Game Theory in Semiconductors, AI, and Energy
Get ready for NVIDIA volatility. LONG: PLTR, TSM , CCJ. LLMs are quickly becoming a commodity. Where’s the edge? Ontologies!
1. Introduction: The Strategic Game in Semiconductors, AI, and Energy
The semiconductor industry isn’t just evolving—it’s shifting the rules of the game. Three forces are driving this transformation:
Disruptions in the supply and demand of chips.
The evolution of AI: from Large Language Models (LLMs) to human-machine interaction.
The relationship between chips and energy: how efficiency gains paradoxically drive higher demand. This isn’t a temporary trend—it’s a tectonic shift in the global economy.
2. Supply Disruption: The End of NVIDIA’s Monopoly
NVIDIA’s dominance once felt unshakable. But monopolies in technology are always temporary.
Big Tech is developing proprietary chips (Google TPUs, Apple, Meta, Microsoft).
Startups like Tenstorrent, Groq, and Rivos are challenging traditional architectures.
The shift toward chiplets and open architectures is reducing entry barriers. Game Theory in Action:
Before: NVIDIA had near-total control. High margins. High barriers.
Now: More competition, lower margins, and a rapidly shifting landscape.
3. Demand Disruption: AI is Getting Cheaper, and That’s a Problem
DeepSeek changed the game.
AI models can now be trained more affordably.
Not all AI applications need ultra-premium NVIDIA chips.
AI is becoming a commodity, driven by OpenAI, Perplexity, Llama, and DeepSeek.
Game Theory in Action:
Before: NVIDIA was indispensable.
Now: Hardware differentiation is fading. The game is shifting to application strategy.
4. AI’s Evolution: From Hype to Human-Machine Reality
The initial AI hype was about replacing humans.
Stories like AI hiring humans to solve CAPTCHAs fueled the fear.
The initial AI hype was about replacing humans.
Yuval Harari Stories like AI hiring humans to solve CAPTCHAs fueled the fear.
However, DeepSeek and others have demonstrated the opposite: AI’s true value lies in enhancing human intelligence, not replacing it. This aligns with the insights of Garry Kasparov in his book Deep Thinking: Where Machine Intelligence Ends and Human Creativity Begins, which explores the symbiotic relationship between human ingenuity and machine efficiency.
The real advantage? Structured knowledge ontologies, not just LLMs.
Conclusion: AI isn’t making humans obsolete—it’s making knowledge structuring more powerful than ever.The real advantage? Structured knowledge ontologies, not just LLMs.
5. From LLMs to Ontologies: The Next AI Frontier
LLMs are quickly becoming a commodity.
Where’s the edge?
Not just in generating text, but in integrating domain-specific intelligence.
Structured Knowledge is the New Moat:
Palantir is leading this space.
China is racing to develop ontologies in defense, healthcare, and finance.
The true AI advantage = context and specialized intelligence.
6. Game Theory in the Semiconductor Trade War
The US tried to slow China’s semiconductor progress.
Classic Game Theory: Barriers breed innovation.
Example: Build a wall, and your opponent builds a taller ladder.
Evidence:
Huawei’s Kirin 9000S, produced domestically despite sanctions.
SMIC rapidly closing the technology gap.
Conclusion: The embargo didn’t stop China—it accelerated its independence. Welcome to the new Cold War of technology.
7. Energy: The Jevons Paradox and the Coming Energy Boom
Investors misread energy trends.
Jevons Paradox Applied:
Greater chip efficiency should lower electricity consumption.
In reality, efficiency often increases total demand.
More efficient chips → more AI → more data centers → more power consumption.
Energy Investment Playbook:
Nuclear (Uranium): The only long-term solution to exponential demand.
Natural Gas: The crucial bridge to a stable energy transition.
Game Theory in Action:
Before: The assumption was that AI would lower energy demand.
Now: AI is triggering an energy super-cycle.
Key Takeaway: The most valuable commodity in the AI revolution isn’t data—it’s electricity.
8. True Investment Opportunities: From NVIDIA to Taiwan Semiconductor, Palantir, and Energy
TSMC: The next undisputed leader.
Advanced node production makes it indispensable.
If NVIDIA loses its premium, TSMC could surpass it in market cap.
Broadcom (AVGO): A long-term play on 5G and infrastructure.
Palantir (PLTR):
AI’s real moat is structured knowledge, not raw computing power.
Palantir’s data integration capabilities make it a long-term winner.
Energy: The overlooked giant.
AI is power-hungry. Data centers need a lot of electricity.
Uranium and natural gas are now strategic assets.
9. Conclusion: Investing in a World of Multiple Nash Equilibriums
This isn’t a traditional market—it’s a multi-layered strategic game.
Investors relying on old frameworks (diversification) will lose.
The future isn’t just about monopolies like NVIDIA—it’s about strategic players like TSMC, Broadcom, Palantir, and Cameco (uranium).
Forget diversification—this is a world of competing Nash Equilibriums.
Investing in semiconductors and energy now means understanding game theory. Some will win. Some will lose. The key is positioning on the right side of the equation.
This research isn’t just a snapshot of where the market is today—it’s a roadmap for navigating a world where technology, energy, and game theory are the defining forces of investment success.
Guillermo Valencia A
Cofounder of Macrowise
January 29th , 2025
Disclaimer: This document is for informational purposes only and should not be considered financial advice. Always conduct your own research before making any investment decisions.