# AI This Week: NVIDIA’s Record Revenue, Elon’s Data Centers in Space & Gemini 3’s Insane Performance Source: https://www.youtube.com/watch?v=Ht9XtcV7ZYk Recap page: https://rapidrecap.app/video/Ht9XtcV7ZYk Generated: 2025-11-22T15:32:36.878+00:00 --- ## Quick Overview Nvidia reported record revenues of $57 billion with 62% year-on-year growth driven by AI compute demand, while Elon Musk detailed plans to launch 100 gigawatts per year of solar-powered AI data centers into orbit using Starship, potentially making orbital transport cheaper than transatlantic flights. **Key Points:** - Nvidia achieved $57 billion in revenue, marking 62% year-on-year growth fueled by AI compute demand, with CEO Jensen expecting $65 billion in revenue next quarter. - One speaker argued that Nvidia has become the "central bank for AI," minting "compute" as its currency, which everyone must buy. - The panelists discussed potential challengers to Nvidia, naming Google's TPUs (powering Gemini) and AMD as non-incumbents moving toward a heterogeneous ecosystem. - Elon Musk proposed a path to put 100 gigawatts per year of solar-powered AI satellite compute into orbit, aiming for this to be the lowest-cost way to operate AI at scale. - Musk's vision for orbital compute involves using Starship, which he estimates could eventually make a seat ticket to orbit cheaper than a transatlantic flight, and potentially mining the moon for silicon. - Google announced a $40 billion investment in Texas through 2027 to build new cloud and AI infrastructure, adding 6.2 gigawatts of new energy generation, because Texas can build energy infrastructure fast enough to meet AI demand. - The discussion highlighted a belief that the future form factor for computing is reversing the trend of miniaturization, moving toward larger, coherent AI data centers and superclusters in space. **Context:** The discussion centers on the explosive growth in AI infrastructure, focusing heavily on Nvidia's financial dominance and ambitious future projects from Elon Musk involving space-based computing. Key figures mentioned include Nvidia's CEO Jensen, Google's Demis Hassabis, Anthropic's Dario Amodei, and Elon Musk, who are all driving fundamental shifts in compute, energy, and sovereign technological stacks. The conversation frames current events within the context of approaching the technological singularity and the race for superintelligence. ## Detailed Analysis The episode confirms Nvidia's current supremacy, reporting $57 billion in revenue and 62% YoY growth, positioning them as the essential supplier of compute currency for AI development, although competitors like Google (with TPUs) and AMD are noted as developing alternative architectures. A major segment focuses on Elon Musk's extreme moonshot: deploying 100 gigawatts per year of solar-powered AI compute in orbit via Starship, estimating that orbital transport costs could eventually undercut terrestrial transatlantic travel costs. This plan hinges on Starship operationalizing soon and solving the challenge of radiative cooling in space, which Alex suggests is entirely feasible by pointing heat toward the cooler cosmic microwave background, noting the first H100 unit is already successfully operating in space using aluminum radiators. Furthermore, the panel explored geopolitical AI positioning, noting Saudi Arabia's $100 billion commitment to become an AI superpower, partly through XAI using an Nvidia-backed data center. Google is also massively expanding on Earth, investing $40 billion in Texas to add 6.2 gigawatts of energy generation, signaling that AI is becoming the largest electricity consumer. The overall theme suggests a fundamental shift where compute form factors are getting larger, moving from personal devices to massive data centers and eventually to space-based superclusters, potentially turning parts of the solar system into 'Computronium.' ### Nvidia's Financial Power - Recorded $57 billion revenue with 62% YoY growth driven by AI compute demand - Jensen expects $65 billion next quarter - Argument made that Nvidia mints 'compute' as its currency. ### Competitive Landscape & Architecture - Non-incumbents like Google TPUs and AMD challenge Nvidia's CUDA dominance - Trend moves toward a heterogeneous ecosystem - A speaker is invested in Standard Kernel, which designs chips for specific AI algorithms. ### Anthropic-Microsoft Partnership - Enthropic agreed to $30 billion spend on Azure, with Microsoft and Nvidia investing up to $15 billion in Anthropic - This forms an 'AI power block' - Prediction made that a merger between Anthropic and Microsoft is likely. ### Ethics and Leadership - Leaders like Demis (Google) and Dario (Anthropic) are highlighted for deeply prioritizing ethics independent of profit maximization, contrasting with more business-focused leaders like Jensen, Sam, and Elon. ### Space-Based AI Compute - Elon Musk plans to launch 100 gigawatts/year of solar-powered AI data centers via Starship, potentially cheaper than transatlantic flight - The first H100 unit is operating in orbit using aluminum for radiative cooling. ### Lunar Expansion and Energy - Musk's plan includes mining the moon for silicon to potentially achieve 100 terawatts of energy capacity per year via mass drivers, a concept rooted in Gerard K. O'Neal's work. ### Energy Constraints and China's Lead - China is quadrupling the US in projected electricity generation growth (325 TW vs. 80 TW) - A large drop in coal usage in China is noted as solar economics become superior - US faces a critical electricity gap between 2030 and 2035 as chip fabrication accelerates faster than fusion/new power source deployment. ### US Infrastructure Investment - Google commits $40 billion in Texas through 2027, adding 6.2 gigawatts of energy generation - Texas is favored due to its ability to build energy infrastructure quickly and its friendly regulatory environment for data centers.