SpaceX Acquires xAI in Historic $1.25 Trillion Merger to Pioneer Orbital AI Data Centers

SpaceX Acquires xAI in Historic $1.25 Trillion Merger to Pioneer Orbital AI Data Centers

SpaceX announced on February 2 its all-stock acquisition of xAI, valuing the combined company at $1.25 trillion—SpaceX at $1T and xAI at $250B—via Elon Musk's memo on the company's site.[1][2][3] The deal unites Musk's ventures to build space-based AI infrastructure, integrating Grok AI with Starlink's satellite network for orbital data centers powered by solar energy and Starship V3 launches starting mid-March.

This merger eyes terawatt-scale compute via million-satellite constellations and lunar factories, marking the largest M&A in history. On X, Musk's announcement fueled hype around "space-based compute revolution," positioning the entity to dominate AI hardware beyond Earth.

Anthropic Unveils Claude Code Security Preview, Erasing $10B from Cybersecurity Stocks

Anthropic dropped Claude Code Security into limited research preview on February 21, an AI tool that autonomously hunts software vulnerabilities and proposes fixes after a year of cyber research.[4][5][6] Within an hour, it tanked stocks like CrowdStrike (-6.5%), Cloudflare (-6%), and Okta (-5.7%), wiping out $10B in market cap.

Traders on X called it a direct threat to the $100B+ cybersecurity industry, with the preview signaling AI's power to disrupt legacy players by spotting flaws humans miss. This isn't full release yet, but it underscores how AI agents could automate security workflows overnight.

OpenClaw Framework Goes Viral as Local LLM Agent Orchestrator, Drawing Karpathy Praise and Security Warnings

OpenClaw, an open-source framework from early February (spun from Moltbot/ClawdBot), lets you run persistent local AI agents for email, calendars, and more via WhatsApp or Telegram—on anything from laptops to Raspberry Pi using LLMs like Claude.[7][8][9] It's exploding in popularity, spawning variants like NanoClaw (500 lines of TypeScript) for home automation.

Andrej Karpathy jumped in, buying a Mac Mini to test it and praising the orchestration and persistence—as @karpathy noted on X—but flagged serious RCE risks from broad agent permissions. Developers are buzzing about "AI house elves," but security experts warn of the dangers in viral, permission-heavy local agents.

Google's Gemini 3.1 Pro Surges 13 Points on Arena Leaderboard with Coding and Math Breakthroughs

Google's Gemini 3.1 Pro preview, out around February 19-20, rocketed to top-3 on LMSYS Arena with 1500+ Elo, up 13 points from before, dominating coding (+5 ranks), math (+4), and expert/IT categories.[10][11][12] It leads MathArena in most contests and shines in complex text/vision reasoning.

The Arena team highlighted its cross-category gains on X, with users dubbing the coding/math leaps a "game-changer." This keeps Google neck-and-neck in the model wars, building on Gemini 3 Pro's momentum.