Every data point must satisfy thermodynamics, biomechanics, and signal physics simultaneously.

Ltotal = Ldata + λ × Lphysics

Ldata

Match sensor readings

Lphysics

Obey physical laws

MAE 0.35 Deep Verify 15K Citations
+
⚡

Neural Operator

Speed Engine (FNO)

Fourier Neural Operator learns the entire mapping PINNs discovers — then applies it 1000x faster. Train once on PINNs-verified data, deploy everywhere for real-time inference.

Gθ : sensor → verified (learned operator)

PINNs solves PDEs from scratch every time. The Neural Operator learns the solution operator itself — mapping inputs to outputs in a single forward pass through Fourier space.

1000x Faster Real-time Zhu-Li et al. 2023

Hybrid Proof Flow — How Every Data Point Gets Verified

📱

Sensor Data

25K nodes

→

⚡

FNO Fast Screen

0.05ms • 97% pass

→

✓ PASS

97%

⚠ FLAG

3% ↓

→

🧠

PINNs Deep Verify

50ms • Physics audit

→

✓ VERIFIED

✗ REJECTED

Like TSA PreCheck: most travelers pass through fast screening. Only flagged cases get the full body scan. Result: real-time verification at 25K-node scale without sacrificing accuracy.

Traditional ML

✗

Learns patterns only.
Can be spoofed.

PINNs Only

✓

Physics-accurate.
Too slow for real-time at scale.

Our Hybrid

✓✓

Physics-accurate.
1000x faster. Both.

Physical Laws Encoded in Our Model

❤️

Cardiovascular Dynamics

Windkessel Model

Heart = pump, arteries = capacitor, vessels = resistor. Heart rate, blood pressure, and SpO2 must follow a specific coupled relationship. If one changes, the others must change accordingly.

🔥

Metabolic Thermodynamics

First Law of Thermodynamics

Energy in = mechanical work + heat dissipation. Running must produce elevated heart rate, body temperature, and metabolic rate — simultaneously. Can't fake one without faking all three.

📡

Signal Propagation

Wave Physics

Biological signals have known propagation delays through tissue. Wrist PPG and chest readings can't appear simultaneously — the signal takes time to travel. Timestamping catches replay attacks.

Live Verification Example

INCOMING SENSOR DATA

heart_rate: 185 bpm
spo2: 98%
skin_temp: 36.2°C
accel_mag: 0.02 g
motion: stationary
→

PINNs
CHECK

PINNS VERDICT

REJECTED

✗ HR 185 + stationary = violates metabolic coupling
✗ No temperature rise at high HR = thermodynamic violation
✗ SpO2 98% at HR 185 without exercise = physiologically impossible

Confidence: 99.7% anomalous. Flagged as spoofed or sensor malfunction.

Try Interactive Demo →

0.35

MAE (clinical: 0.61)

PINNs Engine

0.05ms

FNO fast-screen

Neural Operator

50ms

PINNs deep verify

PINNs Engine

2.3%

False positive rate

Hybrid combined

2MB

Edge model size

Both engines combined

Citations: Raissi et al. (2019), J. Computational Physics, 15,000+ citations • Li, Kovachki et al. (2021), Fourier Neural Operator, ICLR • Lu et al. (2021), DeepONet, Nature Machine Intelligence • IEEE 11073 • MIMIC-III (MIT)

The Foundation Model

Verified Fitness FM

A foundation model trained on physics-verified multi-modal sensor data. Not another chatbot. A fitness intelligence engine.

One founder. Zero engineers. $4M revenue. Powered by a 32-agent AI factory.

Why Fitness First

No FDA. No HIPAA. Largest data volume from 25K nodes. Users willingly share workout data. PINNs verification story is instantly intuitive: “He says he ran 10km. Heart rate 72, temp 36.2°C, zero motion. Physics says: impossible.”

Model Capabilities (MECE)

🏃

Identify

Activity Recognition

Run, cycle, swim, lift, HIIT, yoga. Accel + gyro patterns matched against kinematic models.

🔥

Quantify

Intensity & Calories

MET values, calorie burn, HR zones. Constrained by metabolic equations: VO2 = f(HR, weight, type).

