SpaceX (SPCX) reported Q2 earnings last night, making a dramatic statement around exit revenue of $100B as well as an accelerated ramp of Nvidia compute online through 2027 relative to expectations. Thematically, this is a directionally positive development and shows SPCX’s ability to leverage their access to chips and power to potentially capture 1 year payback opportunities in the compute market. In addition, the continued progress on Starship 3 and what appears to be an accelerating roadmap to ODC’s is positive.
A decidedly bullish lift to exit revenue guidance is going to be met by number crunching around Capex and FCF, which I attempt to do a little thinking around below. My key takeaways are that 2H cash burn is coming down and the anticipated ramp in 2027 capex will be met by a larger revenue base. As an extrapolation of what seems possible, SPCX may be trading <8x revenue in 2028 as the rapid increase in their compute capacity and revenue growth put a $200B revenue year in the realm of possibility. Skeptics will clearly note that many things need to go right.
Liquid cooling and AI infrastructure (+): SPCX indicated they are building out north of 15GW of power capacity through 2027, including cooling/shell equipment to be prepared for additional compute.
Neocloud stocks could see some pressure from their rapid rebound in the past week depending on how the market reads what was a surprisingly large compute capacity ramp announcement last night.
As for Nvidia – it seems like this could be an interesting debate – a clear signal of confidence and capacity with one of the larger compute scalers but also unclear how this signals any broadening vs concentrating of the demand for their chips in the future.
Details from the call and considerations:
“Exclusively Nvidia” was a prolific statement. In a market where compute is scarce and prices continue to rise, SPCX indicated a commitment to rapidly accelerate their installed compute within 12 months by 4x (2GW to 8+GW) with the latest Nvidia compute. This appears to be one of the more bullish outcomes fundamentally from the earnings report along with the upward revenue guidance.
2026 exit revenue was guided to $100B relative to Q4 consensus of $15.8B, which implies a consensus exit run rate just over $65B – this is a 50+% increase relative to market expectations and seemingly driven by compute deals and SpaceXAI revenue post the close of Cursor.
Capex and FCF are interesting dynamics and appear better than knee-jerk reactions would conclude post call. In 2026, FCF has thus far been ~-$25B. Given the guidance for consistent levels of capex through year-end of ~$18.5B per quarter, combined with the ramp in revenue from compute contracts, one could estimate that negative FCF for 2H will be <$5B, a significant improvement.
In 2027, the capex consensus has been $89B, but that is likely to move significantly higher given SPCX’s aspiration to move from 2GW to north of 8GW of IT capacity by the end of 2027. Arguably, this is where the market will have some consternation; however, considering the FCF profile exiting 2026 and continued growth acceleration, it would not be inconceivable that SPCX could generate north of $100–125B of operating cash flow in 2027, thus supporting a pretty extensive ramp in capex.
Could revenue exit in 2027 could approach $200B (consensus $143B)? Using simple compute market math of $20M/MW per year for rental compute (a conservative, tight-market number), if SPCX rents out an additional 4GW of their 6GW added capacity next year at that rate, this is over $80B in annualized revenue, which, combined with an exit of $100B in 2026 and growth in other segments, makes it possible to visualize $200+B in 2028 revenue.
Orbital data centers are coming quickly and this is a theme to watch closely. An Nvidia chip-based reference design is planned to be tested in 2027 – this is moving quickly and I would expect more detail around this in the coming quarters.
Starship 3 commercialization will become a more consistent routine, and the next round of tests will both launch new Starlink V3 satellites into operation at the same time as testing advanced recapture and reusability of boosters. The company focused on the incredible order-of-magnitude increase in payload delivery to space once this is commercialized (thousands of tons/year to millions of tons).
Please let reach out if you have any questions or would like to discuss in more detail.
