Company@X

Signal of the Day

Microsoft has officially taken delivery of its first production Nvidia Vera Rubin GPUs for Azure data centers, marking the first physical deployment of Nvidia’s highly anticipated next-generation AI architecture in the public cloud.

Key Announcements

Starcloud · Source Starcloud has raised $250 million at a $2.3 billion valuation to build orbital AI compute capacity, backed by Manhattan West, Nvidia, Cisco, Benchmark, and others. The space startup is moving into a new 100,000-square-foot facility next week to ramp up satellite production to 100 units per week. This aggressive scale-up aims to bypass terrestrial data center power and real estate constraints by deploying high-density compute infrastructure directly into space.

OpenAI · Source OpenAI is dropping API and credit pricing of its flagship GPT-5.6 Sol model by over 20% for the next three months. The discount is available on the API immediately and is rolling out across eligible plans for ChatGPT Work and Codex credits, while standard consumer subscription pricing remains unchanged. This aggressive temporary cut targets enterprise developers, amplifying the price wars among top-tier foundation model providers.

Waymo · Source Waymo has launched its newest autonomous vehicle, the “Ojai,” which is now active and available to all riders in San Francisco, Los Angeles, and Phoenix. The vehicle features automated sliding doors, extensive legroom, and built-in Gemini in Waymo integration on the cabin’s screen. Riders can use voice commands hands-free to adjust climate controls, adjust seating positions, and ask for localized recommendations.

AWS · Source Amazon Web Services has launched AWS Glue 6.0, introducing a 30% price reduction alongside support for Apache Iceberg v3 with VARIANT shredding. The serverless integration engine now runs Spark 4.1 and Python 3.13 runtimes, supporting Spark Declarative Pipelines and Real-Time Mode. The release is structurally designed to streamline ETL operations, allowing developers to write significantly less boilerplate orchestration code.

Google · Source Google is offering college students free one-year plans for Google AI Pro (US) and Google AI Plus (internationally), complete with a dedicated learning hub and diagnostic study tools. Most notably, Google is introducing Deep Research into Gemini Live, enabling students to trigger comprehensive, multi-step research reports using voice commands. Once initiated, the research agent works in the background, allowing the user to lock their screen or switch tasks while it generates the final report.

xAI · Source xAI has expanded access to its autonomous “Grok Bot” agents to all SuperGrok Plus, Cursor Pro+, and Cursor Teams subscribers, alongside rolling out a limited free trial. Grok Bots are designed to function as always-on AI colleagues equipped with their own virtual computers to execute software workflows without logging off. The deployment signals xAI’s shift from search-centric chatbots toward agentic productivity tools capable of independent computer use.

LightwheelAI · Source In partnership with Hugging Face, LightwheelAI has open-sourced EgoSuite-Open100K, a massive 100,000-hour egocentric human dataset for physical AI training. The commercially licensed dataset features over 15,000 real-world tasks and scenes—spanning factory floors, kitchens, and retail spaces—fully annotated with hand/body pose data and wrist camera coverage. The first 10,000 hours are live immediately, establishing an open public data layer to scale embodied robotics.

Also Noted

• Tesla (Source): Cumulative FSD (Supervised) driving distance in South Korea has officially crossed 50 million kilometers as vehicles continue scaling real-world fleet data collection.
• Prodigy Research (Source): The YC S26 AI trading lab has launched a specialized quantitative foundation model that achieved 140% returns in live trading during its YC batch, claiming to beat Claude Fable and GPT-5.6 Sol at autonomous quant research.
• Array Labs (Source): Awarded a contract extension from the U.S. Navy’s Office of Naval Research to continue development and study of space-based Airborne Moving Target Indication (AMTI) concepts.
• Lamb Labs (Source): Released Woolly, a post-trained Qwen3-8B model optimized to run up to 2-3x faster on math and coding prompts while fitting directly onto their custom silicon.
• Spline (Source): Launched Spline V2, a complete rebuild of the 3D editor introducing an AI Agent Mode, support for the Spline Model Context Protocol (MCP), and a faster WebGPU engine.

📈 I can draft a detailed investment analysis of the space-based compute landscape, comparing Starcloud’s orbital model with traditional terrestrial data center trends from the sources.