💤

Recover

Recovery Prediction

HRV + sleep + resting HR + temp. Parasympathetic recovery curve predicts overtraining risk.

🛡

Verify

Anti-Fraud / Anti-Spoof

Full PINNs constraint check. Data violating physics → REJECTED. Catches GPS spoof, replay attacks, fake workouts.

🧬

Model

Digital Twin

Personal fitness baseline that evolves. Training adaptations follow physiological curves over weeks/months.

Architecture Stack

L4 — APPLICATION

API Gateway + SDK

REST + WebSocket + Mobile SDK
L3 — MODEL

Verified Fitness FM

PatchTST • 50M→200M params
L2 — VERIFICATION

Hybrid Proof Engine

PINNs + FNO dual engine
L1 — DATA

25K Sensor Network

6 device types • multi-modal
L0 — COMPUTE

13-Node GPU Cluster

198 slots • PyTorch + DeepSpeed

Market Segments

Insurance

$2-5/user/mo

Verified workout data → lower premiums. Insurers pay per verification. $300B+ global health insurance wellness market.

Corporate Wellness

$10K-50K/yr

Employee fitness KPI verification. Are they actually exercising? Physics-verified proof for wellness programs.

Fitness Apps (API)

$0.001-0.01/call

Strava, Keep, Nike Run Club integrate our verification API. Anti-cheat for leaderboards. Accurate calorie/distance.

Robotics & Autonomy

Sensor data licensing

Verified sensor data for autonomous systems, robotic perception, and physical AI training.

Why We Win

Oysterworld Apple WBM Google SensorLM WHOOP
Data Source Open network 25K Watch only Fitbit only Strap only
Physics Verified PINNs + FNO No No No
Cross-Device 6 device types 1 2 1
Data Ownership User-owned Apple Google WHOOP
Model Access Open API Internal Internal None

The Data Flywheel

More Nodes

25K → 100K

→

More Data

5K → 50K hr/day

→

Better Model

F1 > 0.92

→

Better Product

API + SDK

→

User Rewards

Network effect

Better product attracts more users → cycle repeats. Self-reinforcing data moat.

Supply-chain-driven hardware matrix. Zero hardware R&D overhead. We do two things: curate the right devices & deploy our AI software stack.

HW

Device sales

60% margin

Data

Sensor data licensing

95% margin

API

Per-call pricing

90%+ margin

License

Enterprise annual

85% margin

Tune

Custom fine-tune

80% margin

“Apple trains on dirty data from one device. Google trains on dirty data from one device. We train on physics-verified data from six device types across 25,000 nodes. Quality × openness > quantity.”

Theory is cheap. Here's what's already built and shipping.

The AI Matrix Platform

Not One Product.
An Integrated AI Matrix.

Hardware, software, data verification, and data intelligence — all working as one platform. Investors back the matrix, not a single device.

AI Smartphone

ClawPhones

25K
Nodes Deployed
Oyster OS Android 14 Distributed Sensor Node

Smart Wearable

Puffy Health

Physics-verified health monitoring in a wearable form factor.

10K Paid Pre-Orders HR + SpO2 + Motion PINNs Verified

AI Glasses

ClawGlasses

First-person AI perception with real-time spatial intelligence.

Beta 2026 3D Spatial Intelligence

Context Sensor

AI Pendant

Coming Soon
24/7 Ambient Intelligence

Hybrid Proof Engine

PINNs + Neural Operator

0.35 MAE • 0.05ms FNO screen • 50ms deep verify

Dual-engine verification: FNO for real-time speed, PINNs for physics accuracy. Combined: 1000x faster than PINNs alone.

Dual Engine FNO + PINNs 1000x Faster

Data Platform

Data Exchange

Real-world asset marketplace for health, mobility, and energy data.

$4T+ Market Health Mobility Energy

AI Distribution

Content Engine

10 Channels 140K+ Community

Economy Layer

Oyster AI Platform

AI Platform Payment + Incentives
60%
Gross Margin
6
Product Lines
13x
Dev Throughput
24/7
AI Agents
32
AI Agents

The AI-Enabled Hardware

Powered by OpenClaw

Every device ships with OpenClaw pre-installed -- turning consumer hardware into AI-powered sensor nodes.

