The Greatest Strategy Game in the History of the World
Seeing the AI race through the lens of chess, statecraft, StarCraft, and Azad
Every time I see a big AI headline, I think in terms of chess moves. Amazon is going to invest $50B in OpenAI, rook to a5. Nvidia is going to invest $100B in new data centers, queen to e8. Nvidia might not invest $100B in new data centers, queen to e3.
It is truly amazing to watch one of the greatest strategic battles play out in real time, with access to the Internet and real-time updates. I do not believe there has ever been such a high stakes match played in history. The only historic precedent for what we are witnessing are the great military battles of history, and certainly regular people didn’t get to observe those from the sidelines, discussing and analyzing each and every move.
One of the reasons the Game of AI is so captivating is that the players are so powerful. It cannot be repeated too many times how unprecedented this is. The Big 7 tech companies represent nearly a third of the stock market. The cloud and advertising monopolies are two of the greatest monopolies in the history of business, with perhaps the only exception being the Dutch East India Company. These cash cows have proven more than sufficient to support many new adventures in the age of AI.
Today’s Magnificent 7 are better modeled as nation-states than as corporate actors. The CEO is the king, the board of directors is akin to the parliament, the shareholders to the population, and the employees to the cabinet and the administrative bureaucracy of the state. With the king and parliament fully committed to war, we are seeing a strategic game being played with dramatic moves taken each day. This is not a cautionary, defensive game being played — it’s all out offense.
Chess and statecraft are two readily accessible metaphors, but I think the closer analogy is actually a resource allocation game like StarCraft. Players are competing over a finite set of resources: AI researchers, GPUs, power, and capital. We have reached a point where this is far from abstract — from cheap power in West Texas to hydropower in Norway, the game has physical dimensions to it. Players are building AI factories, committing resources to units that will go-live at different phases of the game.
To win the game, each player has taken a different approach: Google is focused on vertical integration, OpenAI and Anthropic on model dominance, Meta on talent. But all are fighting for the same physical resources, which has created a heated competition. Meanwhile, the differing strategies have resulted in each player uncovering a dominant mode of operation: OpenAI is winning in chat, Anthropic in code, xAI in image, and Google across a wide variety of good-enough offerings.
Some readers will push back on the game analogy, saying that the stakes are too high for this to be a game. Rather, it is an utterly serious quest to define the future of human civilization. To them, I would point to Azad, Iain Bank’s fictional game in The Player of Games. Azad is a game played for the highest stakes of all: the winner becomes Emperor. The book is a cutting critique of human civilization, and it captures some of the very real drama that defines all great battles for the future of civilization, the AI race among them.
Most strategy games involve opening strategies, mid-game strategies, and end-game strategies. Different player types and strategies perform better at various phases of the game. While it is possible to see who is ahead, it is difficult to forecast the end-game, and this is especially the case when those playing are grandmasters.
It is valid that many of us are constantly analyzing the game board, and either betting on where it will go, or actively making moves to change it. Today, I wanted to simply take a step back and appreciate the beauty of the game, the grandmasters currently engaged in it, and the unparalleled scale of this competition.
All the world's a stage...or to each their own analogy.
Connecting a few observations, it seems reasonable to be very nervous about the foundational model company valuations. My naive logic is that:
1) it seems that the public markets are nervous about the current capex spending;
2) it seems that OAI/Anthropic/xAI must access public markets to 'keep up' in spend;
3) it seems likely (as you noted in your 2026 thoughts) that some of the datacenter buildout will have delayed delivery, extending the lag between capex and revenue.
If these concerns truly kick-in ahead of these IPOs, it seems that there could be some serious holes in various balance sheets. Is this line of thinking overly negative, or missing critical data or thought? I would love your perspective.
It’s funny you compare this to an RTS, because that’s exactly how it feels. If I translate it into DnD terms, I rolled high INT and WIS, but DEX was always my dump stat. I’ve always understood how systems should work and how to architect solutions, but actually executing in code was the bottleneck. AI didn’t change the character, it just removed the execution penalty. In the past few months alone, I’ve built three fully functioning applications to solve problems I ran into directly. Not prototypes, actual working systems. The shift has been from searching for tools to simply creating them. Most recently, I wanted a very targeted SEO and advertising strategy that wouldn’t burn cash, so instead of trusting existing ad platforms to optimize for me, I spent about 12 hours building my own SaaS platform that identifies undervalued ad space and maps it directly to my target demographics. What’s fascinating about the Mag 7 “arms race” isn’t just the scale of their competition, it’s that they’re distributing capability as a side effect. They’re building the infrastructure, but in doing so, they’re enabling a class of builders who previously couldn’t execute at this level. I’m no longer watching the game. I’m building pieces on the board.