AI EMPIRE

How to play

• Roll & move. Land on an unowned asset to buy it. Land on a rival's asset and pay rent. Pass FUNDING ROUND to collect $200.
• Own a full color layer (e.g. all three LLM squares) to double base rents and unlock building: up to four 🤖 MODELS, then one 🏛️ AGI. Rents explode with buildings.
• GPU Datacenters (Jamaica, Trinidad, Barbados, Bahamas) pay more the more of them you own - compute is the railroad of this era. Utilities (Solar Grid, Subsea Cable) charge a multiple of the dice.
• BREAKTHROUGH ✨ and TECH NEWS 📰 cards bring the chaos of the real AI industry: viral models, GPU shortages, privacy fines, NaN losses.
• DEBUGGING 🐞 is jail. Escape by rolling doubles (reproducing the bug), paying $50, or playing a Get-Out-Free card (you wrote unit tests).
• Cash crunch? Sell buildings or mortgage assets - mortgaged assets earn nothing, just like in real life. Run out of everything and you're bankrupt. Last empire standing wins.
• 🖥️ Corner the chip market. Own 3 GPU datacenters and every rival builds models at +25% cost. Own all 4 and you've cornered it: rivals build at +50%, pay you a surcharge on every build, and your datacenter rent jumps to $300.
• 🧠 Hire AI engineers (once per turn, up to 5, price rises each hire). Each engineer boosts ALL your income by 8% - better talent, better products, better markets.
• 🔬 Fund R&D ($120, once per turn). Every innovation point pays royalties each funding round and raises your breakthrough odds. Breakthroughs - cancer detection, hurricane forecasting, antibiotic discovery - save lives ❤️ and earn grants.
• 🏆 The Impact Tally ranks everyone live: net worth + revenue/4 + lives/10 + innovation×25 + talent×120 + (trust−50)×6. Win by bankrupting rivals, or press END & TALLY (from round 8) and let total impact decide. Your strategy - landlord, chip baron, or scientist - shapes your score.
• 🤖 The bots play for real: Compute Cassie hoards datacenters to corner chips, Data Don rushes cheap layers and builds fast, Agent Adrian invests in research and talent and hunts premium assets. All of them will buy a square just to block your monopoly.
• 🎲 3D view - toggle the voxel-style tilted board and tumbling dice with the 3D button.
• 🛡️ Public Trust - risk zones, scandals and war contracts drain it; safety practices, climate AI and discoveries grow it. Drop below 40 and public backlash drains your cash every turn. Trust counts in the final tally.
• ⚠️ AI Risk Zones - the Rogue Agent and Cyberattack squares hit you with real-world AI failures: breaches, bias lawsuits, deepfake scandals. The AI INCIDENT corner (an agent gone awry, ransomware, NaN loss…) sends you to Debugging & Containment.
• 🧪 Tech Tree - innovation points unlock permanent upgrades: ML Mastery, Data Flywheel, Foundation Models, Green Compute, Physical AI, Climate AI, and Quantum AI at the summit.
• 🌐 World Events - pandemics, global AI outrage, innovation booms, GPU gluts, regulation waves and viral adoption strike at random. Research and trust make you resilient.
• 👥 Pass & Play - up to 4 human founders share one device; the panel always manages whoever's turn it is.
• 📜 Scenarios - beyond Classic: survive the AI Winter (half funding, an event every round), escape a Trust Crisis (climb from 25 to 80 trust), or win the Race to Quantum (first to the top of the tech tree takes all).
• ⚔️ Hard choices - war-AI contracts pay big revenue but cost public trust. Climate and poverty programs pay less but build it. Your strategy decides what your empire is FOR.

What it teaches

• The AI value chain: the board's eight color groups walk the real stack in order - Data → Data Infrastructure → Classic ML → Deep Learning → Generative AI → LLMs → Agents → AGI. Each side of the board is a layer; you can't credibly build the top without the bottom.
• Compute economics: GPUs as railroads, energy and bandwidth as utilities, the cloud bill as the tax collector. Compute scarcity is rent-collection in the real world too.
• Vertical integration: monopolizing one layer (a full color set) is what makes building MODELS and AGI possible - exactly how real AI giants compound their advantage.
• Startup survival: burn rate, runway, mortgaging the future for cash today, and why most AI startups die of finances rather than technology.
• Compute concentration: when one player corners GPUs, everyone else's rollout gets slower and pricier - exactly how chip scarcity and export controls tax the real industry.
• Talent economics: engineers are the scarcest resource in AI; each hire compounds revenue across everything you own.
• Science as strategy: research looks expensive turn-to-turn but pays royalties, compounds discovery odds, and saves thousands of lives - AI's highest form of impact. The final tally proves an empire is measured in more than money.
• Industry literacy: every square and card carries a real fact - from Common Crawl and ImageNet to RAG, hallucinations, scaling laws, and GDPR fines. Click any square to read its story.

Credits

AI EMPIRE is an educational game by Adrian Dunkley - The AI Boss (adriandunkley.net). The GPU datacenters honor Jamaica, Trinidad & Tobago, Barbados and The Bahamas - because the Caribbean is not catching up to AI, it is building the version the world will need next. Also play AIMon, the Zelda × Pokémon game that teaches machine learning.