Real Products. Real Revenue. 25K Nodes Deployed.

AI Smartphone

ClawPhones

The backbone of the network. Every ClawPhone is a distributed sensor node running Oyster OS on Android 14.

25K Deployed $4M Revenue Oyster OS

universalphone.xyz →

Smart Health Device

Puffy Health

Smart health device with built-in biometric sensors. Generates physics-verified health data for the health data platform.

10K Paid Pre-Orders Health Sensors Health AI

getpuffy.ai →

ClawGlasses

AI Glasses

ClawGlasses

First-person AI perception with real-time 3D spatial intelligence. Give your claw eyes.

Beta 2026 3D Spatial Intelligence

clawglasses.com →

Health & Wellness

Oral Health Spray

Smart oral care meets AI health monitoring. Sensor-verified formulation integrated with the Oysterworld health data ecosystem.

Shipping Now Oral Care Health Data

Software Layer

OpenClaw

The AI agent runtime that turns any phone into a sensor node.

OpenClaw ships pre-installed on every Oyster device — but it's also a standalone app. One install transforms any Android phone into a physics-verified sensor node for the Oysterworld network. Data collection, PINNs verification, and user rewards built in.

Open Source Android PINNs Built-in User Rewards
View Source on GitHub

5

Device Types

24/7

Data Collection

0.35

MAE Accuracy

60%

Gross Margin

The Market

Three Data Markets Worth $4T+

Health Data

$1.6T

Pharma companies pay $500-5,000/patient for clinical-grade real-world data. Sources: ClawPhones, Puffy, wearables.

Mobility Data

$500B+

Location, movement, and environmental sensing for autonomous systems and smart cities.

Energy Data

$2T+

Real-time consumption, production, and grid status from smart meters and distributed sensors.

Revenue Model

Multiple Streams,
60% Margins

$4M

Current Revenue

Device Sales

ClawPhones, Puffy Health, AI Pendants. 60% hardware margins. 10K pre-orders locked in.

$3M+

Year 2 Target

Data Activation

Per-device activation fees for physics verification and Realm access. 25K active nodes generating predictable recurring revenue at ~$10/device/month.

$1M+

Year 2 Target

Gas & Settlement

Platform fees for device coordination and data settlement. Scales linearly with network activity.

$3M+

Year 2 Target

Realms & Data Licensing

Data licensing to pharma ($500-5K/patient), insurance, and AI training companies. Revenue from Realm access and enterprise contracts.

Multiple revenue streams, 60% margins. Now here's why nobody else occupies this position.

Competitive Landscape

The Physical AI Stack:
Everyone Has a Layer. Except the Eyes.

Actuation

Body

Figure AI $39B Apptronik $5.5B 1X Technologies ~$100B Tesla Bot

Control

Brain

Physical Intelligence (Pi) $5.6B Covariant → Amazon

Modeling

World

World Labs (Fei-Fei Li) ~$5B NVIDIA Cosmos / Omniverse

Perception

Eyes

★ OYSTERWORLD 25K deployed nodes $4M revenue Physics-verified (PINNs) NOBODY ELSE HERE

Oysterworld × World Labs — Upstream / Downstream

World Labs
Oysterworld
Role
Downstream consumer
Upstream data supplier
What They Build
Synthetic 3D worlds from text
Real-world ground truth data
Funding
$1.23B raised
$0 raised (bootstrapped)
Deployed Devices
0
25,000
Revenue
Pre-revenue
$4M
Data Verification
None (synthetic)
PINNs physics-verified
Critical Need
Real-world training data
We provide exactly this

World Labs generates synthetic worlds. But synthetic models need real-world ground truth to validate and calibrate.
We are the upstream data layer that World Models depend on — 25K sensors capturing verified physical reality.

Not competitors. Supply chain partners.

Roadmap

The Path to 100K Nodes

2026 H1

Consolidate

Data Activation pilot with 3-5 pharma partners. Health data marketplace beta. ClawGlasses prototype.

Target: 35K nodes

2026 H2

Expand

AI Pendant launch. ClawGlasses beta. Oyster SDK for third-party hardware integration.

