8. DGX Spark = personal AI supercomputer Imagine training your own LLM at your desk. DGX Spark is the dev-friendly AI rig built for workstation scale. → Shipping via ASUS, Dell, MSI → Trains large models locally → No server farm needed

🚨 BREAKING: NVIDIA JUST announced roadmap for physical AI, robotics and national-scale AI factories. Here’s a breakdown of the top important announcements: 🧵👇 1. DeepSeek R1 is now 4x faster, setting the standard for AI in inference and reasoning.

2. NVIDIA is building the GPT of humanoid robots. They just launched Isaac GR00T N1.5 - a foundation model for general purpose robotics. Here’s how it works: → A human demos the task once → Cosmos (their physics AI model) generates 1,000s of variations → Omniverse simulates the motions in high fidelity → The robot trains entirely in simulation → Then fine-tunes itself in the real world Robots can now learn general skills across tasks, tools, even body types with just one human demo. AI isn’t just thinking in text anymore. It’s perceiving. Reasoning. Moving. Physical AI is here and it’s training itself.

3. Open-source physics engine for robotics (July launch) Built with Disney + DeepMind. → GPU-accelerated → High-fidelity soft + rigid body simulation → Differentiable → Real-time training Will become the training ground for physical AI—humanoid robotics, drones, self-driving systems, and more.

4. Grace Blackwell is the new industrial brain → 1.5x inference speed → 2x networking bandwidth → 1.5x memory → Fully liquid-cooled rack-scale system → 40 PFLOPS per node—replacing supercomputers like Sierra Already live at xAI, CoreWeave, Oracle, and others.

5. CUDA-X is quietly eating every domain CUDA isn’t just about graphics anymore. It’s powering: → 6G radios (Aerial) → Genomics (Parabricks) → Weather sim (Earth-2) → Quantum computing (cuQuantum) → Chip lithography (cuLitho) → Supply chains (cuOpt) → Medical imaging (MONAI) → Sparse simulation (cuSPARSE) CUDA-X is now the OS layer for accelerated science.

6. AI factories are the new industrial plants. In the past, you built factories to manufacture cars or electronics. Today? You build them to manufacture intelligence. These aren't your typical data centers. They’re purpose-built to train and run massive AI models. Think: → Tens of thousands of GPUs → Liquid cooling racks pulling 100kW+ per unit → Token output as the new “productivity” metric Jensen Huang put it simply: “You apply energy in, and it produces something incredibly valuable—tokens.” This shift is already real: → xAI’s Stargate: 4 million sq ft, 1 gigawatt, $50B+ compute → CoreWeave, Oracle, Microsoft: spinning up multi-billion dollar AI facilities → Foxconn x NVIDIA x Taiwan Gov: building a national AI supercomputer for Taiwan It’s an arms race but with GPUs instead of missiles. We’re entering a world where: - Every country will want its own AI factory - Every major company will run its own internal model - Every startup will rent capacity like they rent AWS today And just like the industrial revolution needed electricity, This one needs Blackwell.

7. In collab with NVIDIA + Foxconn, Taiwan is building a national-scale AI factory. Blackwell-powered. Open to researchers, startups, and industrial use...

I hope you've found this post helpful. If you want to level up with AI in your daily life, subscribe here:

Automated GitHub, GitLab, Bitbucket and Jira backups, security compliance, data migration and every-scenario-ready Disaster Recovery for 360 cyber resilience. Schedule a custom demo or try 14 days for free.

Preeeeetty good pitch

@Threadreaderapp unroll

I can imagine someone powerful with nasty intentions and a lot of many training humanoids for awful purposes. I hope I'm wrong