Welcome to the latest AI News Roundup! Today, we’re diving into major moves from Google, OpenAI, Apple, Microsoft, Amazon, Sora, and more—covering futuristic data centers, billion-dollar AI deals, next-gen image/video models, and the latest legal drama around AI-powered browsing. Let’s jump right in!
🔠Google’s Suncatcher: AI Data Centers... in Space?!
Google revealed “Project Suncatcher”—a vision to build massive AI data centers in space! Why? Here’s the big idea:
- Solar Power in Space: Solar panels in the right orbit can be 8x more efficient than on Earth.
- Solving Earth’s Power Bottleneck: Data centers on Earth are hitting power limits; in space, energy is abundant.
- Communication Challenge: High-bandwidth satellite clusters will beam signals between arrays—satellites flying just a few hundred meters apart!
- Surviving Space: Google’s custom AI chips, tested against space radiation, survived intense conditions.
- Cost Factor: By 2030s, launch costs may drop to $200/kg with reusable rockets, thanks largely to SpaceX.
Google bets big on the future—could this revolutionize the AI world?
🤝 OpenAI’s Mega Deals: The TEAR is Real
OpenAI is on a “TEAR” with a string of massive compute infrastructure deals:
- $500B Stargate deal
- $100B Nvidia / $100B AMD
- Fresh deals with Intel, TSMC, Amazon, Microsoft, Oracle
- Considering a trillion-dollar IPO
- After restructuring, OpenAI partnered with AWS for compute—the fight for processing power is fierce!
🍏 Apple Teams Up With Gemini for Next-Gen Siri
Apple is reportedly paying Google $1B per year to power the next Siri (“Apple Intelligence”) using a fully custom Gemini model:
- Apple runs this tech in its private cloud (not Google’s!), optimizing control and privacy.
- OpenAI was a contender, but Apple chose Google for key gen-AI features.
- Users can expect a smarter, more natural Siri as soon as next spring.
⏸️ OpenAI’s New Feature: Interrupt Long Queries
OpenAI shipped a must-have quality-of-life feature: you can now pause/interrupt long-running queries, add more context, and resume from where you left off—a real win for deep research and productivity!
🎨 Microsoft’s MAI Image One: Next-Gen Image Creation
Microsoft launched MAI Image One:
- In-house image generation model, top 10 on LM Arena.
- Prioritizes creativity, avoids generic outputs.
- Designed with feedback from professional creators.
- Excels at photorealism: landscapes, lighting, nuanced detail.
- Try it on bing.com/create!
🎥 Vidu Q2: The Video Gen Model to Watch
- Vidu Q2 launched at #8 on the text-to-video leaderboard, surpassing Sora 2 and others.
- Supports multiple reference images for controllable results (e.g., character consistency across scenes).
- Variants: up to 8s of 1080p video. Pro version: $6–10 per minute.
- Standout features: great lighting, realistic water/physics, higher creativity.
Sample: “A young elf rides a unicorn through a glowing enchanted forest.” Magical!
⚖️ Amazon Threatens Perplexity Over Agent Buying
Amazon issued a legal threat to Perplexity: stop using AI agents to buy Amazon products for users.
- Why? If AI agents shop for us, we bypass Amazon’s ad-driven business model (no chance to serve you ads as you shop).
- Larger implications: as AI agents grow, expect more legal clashes between tech giants and the new agentic internet.
🧑‍🎤 Sora Updates: Character Cameos!
OpenAI’s social video network Sora now lets you:
- Upload characters, pets, or designs to appear as cameos in your AI-generated videos.
- Feature your custom character in any scene—cool, creative, and fun!
Top comments (0)