• AI Academy
  • Posts
  • 🤖 As Trump Invests Big China Innovates Smart

🤖 As Trump Invests Big China Innovates Smart

Plus: The Race for AI Supremacy Takes a New Turn

January 23, 2025

Hello AI Enthusiast,

The AI landscape is shifting dramatically this week. In Washington, a $500 billion mega-project signals America's ambitions for AI dominance, while from China, an impressive open-source release shows they won't be left behind. It's a week that perfectly captures how the global AI race is heating up.

The Big Picture 🔊

White House Welcomes Stargate

President Trump welcomed OpenAI, SoftBank, and Oracle executives to the White House to announce Stargate, dubbed "the largest AI infrastructure project in history." The initiative will start with a $100 billion investment, potentially growing to $500 billion, and aims to create 100,000 US jobs. The announcement came just a day after Trump revoked Biden's 2023 AI safety regulations that required companies to share safety test results before public releases. OpenAI CEO Sam Altman, who attended Trump's inauguration, stated "We wouldn't be able to do this without you, Mr. President."

Gianluca Mauro
Gianluca MauroFounder and AI Rockstar

It's fascinating to see how "national security" is being used to justify two completely different approaches. Biden used it to justify regulations, while Trump cites it to remove them and accelerate AI development. The urgency seems driven by competition with China as OpenAI warns that without U.S. investment, $175B in global funding could shift to Chinese projects. The irony? Stargate aims to create AI systems that outperform humans in key tasks but is being sold as a job creator.

As we noted in our previous newsletters, this shift in AI regulation was predictable with the change in administration, showing how deeply AI policy is now tied to political cycles.

What's your take on Trump's new direction for US AI?

Login or Subscribe to participate in polls.

DeepSeek Releases Open-Source Rival to o1

DeepSeek, the Chinese AI lab, has released R1, an open-source reasoning model claiming to match OpenAI's o1 performance at a fraction of the cost. The model is freely available for commercial use and comes in various sizes. While R1 must comply with Chinese content restrictions, its release amid tightening US export controls signals China's growing capability to develop competitive AI models independently.

James Varnham
James VarnhamCEO and Rainmaker

While America plans AI dominance, China isn't standing idle. DeepSeek recently showed this in two different ways: their V3 model caught attention by creating a more efficient training system, cutting costs by 90%. But their new R1 model is truly innovative in how AI thinks through problems, matching OpenAI's o1 performance while being free to use and running on cheaper computers.

By running their own data centers, DeepSeek turned America's chip restrictions into motivation to find smarter solutions. This shows China isn't just copying anymore, they're pioneering new approaches to AI.

Nations are racing for AI dominance, but this technological race only matters if companies can effectively implement these advances. Join our free webinar "Why Traditional AI Training Fails (And What Works)" on January 27th at 6 PM CET, where James, our CEO, will show you why traditional AI training approaches aren't enough in today's fast-moving landscape. He'll break down why certifications often lead nowhere and share a practical framework that actually works.

Bits and Bobs 🗞️

  • Google has simplified access to Workplace AI tools by integrating them into the standard Workspace Business plan at $14 per user per month, removing the separate $20 Gemini for Workplace fee.

  • Google researchers introduced Titans, a neural-network architecture that integrates attention mechanisms with a “neural memory” layer, enabling language models to handle millions of tokens efficiently while reducing memory and compute costs.

  • Luma Labs has launched Ray2, a text-to-video platform that quickly transforms descriptions into short, realistic video clips.

  • OpenAI has revamped the ChatGPT custom instructions feature, allowing users to easily personalize how the AI responds.

  • Adobe has introduced new generative AI tools, including "Firefly Bulk Create," which can automate image editing tasks, like resizing or replacing backgrounds for up to 10,000 images at once.

  • OpenAI has developed a new AI model, GPT-4b micro, which has significantly improved the efficiency of protein factors in converting regular cells into stem cells, beating human attempts by making the proteins over 50 times more effective.

  • Windows Insiders with Copilot+ PCs can now enjoy enhanced search capabilities in Windows 11, thanks to semantic indexing that allows users to find documents, photos, and settings using natural language descriptions.

Tribal News🫂

Two days ago, our Members had the chance to ask Gianluca, our founder, anything about AI during our live Ask Me Anything session. From market trends to technical doubts, no question was off limits, though we had to smile when half the session naturally drifted into a deep dive about AI agents.

One of the slides shared during the session

From Our Channels 🤳

With TikTok's fate being one of the most talked-about topics of Trump's first days in office, Gianluca explains why the ban isn't quite what it seems.

@gianluca.mauro

Why the tiktok ban is not about data. BTW, all examples and scenarios I make are just examples, I’m not suggesting any of this actually ha... See more

LOLgorithms 😂

With both Musk and Altman backing Trump, they seem to be getting along just fine.

That's a wrap on our newsletter! Before you go, here’s a quick recap of our offerings:

Catch you next week! 👋