Target: 50K nodes

2027 H1

Scale

Third-party hardware onboarding. Mobility data marketplace. Enterprise data contracts.

Target: 75K nodes

2027 H2

Platform

ClawGlasses GA. Energy data marketplace. Open API for AI agent frameworks.

Target: 100K nodes

Team

Builders Who Ship.
$4M Revenue Proves It.

Howard J. Li

Co-Founder & CEO

Serial builder. $4M hardware revenue with zero funding.

Co-founded MPCVault (digital asset custody, $5B AUM, 1,000+ clients).

LP at Berkeley SkyDeck. Founded Erupture Angel Network at UC Berkeley.

Wharton MBA  |  UC Berkeley Haas.

Full Story →

Eric Zhu

Co-Founder — Supply Chain & Global Business

Shareholder of Shenzhen Energy Southern Control.

Family enterprise ownership in power generation, solar, wind & energy storage.

Extensive robotics experience. China manufacturing & global distribution. UC Berkeley, Environmental Economics & Policy.

Stephen Vu

Advisor (GTM)

Veteran in Renewable Energy. 100M+ Projects building with Verizon, SCE & Orange County Transition. Designs power stations for Southern California Edison & Verizon. US energy & telecom industry connections.

Bruno Moreira

Head of BD  AI-AUGMENTED

Runs 24/7 outreach, partnership qualification, and deal pipeline with AI-augmented automation.

Driving enterprise partnerships, pharma pilots, and go-to-market strategy across the Oyster ecosystem.

AI Agent Factory

Secret Weapon

32

AI Agents

AI development system producing code at 13x throughput. 13 compute nodes. 24/7 operation.

"We don't hire 50 engineers. We built 32 AI agents that work 24/7."

Strong team, clear roadmap. Now let's address what keeps skeptical investors up at night.

Hard Questions

Every Investor Will Ask These.
Here Are Our Answers.

We don't hide from risk. We name it, measure it, and show you how we solve it.

"What is Physical Intelligence? Isn't this just IoT with a new name?"

Fair question. The IoT industry has been over-hyped for a decade. "Physical Intelligence" can sound like a rebrand of the same old smart-device playbook.

IoT collects data. Physical Intelligence understands it. The difference is the verification layer: our PINNs engine encodes the laws of physics — thermodynamics, biomechanics, fluid dynamics — directly into the data pipeline. IoT gives you raw sensor readings. We give AI agents verified, physics-consistent perception of the physical world. GPT was trained on the internet. The next frontier of AI needs physical-world training data — and that data has to be trustworthy.

Key insight: LLMs know what a heartbeat is from text. Physical Intelligence knows if this specific heartbeat is physiologically real. That's the gap we fill.

"Apple has a billion devices. Google has sensors everywhere. Why can't they do this?"

Big Tech has massive distribution, better hardware, and unlimited R&D budgets. Competing with them on devices seems suicidal.

They can build devices — they can't build a decentralized, user-owned data network. Apple's health data stays in Apple's walled garden. Google's data feeds Google's ad engine. Neither company can credibly promise users sovereignty over their own data, because their business model depends on controlling it. We're not competing on hardware specs. We're competing on data ownership. Our platform lets users own and monetize their data. Big Tech structurally cannot offer this.

Proof: Apple Health data: locked in iCloud. Oysterworld data: user-owned, PINNs-verified, tradeable on open marketplace.

"Hardware networks fail at scale. Why won't Oysterworld?"

The risk is real. Most hardware startups fail because they front-run network effects with hype before proving product-market fit.

We built $4M in hardware revenue before even launching a platform. Revenue first, platform second. We're not raising on a whitepaper. We're raising on a P&L.

Proof: $4M revenue. 60% hardware margins. Revenue-funded growth.

"One founder? How do you ship like a team of 50?"

Solo founders are a red flag for VCs. Key-person risk. Limited bandwidth. Most solo-founded companies fail to scale past seed.

We built a 32-agent AI development factory that runs 24/7 across 13 compute nodes. It produces code at 13x the throughput of a traditional engineering team. This isn't vaporware — it's how we shipped 5 hardware products, a data platform, and a 140K-person community with zero employees.

