Character AI = Bankrupt

This link leads to one of the most interesting things I’ve seen in AI recently.

What’s cookin’? Character AI, one of the biggest AI companion platforms, has filed for Chapter 11 bankruptcy after struggling with high operating costs, lawsuits, and growing regulatory pressure. The app lets people chat and roleplay with AI-powered characters, but the company says building and running the service became too expensive to sustain. Character AI has also faced multiple lawsuits over chatbot safety, including allegations involving self-harm to younger users — some even resulting in death. As part of the bankruptcy process, Disney has agreed to serve as a $20 million stalking-horse bidder for some of Character AI’s intellectual property. The platform will remain online for now, although the company says it could eventually be shut down.

🤔 Hunter’s take: This is a useful reminder that massive user engagement does not automatically make an AI business healthy. Companion apps are expensive to run, difficult to monetize, and unusually exposed to safety risks. In AI, getting millions of people hooked on a product may actually be the easy part. Keeping it safe? That’s the real challenge.

What’s cookin’? Nvidia and Cisco are working on a new piece of networking hardware (or “optical connector”) that could help AI data centers connect more chips without using as much space or electricity. The technology, called USFP, is basically a much faster connection between servers and networking equipment. Nvidia says it can move four times more data through the same amount of space than current systems, while also being more energy efficient. That matters because modern AI systems rely on thousands of chips constantly sending huge amounts of information back and forth. If those connections get too crowded or slow, the entire system can become less efficient.

🤔 Hunter’s take: The space part is what jumps out to me. AI companies are spending billions building enormous data centers, so every extra rack, switch and networking layer takes up valuable room. If you can move four times more data through roughly the same physical footprint, you can pack more AI computing power into buildings you already have. That could make existing data centers much more valuable without simply making them bigger or building more.

Here’s your chance to get a straightforward answer, from an expert in the field.

Hi — I’m Hunter, a PhD candidate whose work has appeared in major academic journals and popular tech outlets. I founded FryAI to make staying ahead of AI clear, accessible, and fun.