🧠 Food for thought

Implications, context, and why it matters.

Meta is buying data while it considers licensing AI models

• Meta is weighing a license for Google’s Gemini while it ramps up efforts to buy high-quality training data 1.
• Meta signed a multiyear contract with News Corp that pays up to $50 million a year for content from outlets like The Wall Street Journal, to pull fresh information for users of its AI products and to train on other material such as story archives 2.
• The News Corp contract sits within a wider plan that includes similar content licensing agreements with USA Today, People, CNN, Fox News, plus other outlets 3.
• This mix of data buying and possible model licensing looks like a bid to close an AI performance gap. Meta’s leaders have not made a decision on licensing Gemini 4.

Even tech giants may become customers in the AI arms race

• Meta’s reported struggles add to the case that even with huge capital spending, building a state-of-the-art AI model that can compete with the top tier remains exceptionally hard 4.
• A market shift may follow where a small set of firms such as Google and OpenAI supply core models to more of the industry, including other tech giants 1.
• Other companies could then put more energy into products built on licensed models, including Meta’s AI-powered shopping research tool 5.
• This pattern could concentrate influence among a few foundational model makers, leaving some of the biggest tech firms relying on rival suppliers.