Daily TEA – Code Reds, DePIN Dreams, and AI Superchips
Altman, Telegram, Bitcoin, Microreactors, AWS chips and more.
Hello, dear TEA-mates, here’s what you need to know today.
1. ⚠️ Altman Declares ‘Code Red’ To Turbocharge ChatGPT
OpenAI CEO Sam Altman has reportedly sent an internal “code red” memo instructing staff to urgently improve ChatGPT’s speed, reliability, personalization, and answer quality amid intensifying competition from Google’s Gemini models and other rivals. The directive calls for delaying plans such as ads, AI shopping tools, and other experimental products so the company can focus on its flagship chatbot experience. Altman is said to have told employees that OpenAI will ship an upgraded ChatGPT intended to surpass Google’s Gemini, underscoring how quickly the generative AI race is evolving. (Read More)
🫖 TEA For Thought: With Google’s Gemini and a wave of free, open-source models from China, it makes sense for OpenAI to pivot back to its core product, ChatGPT, instead of prioritizing monetization.
2. 🕸️ Telegram’s Durov Pushes Cocoon, A Decentralized AI Network
Telegram founder Pavel Durov highlighted the launch of Cocoon, a decentralized confidential computing and AI network built on The Open Network (TON), which lets GPU owners earn tokens by providing private compute for AI requests. Durov has pitched Cocoon as an answer to privacy and cost concerns around centralized AI services run by major cloud providers, with the network already processing its first user AI queries and paying GPU providers in TON. The project taps into growing enthusiasm for DePIN and decentralized AI infrastructure, positioning Cocoon alongside other emerging efforts to blend blockchain and AI at scale. (Read More)
🫖 TEA For Thought: DePIN and DeAI — decentralized AI built on blockchain networks — looks like the future, and projects like Sentient AI and Planck suggest this is just the start of a long fight.
3. ₿ Michael Saylor’s Firm Builds $1.4 Billion Bitcoin Shock Absorber
Michael Saylor’s Bitcoin treasury firm Strategy has assembled a roughly $1.4 billion cash reserve, funded via new stock issuance, to cover dividend and interest payments and avoid having to sell Bitcoin during sharp market downturns. The company, which holds around 650,000 BTC (about 3% of total supply), framed the move as a shift from its prior strategy of relentlessly buying coins, opting instead to prioritize balance sheet resilience as Bitcoin trades about 30% below recent highs. The reserve is designed to cover at least 21 months of obligations, even as the firm cut its year-end Bitcoin price forecast to as low as 85,00085{,}00085,000 from a previous 150,000150{,}000150,000. (Read More)
🫖 TEA For Thought: Digital Asset Treasury companies will eventually become obsolete; as it gets even easier for people to hold crypto in their own wallets, why buy shares just to hold the assets indirectly?
4. ⚛️ Microreactor Startup Antares Raises $96M For Land, Sea, And Space Power
U.S. nuclear startup Antares has secured a $96 million Series B round, including $71 million in equity and $25 million in debt, to advance its R1 microreactor capable of producing between 100 kilowatts and 1 megawatt of electricity. The company plans to deploy its TRISO-fueled microreactors for commercial, defense, maritime, and space applications, and was selected as one of 11 participants in a U.S. Department of Energy pilot program targeting at least three operational reactors by July 4, 2026. Antares aims to demonstrate its reactor for the DOE next year and switch on a full-power unit by 2027, betting that compact, modular nuclear systems can meet demanding off-grid and off-planet energy needs. (Read More)
🫖 TEA For Thought: The future of energy is above us, in the skies and in space.
5. 🤖 Amazon Unveils Trainium3 AI Chip And Teases Nvidia-Friendly Future
Amazon Web Services introduced Trainium3, the latest version of its in-house AI training chip, promising up to four times the performance and around 40% better energy efficiency than the previous generation, which could cut training and inference costs significantly for customers. AWS also previewed Trainium4, which is in development and will support Nvidia’s NVLink Fusion interconnect, allowing future systems to combine Amazon’s custom chips with Nvidia GPUs in the same high-performance clusters. The roadmap signals AWS’s strategy to control more of the AI compute stack with its own silicon while remaining tightly integrated with Nvidia’s ecosystem to attract large-scale AI workloads. (Read More)
🫖 TEA For Thought: Like Google, Amazon is positioning itself to own the full AI stack — from energy and chips to cloud infrastructure and applications via bets like Anthropic.
Prompt Tip of the Day: Act as a thoughtful end‑of‑year reflection coach
Create 31 daily reflection questions for the last month of the year (one for each day), designed for a busy, growth‑oriented adult who wants to honestly review the past year and set intentions for the next.
Requirements:
Organize questions in order, labeled Day 1 through Day 31.
Make each question open‑ended and specific (avoid yes/no).
Ensure coverage of:
Gratitude and highlights of the year
Challenges, failures, and lessons learned
Relationships and community
Work, creativity, and impact
Health, habits, and routines
Identity, values, and meaning
Hopes, goals, and priorities for next year
Vary the depth: mix a few light, easy prompts with deeper, more introspective ones.
Use clear, simple language that invites honest self‑inquiry.
Avoid clichés and generic “New Year resolutions” phrasing; make prompts feel personal and reflective.
After listing the 31 questions, add a short note suggesting how to use them as a daily practice (e.g., how long to write, how to revisit answers at year‑end).
TEAHEE Moment
Stay sharp, stay informed. See you tomorrow.
And if you enjoyed today’s TEA, follow along on social for more: Twitter/X.