The Core Thesis
"Whoever controls the distribution of tokens (data, relationships, information) in a vertical will own that vertical. Build the tokenizer, become the operating system."
Distribution Control
in the Age of AI
A framework for building vertical token systems that capture lasting value
What Are Tokens?
Tokens are discrete units of encoded information that make the world accessible and legible to machines.
Access Tokens
Credentials and permissions that grant entry to systems and services
Memory Tokens
Personal context and history that AI uses to understand and serve individuals
Expert Tokens
Domain-specific knowledge encoded into systems for specialized tasks
Identity Tokens
Personal information establishing who someone is across digital systems
Knowledge Tokens
Company and organizational data powering business operations
Asset Tokens
Digital representations of ownership rights and value
Why Now?
1000x
Reduction in inference costs over 3 years
500T+
Tokens processed monthly by major platforms
80%+
New vehicles with ADAS features
$0.02
Blockchain transaction fees collapsed
The PowerOutage Example
A case study in distribution control
927+
Utilities tracked
213M+
Customers covered
10 min
Real-time data updates
The Insight
"VCs with 10 PhDs who built amazing hardware/software and raised $100M... go to PowerOutage for distribution."
Everyone is struggling with problem awareness. Control distribution, and they come to you.
The Mental Model
Common Sense + Determinism Architecture
LLMs are "common sense engines" while code is deterministic. Mixing these creates powerful automation.
Common Sense
Local LLMs (Qwen 80B)
Pattern recognition, data enrichment, ambiguity handling
Determinism
Traditional Code
Reliable execution, data pipelines, business logic
Infrastructure
Mac Studios 128-512GB
Self-hosted compute at $4K-10K/unit, years of runtime
The Cap-Ex vs Op-Ex Arbitrage
The most powerful tactical insight: understand how your customers make money.
Investor-owned utilities make profit by building their rate base (capital investments). Operating expenses are pass-through costs with no profit margin.
By repackaging SaaS as on-premise infrastructure, you convert a cost center (Op-Ex) into a profit center (Cap-Ex)—and can bill 2-3x.
"You only learn insights like this by controlling distribution in an industry."
3-Year Distribution Capture Playbook
YEAR 1
Build & Give
Build platform, give away distribution free. Invest in user/entity acquisition.
YEAR 2
Sustain
Small SaaS or get major sponsor. Break even or slight profit.
YEAR 3
Monetize
World has changed, you own the relationships. Monetize distribution, expand vertically.
Example: Get Sig Sauer to sponsor a firearms site, offer free listings to instructors—while capturing all relationships and data.
Selection Criteria
Target These
- Already unaggregated industries
- Verticals big business won't touch
- Real-time data-sensitive industries
- Industries with cap-ex budget structures
- Obscure data aggregators (ship movements, etc.)
Avoid These
- -Heavily regulated (finance, healthcare) without resources
- -Anything a big tech company cares about
- -Static directories (live data has defensibility)
- -Consumer-focused without institutional buyers
The Window is Closing
"SEO-style cheap distribution is going to be dead in 3-5 years. Get distribution while you can."
LLMs are changing how people form consideration sets. The way discovery works is fundamentally shifting. Capture relationships now.
Cap-Ex Opportunities
Industries where customers make money from infrastructure investment, enabling 2-3x pricing and stickier relationships.
Automotive Services
Energy
Infrastructure
Healthcare
Commercial Real Estate
Manufacturing
Agriculture
Utilities
Auto Glass & ADAS Calibration
Market: $20.2B aftermarket glass, ADAS calibration growing to $5.8B by 2030. Calibration costs $300-800 per vehicle.
Problem: Safelite dominates but only 26% of shops have calibration equipment ($15K-50K/unit). Insurers need certified networks.
Cap-Ex Angle: Equipment financing bundled with network membership. Insurance companies fund rollout to ensure calibration compliance.
Distribution Path
- Partner with insurance carriers (70%+ of claims)
- Become preferred network for claims routing
- Control insurer-to-shop relationships
Sponsors: State Farm, Progressive, Autel, Hunter
Generator & Backup Power
The Opportunity
Generac dominates hardware but installers are completely fragmented. No standardized quoting, wildly variable quality.
Token Play: Aggregate installer network, track every installation. Know where every backup power system is = know grid resilience gaps.
Cap-Ex Angle
Municipalities and utilities can fund backup power programs as infrastructure investment. HOAs and commercial properties capitalize generator systems.
Synergy: Direct tie to PowerOutage data—outages drive generator demand.
EV Charging Infrastructure
ChargePoint, EVgo exist but 30%+ of chargers broken at any time. No aggregated reliability data.
Token Play: Aggregate real-time status, reliability scores, wait times. Become the "TripAdvisor for EV charging."
Cap-Ex Angle: Pure cap-ex for utilities, retail, municipalities. Reliability data influences where billions in federal infrastructure money flows.
30%+
Chargers broken at any time
$7.5B
Federal EV infrastructure program
More Cap-Ex Opportunities
Solar & Battery
30% ITC flows to capital investors. Fragmented installers need quality tracking.
Elevator Maintenance
$100K-500K modernization projects. OEMs have 80% margins on service.
Medical Equipment
Imaging equipment $500K-$3M. Third-party service saves 30-50%.
Commercial HVAC
Building controls are capitalized assets. Utility efficiency programs.
Water Infrastructure
$625B infrastructure bill. Lead pipe replacement mandated.
Data Center Edge
Edge facilities $1-10M each. Reliability affects financing terms.
The Common Pattern
1.
OEMs protect service revenue, independents are fragmented
2.
Capital intensity of equipment creates financing opportunities
3.
Financing bundled with network membership = distribution moat
4.
Real-time data creates defensibility static directories lack
On Self-Hosted Models
"Self-hosted non-frontier models will be the workhorses and steam engines of this era."
The Economics
A Qwen 80B at 6-bit quant on a $4K Mac Studio is "like having a smart high schooler do data entry 24/7" for years at minimal marginal cost.
The Strategy
Not about cost savings (5x more than OpenAI). It's about ownership, learning from first principles, and protection from vendor lock-in.
Key Principles
"Figure out how to control the relationships with the biggest wallets in the industry."
"Aggregate an entire industry. Build a platform. Get everyone on it. Control distribution."
"Golf courses, wedding venues, birthday places... all of that will go on and no one sees it."
The Playbook
- Pick a vertical where distribution is capturable
- Aggregate the fragmented—the messier, the better
- Capture real-time data (not static directories)
- Control relationships with institutional buyers
- Use local LLMs for data enrichment
- Think Cap-Ex, not Op-Ex
- Structure around how customers make money
- Move now—the window is closing