I don't think it's crazy to think $SPCX can do ~$300bn+ of Cloud revs in 28/29, which would make it the 4th biggest CSP and within shouting distance of the Big 3. Assume 8GW capacity is online in 28 (delay vs Elon targeting YE27). Leading-gen $NVDA systems (GB+VR) should monetize toward high-end of $30-50bn/GW range (inline w/ Anthropic/Google deals), especially as model capabilities continue to improve and system optimizations drive *much* higher tokens/node (supporting further rental rate upside). Assuming 80% is rented to third parties at 50-60% margin, Cloud can do $150-180bn of EBITDA (giving zero credit to 1P) and total company EBITDA can reach $180-230bn, vs street at $80/130bn in 28/29 and buyside even lower than that (which has limited confidence in any numbers at all). Would imply SPCX is trading 6.5-8x which seems cheap relative to the Cloud opportunity on its own, let alone Starship and Orbital, or even Grok 5+ w/ all of the SPCX engineering data which sounds unique. I have questions around financing the company will need to answer. Would expect a substantial portion of the GPU-system capex to be financed through vendor-, lease- or GPU-backed structures. But if you believe in the value of GB's/VR's, which are supported by increasing spot rental rates and Elon's comment around <1-year paybacks, then the financing should be very achievable without undermining the return profile. At a higher level, this was one of the more fascinating Elon calls I've covered and the budding @elonmusk + @JensenHuang partnership seems like it might have major implications for the entire industry. SPCX building exclusively on NVDA for the foreseeable is a big deal. And big credit to @FundaAI for nailing the 8GW number in their preview. Very impressive.