Proof: 5 products live. $4M revenue. 25K nodes. All built with AI agents, not headcount.

"How do you ensure data from consumer devices is actually reliable?"

Consumer sensors are noisy, inconsistent, and easily spoofed. Pharma companies need clinical-grade accuracy. The gap between a Fitbit and a medical device is enormous.

PINNs (Physics-Informed Neural Networks) encode the laws of physics directly into our verification model. Heart rate × SpO2 × motion must be biologically coherent — if it violates pharmacokinetic curves, it's automatically rejected. Traditional ML learns patterns from data alone. PINNs constrain predictions to obey physical laws — Newton's mechanics, conservation of energy, biological rate equations. This makes verification fundamentally harder to fool.

Proof: PINNs MAE 0.35 vs clinical threshold 0.61. Methodology: Raissi et al. (2019), Journal of Computational Physics.

"Your revenue is 100% hardware sales. Where's the recurring software revenue?"

Hardware-only revenue means no SaaS multiples. Investors want recurring revenue, not one-time sales. Without software monetization, this is just a consumer electronics company.

We built hardware first because you can't sell data from a network that doesn't exist. Others tried hype-first and got speculation. We went revenue-first and got 25,000 deployed nodes. That's the hard part done — the infrastructure is live, collecting data today. Data activation (recurring revenue per node) launches H1 2026. But here's the honest answer: even without data revenue, $4M in hardware at 60% margin with $0 funding is a real business. Data monetization is upside, not a requirement for survival.

Proof: 25K nodes deployed and collecting data. Hardware sustains the business. Data revenue is additive, not existential.

"You sold 25K devices but your community seems disengaged. What happened?"

This is the elephant in the room. Revenue grew fast, but community incentives and communication didn't keep pace. Users bought hardware but weren't given clear reasons to stay engaged. Trust eroded. That's on us.

We're not hiding from this. We shipped product before community — the right order for revenue, the wrong order for retention. Here's what we're doing: OpenClaw pre-installed on every device — gamified, incentivized, giving users a reason to turn on their nodes daily. Platform rewards give early adopters real upside. New community program: transparent roadmap updates, direct founder access, and retroactive rewards for Day 1 supporters. We broke trust by going quiet. We're rebuilding it by shipping loud.

Proof: OpenClaw beta live. Community rebuild program launching Q1 2026. 140K+ across X, Discord, Telegram — the base is still there. They need a reason to care again. We're giving them one.

"$30M valuation for a hardware company with $4M revenue?"

7.7x revenue multiple is aggressive for hardware. Most hardware companies trade at 1-3x. Even SaaS companies at this stage are 5-10x.

The hardware is the distribution channel, not the product. The product is a 25,000-node data network that's already deployed and collecting. Comparable AI hardware networks traded at 50-100x revenue. At $30M, you're buying into the network before data monetization kicks in. World Labs raised $1.23B with zero revenue and zero deployed devices. We have $4M revenue, 25K live nodes, and 60% margins.

Proof: $4M revenue (7.7x multiple). Bootstrapped, profitable. World Labs: $5B valuation, 0 devices, 0 revenue.

Community

Follow the Build.
In Real Time.

We ship in public. Every product launch, partnership, and technical milestone — across 10 active channels.

X / Twitter

4 Accounts

@oysterecosystem
@ClawGlasses
@puffy_ai
@UBSPHONE

LinkedIn

6 Pages

ClawGlasses
ClawPhones
Puffy AI
Oyster Republic

Bluesky

4 Accounts

oysterecosystem
clawglasses
puffyai
clawphones

Discord

Community

Announcements
Dev Updates
Hardware Support
Open Forum

Telegram

Updates

News & Alerts
Product Updates
Product Drops
AMA Sessions

14

Active Channels

140K+

Total Reach

37

Posts / Day

AI

Powered Engine

Now Raising

$1M Builder Round

$30M post-money valuation. SAFE.
80% Go-to-Market. 20% Engineering.

Schedule a Call with Howard
Howard J. Li  |  Co-Founder & CEO  |  [email protected]