Jon Smith π²

This Week in Theta — what shipped, what's getting attention. Two substantial stories this week: - Theta EdgeCloud disaggregation benchmark - GPU Node Browser launches on thetaedgecloud.com Video below.

🆕 Just launched: The ThetaNext FAQ page! A brand new FAQ page is now live on thetanext.io/faq covering everything you need to know about staking #THETA and #TFUEL with ThetaNext. From getting started to Booster Nodes, rewards, and more, it's all in one place now. Something missing or unclear? Drop it in the comments 👇

We're looking forward to joining @mms_Blockchain at Berlin Blockchain Week next month! It's going to be a great day of talking about the future of our industry 🚀

Decentralized networks are powering the future of AI, reshaping how the world accesses compute. @Theta_Network is joining us at Blocks & Brews and we can't wait to talk about what's next. June 18 | Berlin Blockchain Week RSVP👇 luma.com/gu20fy2p

🟢 Our new GPU node browser is live on thetaedgecloud.com. It's now even easier to browse AI compute and filter by GPU count, memory and reliability, then see exactly what each node offers before you rent. Discover for yourself: thetaedgecloud.com/dashboard/ai/s…

Brand new GPU browser page, 1-click to one of the most sought after assets. thetaedgecloud.com/dashboard/ai/s…

Great start to the Western Conference Finals for the @GoldenKnights. 🏒 We're proud to be partnered with VGK as Sir Lancelbot powered by @Theta_Network continues helping fans all season long, already answering thousands of questions from 1600+ fans.

For production AI, that means more predictable performance under uneven workloads. Operators can scale prefill or decode separately, reduce interference between users, and get more useful output from existing GPUs. Read more: medium.com/theta-network/…

The main result was that response start times stayed stable as prompts got longer. ✅ At 1,000, 2,000, and 4,000 words, response start times stayed around 0.8 seconds. Heavy prompts did not disrupt generation for other users. ✅

We ran an experiment at Theta EdgeCloud to test a more efficient way to serve LLMs in production. Long prompts can slow down everyone sharing the same GPUs. We tested an architecture designed to keep latency/throughput stable as prompt length and traffic increases. 🧵

BlackRock's CEO thinks compute will become a tradable asset. CME Group has just announced the first contracts. We've written about why this is happening, and why pricing the shortage is not the same as solving it. medium.com/theta-network/…

Theta offers GPU compute matched to your workload, instant and affordable. No more paying for what you don't use. No more H100s when distributed inference will do. Sign up at thetaedgecloud.com. No call required. 🖥️

La Gran Migración: Del “Entrenar” al “Ejecutar” (y por qué Theta es la autopista) 1/6 Hay un dato que McKinsey acaba de soltar y que redefine la década de la IA: Para 2030, la inferencia superará al entrenamiento como la principal carga de trabajo. Traducción: La era de construir modelos gigantes está dando paso a la era de ejecutarlos millones de veces al día en cada app, chatbot y recomendación. El foco se mueve de la fábrica a la calle. 🧠➡️🌍

McKinsey expects inference to overtake training as the main AI workload by 2030. That means more AI spending will shift from building models to running them every day in products like chatbots, search, recommendations and creative tools. 🧵

We're pleased to welcome City St George's, University of London to Theta EdgeCloud's academic network, our 34th university partner globally & our 2nd in the UK. 🧵

This Week in Theta — what shipped, what's getting attention. Three substantial stories this week: - PageIndex shipped for EdgeCloud AI agents - Theta × Peking University: two WWW'26 papers - Yonsei University: research powered by EdgeCloud Video below.

Wow, THIS is what frontier #AI infrastructure looks like. @yonsei_u Yonsei University in Korea just dropped TWO papers on personalized reward modeling - trained on @AWSAI @awscloud #Trainium via @Theta_Network #EdgeCloud Hybrid. 🔥🔥🔥😎😎🤠🤠🤠 🔥First decentralized platform AWS approved for custom AI silicon. 🔥First blockchain network to deploy Amazon's next-gen AI chipsets. 🔥World-class academic AI is moving onto decentralized rails. Let's JUST GOOOOOOOOOOOOOOOOOO #THETA $THETA $TFUEL $TDROP #TDROP #TFUEL 🔥🔥🤠🤠😎😎🔥 read more here : medium.com/theta-network/…

Theta EdgeCloud is the first decentralised compute platform approved by @AWS to run their custom AI silicon. We did it in collaboration with @yonsei_u, one of Asia's top AI research institutions. 🧵

@HighAltitudeInv 🙏

#Theta super moon incoming!

🚨🚨🚨 Theta EdgeCloud just got a massive upgrade signal 🚀 Two groundbreaking papers with Cornell accepted to The Web Conference 2026... Alzo - Hierarchical RL that auto-tunes DAG blockchains in real time. No more manual guesswork, self-optimizing throughput & latency like never before. DARA - RL-finetuned LLMs mastering few-shot budget allocation with lightning adaptability. This is Theta building intelligent autonomous decentralized infrastructure. EdgeCloud isn’t just decentralized compute. It’s AI-powered, self-healing, and ready to dominate. The future of Web3 Edge is here 🔥 x.com/i/status/20524…