Reid Hoffman says SpaceX 'not an AI company', xAI 'complete train wreck'

Original Article ↗ Hacker News Discussion ↗

A prominent tech investor’s claim that SpaceX is “not an AI company” and that xAI is a “train wreck” has triggered wider scrutiny of SpaceX’s post-IPO narrative, which leans heavily on AI revenues, GPU data centers, and sky‑high total addressable market figures. Commenters argue that rocket launches and Starlink can’t justify the valuation and see the AI story as bubble‑era financial engineering, while others counter that SpaceX’s launch record and infrastructure make it uniquely valuable even if the stock is overpriced. The critic’s deep financial ties to rival AI labs, along with broader unease about AI hype, generational attitudes toward AI, and billionaire score‑settling, lead many to question his motives even as they agree xAI looks strategically and commercially weak.

SpaceX as an “AI company”

  • Several commenters say SpaceX’s IPO materials present it primarily as an AI company with rockets/Starlink as side businesses.
  • Others argue the real substance is rockets and satellite internet, and that “not an AI company” is not an insult.
  • Some think tying a fundamentally space/communications business to the AI hype cycle is dangerous and could harm an otherwise solid but less spectacular company.

Financials, TAM, and valuation

  • Cited figures from the S‑1: 2025 revenue ~$18.7B total, with ~$11.3–11.4B from Starlink, ~$4B from launch, and ~$3.2B from “AI” (mostly the social network).
  • There’s heavy skepticism about internal accounting (SpaceX paying itself via Starlink launches) and whether that revenue is economically meaningful or just recycling losses.
  • The IPO pitch reportedly claims a $28.5T AI TAM and $320B AI revenue by 2030 with very high margins; many call this absurd relative to global GDP and food spending.
  • Some think current valuation is pure bubble behavior—investors buying because they expect to sell to someone even more optimistic later.

xAI’s competence and products

  • Many describe xAI as a “train wreck,” “failure,” or irrelevant, arguing its main “AI” value is renting GPUs to stronger labs.
  • Others counter that Grok 4.3 and Composer are competitive for coding and conversation, with Cursor (post‑acquisition) having real developer traction.
  • Still others say that, in practice, few professionals actually use Grok compared to leading models.

Datacenters vs. AI

  • Debate over whether renting GPU clusters makes a firm an AI company or just a datacenter operator.
  • Some argue that if you have large idle compute to rent out, you’re probably not a top‑tier AI lab.
  • Others note that, like cloud providers, it’s possible to be both infrastructure and AI, though the market may not justify AI‑style multiples for commoditizing compute.

Credibility and bias of critics

  • Many highlight that prominent critics in the article have substantial financial stakes in competing AI labs and in a major cloud provider.
  • Some see their comments as predictable “trash talk” with low information content; others argue that board‑level investors do have meaningful insight, but their incentives must be heavily discounted.

Gen Z and AI “generation” framing

  • The quoted idea that Gen Z should embrace being “generation AI” is widely criticized as patronizing and self‑serving.
  • Several commenters say young people have valid reasons to “boo” AI: job erosion, hype, circular investment, and looming crash risk.