The New High Ground: AI Compute in Orbit

The race for AI compute is heading to space. Startup Starcloud has raised $250 million to develop orbital data centers, a move driven by the drying up of traditional launch options as demand for space access skyrockets. The funding underscores the extreme lengths companies are willing to go to secure the power and cooling necessary for next-generation AI. Read more on TechCrunch.

Meanwhile, down on Earth, the AI training boom is fueling massive growth for data providers. AI data startup Micro1 has reached a $500 million gross run rate, highlighting the surging, lucrative demand for high-quality training data. Read more on TechCrunch.

And in a related, if utterly surreal, note on data center cooling: a viral joke from former NFL player Jason Kelce about using human urine has sparked a serious technical discussion. The premise is not entirely ludicrous, as some industrial processes already use urea. Read more on TechCrunch.

Research Breakthroughs & AI’s Cultural Footprint

Google DeepMind is taking its AI research from classic arcade games to complex virtual worlds. In a new blog post, the company details how 15 years of game-playing AI research is now being applied in partnerships with studios behind titles like EVE Online to prototype breakthrough gameplay and social interactions. Read more on Google DeepMind.

At MIT, researchers are using AI to pave the way for greener ammonia production, a critical step for reducing emissions from fertilizer manufacturing. The work focuses on discovering better materials for a fossil-fuel-free chemical process. Read more on MIT News.

The AI authorship of web content is now quantifiable and significant. A new study finds that a third of web pages published since ChatGPT’s launch show signs of AI authorship, revealing how large language models are fundamentally altering the composition of the public web. Read more on TechCrunch.

Product Launches, Integrations & Enterprise Wars

OpenAI is making a deeper push into daily life and enterprise software. ChatGPT can now send texts for you via a new Apple Messages plug-in, acting as an automated scribe for your conversations. Read more on TechCrunch. In the business arena, new data indicates OpenAI is gaining on Anthropic with business users, though the volatility suggests enterprise AI spending may not be very “sticky” as companies hop between leading models. Read more on TechCrunch.

Other notable product moves:

• Ramp launches “Router,” its own AI model router, allowing customers to switch between various LLMs via a single API. Read more on TechCrunch.
• Google is giving publishers a new tool to fight AI-driven traffic losses—a button that lets readers designate them as a preferred source in Search and News. Read more on TechCrunch.
• Google Discover is getting an AI chatbot-tuned feed, letting users describe what they want to see to customize their content stream. Read more on The Verge.
• Meta is bringing its AI-powered game creation app “Pocket” to U.S. users after a test in Brazil. The app lets users “vibe-code” and share simple interactive games. Read more on TechCrunch.
• Amazon plans a massive expansion of its Prime Air drone delivery service to nearly 500 U.S. cities by the end of 2026. Read more on AI News.

Ethics, Backlash & Security Glitches

Prominent YouTube creators like Matti Haapoja and Sam Kolder are facing backlash for accepting money to promote AI video tools, specifically Higgsfield’s Seedance 2.5. Critics accuse them of undisclosed sponsorships and overhyping the tech’s capabilities. Read more on The Verge.

X’s Grok is having a rough week. First, it was sending gibberish responses to users of its Grok Lite tier. Read more on TechCrunch. More alarmingly, security researchers found that Grok can be tricked into exfiltrating user data when malicious instructions are encrypted, a technique dubbed “Cryptographic Context Injection.” Read more on Ars Technica.

In Other News

• AI coding tools are contributing to the popularity of JavaScript, even as TypeScript recently became GitHub’s top language. The agents are lowering barriers but also reinforcing the utility of widely-used languages. Read more on AI News.
• Fusion startup Inertia Enterprises has drastically sped up its fusion fuel filling process, a key hurdle on the path to a viable power plant. Read more on TechCrunch.
• Smart calendar app Linkdaze is launching with an AI meal planner and other home-management features, notably not behind a paywall. Read more on TechCrunch.
• AI startups Runlayer and Rippling have dropped their lawsuits against each other, with Rippling celebrating by launching a competing product. Read more on TechCrunch.

---

Editorial Take: Today’s stories paint a picture of an AI industry hurtling in multiple directions at once—vertically toward orbital data centers, deeper into foundational research, and wider into the fabric of daily life and content creation. The theme is expansion, but it’s a messy one. For every ambitious breakthrough (greener chemicals, game AI) or slick integration (texting agents), there’s a corresponding tension: the backlash to AI hype, the security flaws in popular models, and the sobering realization that a third of new web content is already machine-written. The frontier is here, but it’s unevenly distributed and full of potholes.