Introduction
On June 12, 2026, investors decide whether to pay $135 a share for SpaceX, a valuation near $1.75 trillion. Part of that price is AI, and a small part of the AI is Grok. This report breaks the valuation into its component pieces and tests the portion that depends directly on Grok.
Recon Analytics breaks Goldman's aggregated 2030 AI projection into its components and sizes the Grok-dependent piece at roughly 14% of projected AI revenue. The rest is mostly compute that SpaceX rents to other labs and operational AI embedded in its vehicles and network, neither of which depends on how good Grok is.
Grok appears across SpaceX, Tesla, and X, but mostly in bounded, lower-consequence roles, and it earns little today. Recon Analytics AI Pulse data show Grok's primary users as the most male-skewed audience of any major AI tool, with the lowest workplace adoption and the highest concern about bias. At the same time, Grok's accuracy ratings improved through spring 2026, and paying users report good value for the money.
The investor question is not whether Grok decides the entire SpaceX valuation. It does not. The question is how much of the AI premium rests on Grok, how much rests on compute and operational AI, and whether Grok's rebuild can produce a materially stronger model rather than simply a stronger narrative.
Table of Contents
- Executive summary 2
- Why Grok matters even though it is small 2
- How Grok is used across SpaceX, Tesla, and X 2
- Breaking the AI number apart 3
- The industry shift, training to inference 4
- What the AI Pulse data say about Grok 5
- Monetization: free, paid, and willingness to pay 8
- The likely path forward 8
- Risks specific to Grok 9
- Bottom line 9
- Data Provenance and Methodology 10