SpaceX S-1: xAI had a $6.4B operating loss on $3.2B in revenue in 2025; Grok and X had 550M MAUs combined as of March 2026, and 117M used Grok's AI features (Rebecca Bellan/TechCrunch)
THE SO WHAT
A $6.4B operating loss on $3.2B in revenue with 117M Grok users shows the new AI playbook — burn heavily to wire an assistant into an existing distribution surface, then figure out monetization. If you don’t already control a large daily-use channel, your AI economics will look very different.
READ THE SOURCE
MORE FROM THE WIRE
Applied AITrump Set to Sign AI Cybersecurity Directive as Soon as Thursday
AI security is about to get federalized — an executive order on AI cybersecurity means your model, data, and agent stacks will be judged against government-grade expectations, not just SOC 2 checklists. If you’re shipping AI into regulated sectors, assume red-team, logging, and provenance requirements harden from “nice to have” to procurement gate.
Applied AISpaceX Listed Grok's ‘Spicy’ Mode as a Risk in Its IPO Filing
NSFW and exploitative outputs are now line items in S-1 risk sections — not hypothetical PR issues. If your product can generate sexualized or nonconsensual imagery, you’re underwriting legal reserves and brand risk whether you acknowledge it or not.
Applied AIJensen Huang says he’s found a ‘brand new’ $200B market for Nvidia
Calling AI-agent CPUs a $200B market is Nvidia telling you the stack is re-bundling — inference isn’t just GPUs, it’s heterogeneous systems tuned for agents. If you’re designing infra or software, assume CPU/GPU co-design and agent-specific hardware features become table stakes over the next hardware cycle.
Applied AISources: the Pentagon is launching a task force to study how to safely deploy leading AI tools with hacking capabilities across Cyber Command and NSA missions (Politico)
Offensive cyber is moving to AI-native — once Cyber Command and NSA normalize AI-assisted hacking, the baseline threat model for every network changes. Treat LLMs and agentic tools as both a defensive requirement and an assumed attacker capability, not a future scenario.