• The Current ⚡️
  • Posts
  • đź“– Long Reads: Technical Underpinnings of the USA vs. China AI “Contest”

📖 Long Reads: Technical Underpinnings of the USA vs. China AI “Contest”

Also, a sober + mathematical look at the $500b Stargate Project

Author

Jack Lajoie
January 26, 2025 • read time ~ 4 minutes

Deeper dives 🤿 on pertinent issues that are well worth your time today 📚

DeepSeek Is Chinese But Its AI Models Are From Another Planet

OpenAI and the US are in deep trouble

www.thealgorithmicbridge.com/p/deepseek-is-chinese-but-its-ai-models

This one is more for the engineers out there (deep technical analysis incoming) but it’s useful for anyone to have a look just to get a little more grounded on the USA vs. China narratives/realities (generally and specifically related to AI).

As my #1 prediction for AI in 2025 I wrote this: “The geopolitical risk discourse (democracy vs authoritarianism) will overshadow the existential risk discourse (humans vs AI).” DeepSeek is the reason why.

Simply put, tech accelerationists are going to prefer and generally upvote the narrative that the most important thing to think about with AI is “who wins”.

If China (the authoritarian “other” to the democratic American main character) wins then the world will surely be controlled by the dystopian singularity-type thing we’ve all sorta had in the back of our minds lately. If the USA wins, it’ll mean economic prosperity and happiness for all.

On the flip side, tech deccelerationists (that’s not a word but I made it up since technophobes is too reductive) present a more global dichotomy — AI development as a whole, whether Chinese or American, is seen as leading to social chaos and inevitably authoritarian rule from whoever controls it, regardless of national identity (or humanity, for that matter).

Beneath all of that is a lot of really interesting math and mind blowingly fast improvements on machine learning.

The article is paywalled towards the end, but it provides plenty for free to get the jist. It’s a great Substack to follow if this is your cup of tea.

Doing the math on the Stargate Project; Yann LeCun predicts new AI paradigm in five years; revenue is booming for AI startups; game developers push back against AI projects

Immigrant AI workers worry about the new US administration; Huawei aims to replace NVIDIA in China; the second wave of AI coding is here;

www.computerspeak.co/p/the-math-on-stargate-project

One of the biggest announcements this week during a flurry of Trump EOs aimed at establishing his tech agenda was Project Stargate, a $500 billion infrastructure investment plan for the next four years backed by OpenAI, SoftBank, MGX, and Oracle.

Elon Musk threw some doubt on the plan by claiming the backers don’t actually have the money for it, an allegation that, while certainly laced with personal reasons, isn’t totally crazy to make.

Adjusted for inflation, Stargate’s $500 billion investment would be more than twice as much than what the US spent to put a man on the Moon and nearly as much as the cost of building the Interstate Higher System (that spend was spread over 36 years, however, vs. just four years for Stargate). source

While people got lost in debating the reasons why Elon would make such a claim, some people did math. This article is for anyone wanting a sober look at this massive initiative that, if fully undertaken, would clearly articulate the development of AI as the most significant (if not purely “expensive”) commitment our country has undertaken.