AI with Kyle Daily Update 021
Today in AI: West Concedes Open Source Race to China, Amazon Pays NYT $20M for AI Training
Watch this video on our dedicated watch page
Better viewing experience with related videos and full-screen player
The skinny on what's happening in AI - straight from the previous live session:
Highlights
🇨🇳 China Owns Open Source AI (And The West Just Gave Up)
The open source AI leaderboard tells a stark story: Chinese models dominate the top spots whilst Meta's Llama sits at #15. Qwen 3 (Alibaba), DeepSeek R1, GLM 4.5 - all Chinese.
Meanwhile, Zuckerberg's quietly preparing to abandon open source commitments, talking about "safety concerns" requiring them to be "careful about what we choose to open source."
Kyle's take: This is massive and most people are missing it. The West has essentially ceded the entire open source AI movement to China. Meta was our only champion for open models, and now they're moving away from it.
China named AI a national strategic priority in 2021 - Trump only just did it last week.
While we in the West are regulating and wringing our hands, China is busy building better models and giving them away for free.
To be fair: this isn't charity - it's soft power, platform control, and global influence. Smart play from China, terrible strategic blunder from the West.
💰 Amazon Pays New York Times $20M Annually for AI Training Data
Amazon's cut the first major deal with The New York Times, paying $20+ million per year to use their content (presumably for training Claude and Alexa).
This follows the classic AI playbook: scrape first then pay up when threatened with lawsuits. Don’t ask for permission, beg for forgiveness.
Kyle's take: OpenAI and others scraped everything without permission, now publishers are coming for their payday. For The New York Times ($2.5bn revenue, 15% margin), this $20M is basically pure profit since there's no additional cost.
Expect every major publisher to now cut similar deals. The playbook the AI companies are using is basically saying "sorry we nicked your stuff, here's some cash to make it legal."
Source: Wall Street Journal
Member Question from Mankat: "Do you think companies CISO will pass China open source?"
Kyle's response: Nope, probably not! There's a lot of anti-Chinese sentiment (a lot of it just sinophobia unfortunately…), especially in America, plus legitimate security concerns.
The thing about open source models though - when you deploy them locally, you can set them up with no connection back to the original company. That's kinda the whole point of running locally!
But most companies won't even consider it. They'll just hear "China" and say no, we're not using that.
They'd rather trust Meta than the Chinese government, which may be foolish. But better the enemy you know and all that!
This question was discussed at [25:00] during the live session.
Member Question from Tech Roy: "I've really got to start doing content right?"
Kyle's response: Content is becoming more powerful precisely because building is getting easier. When anyone can vibe code an app in hours, distribution becomes the only moat that matters.
Look at Pieter Levels - built a flight simulator for fun, turned it into £1M ARR in 17 days because he has a massive audience.
Meanwhile, I accidentally won a trip to Sweden just by posting one video about Lovable's buddy codes. I’m off to visit Lovable’s HQ and meet their team in a few weeks because of a 2 minute one-take video.
The future belongs to builders with audiences, not just builders. Start now, be consistent, and remember - the video that takes you viral will always be the one you least expect.
FYI if you are making content in the AI space I have an AI Authority Accelerator that may be useful to get started.
This question was discussed at [38:44] during the live session.
Want the full unfiltered discussion? Join me tomorrow for the daily AI news live stream where we dig into the stories and you can ask questions directly.