Daily TEA – Orbiting GPUs, AI Futures, and On-Chain Capital
Space data centers, 2027 scenarios, AI licensing, RWAs, AI Santa and more
Hello, dear TEA-mates—here’s what you need to know today.
1. 🛰️ Nvidia-Backed Starcloud Trains First AI Model in Orbit
Nvidia-backed startup Starcloud has successfully trained an AI model in space using its Starcloud-1 satellite, which carries an Nvidia H100 GPU with roughly 100 times the compute of previous space-based chips. The system is running and querying Gemma, Google’s open-source large language model, marking the first time an LLM has been trained from orbit and demonstrating the feasibility of space-based AI data centers for high-intensity workloads like satellite imagery analysis. (Read More)
🫖 TEA For Thought: This is a huge milestone—space is no longer just for exploration, it is becoming the front-line battleground of the AI era, where computation and energy are desperately needed and everything in the sky points to the future.
2. 🔮 AI-2027 Scenarios Warn of Escalating Cyber Risks and Soaring AI Capex
The AI-2027 scenario report outlines a 2026 world where cybersecurity threats intensify alongside industrial espionage, while China consolidates national AI efforts under a central “DeepCent” hub that controls a large share of the country’s compute. The report also projects agentic models like “Agent-1-mini” displacing many entry-level roles and global AI capital expenditure reaching on the order of $1 trillion, with power demand measured in tens of gigawatts, underscoring how rapidly AI infrastructure could scale over the next few years. (Read More)
🫖 TEA For Thought: The AI wave is moving fast and everyone feels it, but it is still not as fast as the hype suggests—there is still time to jump in and ride the trend.
3. 📜 RSL 1.0 Launches as Web’s New Pay-to-Scrape AI Licensing Standard
A new open licensing framework called Really Simple Licensing (RSL) 1.0 has been formally recognized as an industry standard, giving publishers a structured way to set rules and pricing for AI companies that scrape and use their content. RSL lets sites stay in traditional search while opting out of AI training and generative features, and supports models like subscription fees, pay-per-crawl, pay-per-inference, and “contribution” payments for nonprofits and independent creators. (Read More)
🫖 TEA For Thought: RSL 1.0 is a real milestone—pairing a pay-per-crawl model with something like an x402-style blockchain micropayments rail could finally make fast, cheap, high-volume, on-chain payouts to publishers practical, cutting out intermediaries while keeping every transaction transparently recorded.
4. 🏛️ Superstate Opens On-Chain Capital Raises for SEC Firms
Superstate has introduced an “Opening Bell”–style infrastructure that lets SEC-registered public companies issue tokenized shares on blockchains like Ethereum and Solana, settling investments in stablecoins from KYC-verified investors in real time. By keeping offerings fully compliant with U.S. securities rules while moving issuance and settlement on-chain, the platform aims to make onchain capital markets a mainstream option and tighten the link between public equities, crypto rails, and global investors. (Read More)
🫖 TEA For Thought: This is essentially real-world assets in action, creating a practical bridge between DeFi and TradFi.
5. 🎅 Tavus’ AI Santa Keeps Users Chatting for Hours
AI startup Tavus has rolled out an upgraded “AI Santa” experience that lets users talk to a video, phone, or text-based Santa powered by the company’s real-time agent stack, which can see, hear, and respond with humanlike expressions. The new version remembers users’ interests, can search the web or draft emails, and has seen such strong engagement that many users spend hours per day chatting and hit daily limits, raising both growth opportunities for Tavus and questions about how young children interact with emotionally convincing AI. (Read More)
🫖 TEA For Thought: This is a classic GPT-wrapper that has clearly found product–market fit—especially during the holidays, who doesn’t want a little extra AI-powered magic?
Prompt Tip of the Day: Say “Roast this, then fix it”
“Roast this email draft, then fix it”
Gets brutal honest critique first (what’s weak, awkward, unclear). Then provides the improved version with those issues solved. Two-phase feedback.
TEAHEE Moment
Stay sharp, stay informed. Have a great weekend and see you Monday!
If you enjoyed today’s TEA, follow along on X for more real-time links and threads: the_era_arc.






