Innlegg

When a very busy big shot suddenly posts a long article, there’s usually a reason. First, Huang went on Twitter to post, then Zuckerberg wrote a 6,500-word long article in one go. Recently, both have been emphasizing the same thing: open source. Yesterday, Zuckerberg published a long article titled "The Future Belongs to Everyone," and on the same day, Meta open-sourced the new model Muse Glimmer, also previewing that stronger model weights will be released soon. The core of the long article is one sentence: superintelligence should not be concentrated in the hands of a few companies; it should be distributed to everyone, becoming a personal super assistant. It sounds great. But looking at Meta’s overall actions, it’s very clear. Open-sourcing model weights means actively lowering the barrier for others to make products; At the same time, they continue to lock in computing power, glasses, social data, and advertising systems. The easier it is for others to use your model, the harder it becomes to bypass your ecosystem for monetization in the end. This is completely opposite to OpenAI and Anthropic’s insistence on closed source and emphasis on "safe and controllable" approaches. Both sides use the "safety" narrative to serve their own business models: one says centralization is safe, the other says decentralization is safe. The louder they say "The future belongs to everyone," the more you need to see clearly who is defining the rules and who is collecting the toll.#AIInfraEarningsWatch #CPIToResetFedBets #Nvidia500BAIInfra