xAI’s $6.4 B Loss and SpaceX’s IPO Reveal Massive Future AI Spend
Elon Musk's AI venture xAI recorded a $6.4 billion operating loss on $3.2 billion of revenue in 2025, according to SpaceX’s recent IPO filing. The same filing details an aggressive roadmap to scale the Grok model to “multiple trillions of parameters,” signaling that the current spending trajectory is far from over.
Scale‑Up Plans for Grok Signal Massive Compute Investment
The filing reveals that SpaceX intends to push Grok’s architecture to a size measured in multiple trillions of parameters, a step the company describes as a “step change in reasoning in depth and overall intelligence.” This ambition will require a substantial expansion of compute infrastructure.
Financial Snapshot: Revenues, Losses, and Capital Expenditure Trends
- 2024: $1.56 billion loss on $2.62 billion revenue.
- 2025: $6.4 billion loss on $3.2 billion revenue.
- AI‑related revenue grew to $465 million, split into $365 million from X and Grok subscriptions and $88 million from data licensing.
- Advertising contributed an additional $116 million.
- Capital expenditures rose from $12.7 billion in 2025 to an annualized run rate of $30.8 billion in Q1 2026.
- Monthly active users for Grok AI features reached 117 million in March 2026, out of 550 million total MAUs across Grok and X.
Strategic Implications for the AI Industry and Investor Sentiment
The disclosed losses and soaring capex underscore the high‑cost nature of frontier AI development. While competitors such as OpenAI and Anthropic are eyeing public listings in 2026, SpaceX’s anticipated valuation of up to $1.75 trillion positions the combined entity as one of the largest tech IPOs ever. The vertical integration of compute—via the Colossus and Colossus II data centers delivering roughly 1 GW of power—aims to lower training costs, but the scale of spending may test investor tolerance.
Outlook: Orbital Compute Satellites and Valuation Targets
The filing’s “use of proceeds” section earmarks expansion of AI compute infrastructure, including a long‑term plan to deploy orbital AI compute satellites as early as 2028. Although the satellite strategy is unlikely to materialize in the near term, it signals Musk’s intent to control the physical AI stack, a factor that could reshape cost dynamics if realized.