Apple unveils MLX with no major headlines for the end user, Anthropic pulls Fable within hours under government pressure, and SpaceX debuts on the stock market on its way to a trillion. Episode 14 of Tokenizados reviews the week AI collided with regulators, markets and the physics of hardware.
Apple WWDC: MLX infrastructure
A WWDC without major consumer announcements hides a relevant technical advance: MLX, Apple’s infrastructure for training and inference on its own silicon. The most striking part: connecting several Mac Minis over Thunderbolt to pool RAM and run heavy models as if it were a rack of servers.
Frontier Ecosystem: the learning loop
Starting from a post by Satya Nadella, we explore why the competitive advantage isn’t in choosing the best model, but in building a learning loop that feeds back human capital and token capital. Companies that build their own AI sovereignty through private EVALs turn their processes into proprietary IP and stop depending on a single provider.
The Fable ban and the power of OpenRouter Fusion
We debate the swift withdrawal of Fable after a supposed warning from Amazon about hacking vulnerabilities. As an alternative, we present OpenRouter’s Fusion API, which combines frontier models — like GPT-5.5 and Opus 4.8 — to get Fable-level results at half the cost.
Agentic dating: Charisma as a Service
The dating-assistant segment has grown 333% in a year. Tools like Rizz (+7.5M downloads), YourMove AI or Plug AI automate everything from the first message to replies in stalled chats. We explore the future of “agent-to-agent” relationships, where profiles filter affinities before the humans even get to meet.
SpaceX goes public and SoftBank’s troubles
The SpaceX IPO debuts at $135 per share and quickly reaches a $2 trillion valuation, cementing Elon Musk as the first trillionaire. By contrast, SoftBank’s shares fall due to the difficulties in closing a $6 billion loan tied to OpenAI’s valuation.
Ilya Sutskever: from OpenAI to SSI
We close with the profile of Ilya Sutskever, co-author of AlexNet and a central figure in deep learning. After leaving OpenAI he has taken on the role of CEO at Safe Superintelligence Inc. (SSI), valued at $32 billion with no commercial product, with a single goal: building a safe superintelligence through algorithmic innovation.
Thanks for joining us one more week on Tokenizados. Subscribe so you don’t miss any update from the world of AI.