A sovereign social network
powered by Bitcoin

Grok bot summarized the spcx report. The line that jumps out at me....a planned Enterprise version of Grok Bot. Grok bot is the best user level AI available and no slouch on harder tasks. The user level part is key because Elon is going to hit that sweet spot that leapfrogs Grok ahead of everyone. Grokbot is already leagues better than Hermes etc. It's also better...oddly enough, than running Grok locally. Grok Bot has the special sauce. Don't sleep on it. I recommend everyone get Grok Bot and some SPCX / TSLA.

SpaceX had a big beat, and the forward guidance was the real shock. Revenue was $14.58B, up 161% on last year, against $12.72B expected. Standard EPS was $0.35 against $0.13 expected. Guidance for Q1 2027 revenue is $33–37B, while analysts expected $22.2B. After hours the stock is at $159.70, up 8.1% on the day.

Where the money came from (Q3)

• AI segment (called SI): $7.56B, up 830%. It's mostly renting computing capacity to outside customers like Google and Anthropic, plus $608M from Cursor. It made $431M of operating profit, or about $1.6B before $1.2B of Cursor stock pay.
• Starlink: $5.25B, up 69%, and still the profit engine with $2.37B of operating income. It has 13.6M consumer subscribers, and enterprise and government revenue grew 91%.
• Launch: $1.77B, up 6%, and roughly breakeven.
• Margins: gross margin was about 62% (Q2 was about 54%), and adjusted EBITDA margin was 60%. Management says Q4 gross margin will drop back toward Q2's level.

Future revenue and timelines

• Backlog: $63B, with about $42B expected to become revenue in the next 12 months.
• Revenue target: $100B annualized by December 2026, repeated on the call. Musk's $500B–$1T run rate by end of 2027 is an ambition, not guidance.
• Computing power: 1.7 GW now, rising to 2.5–3 GW by year-end. An enterprise rollout of Grok Bot is due within weeks.
• Starship: Flight 15 is next month and will try to catch both stages, with weekly flights targeted for the first half of 2027.
• Starlink phone service: new satellites start in 2027, and the spectrum deal should close in the first half of 2027.
• Longer term: chip production at the Terafab factory in Q3 2028, and the first launch from the Louisiana site in 2029.
• Tesla merger: independent committees at both companies are studying it. Musk called it a "good idea" but "not a certainty."

Valuation

• Size: about $2.17T after hours.
• Price to sales: about 33x enterprise value to Q3 sales annualized. That falls to about 19.5x on the $100B run rate and about 14x on the Q1 guidance.
• Peers: CoreWeave about 10.5x, Nvidia 17x, Tesla 14x.

My view is that it's expensive today but fair if they deliver Q1. At that point it trades like Nvidia or Tesla while growing over 100% with 60% EBITDA margins. The price leaves no room for misses, though.

Risks

• Cash burn: free cash flow was about −$35B over nine months, and Q4 capex is guided to $35–40B. Two analysts expect a capital raise soon.
• Earnings quality: about $2.3B of net income came from other income, interest income and a tax benefit.
• Share supply: about 1.3B shares unlock Thursday and 328M more on Oct 24.
• Customer concentration: AI revenue depends on a few big customers.
• Starlink pricing: revenue per subscriber fell from $83 to $69.
• Related parties: about $12.9B of debt is owed to related parties, and Musk sits on both sides of the Tesla merger talks.

Analysts: Citi raised its target from $200 to $240, and the consensus target is about $228.