Contents
Key Summary
2026-09-01 Current view: Phase 1: SpaceX is going to IPO at a 1.78T market cap, and we think that market cap is clearly overvalued. But because of the way the IPO is structured, plus Musk's reputation, this already-overvalued 1.78T could get pushed even further into ridiculous territory. We think there is room to profit from a short in there. | Phase 2: The company's market cap is severely overvalued, so there is room to profit from a short, and it also helps hedge overall market and AI risk.
2026-10-05 Current decision: Increasing the short from -14.02% to -18%. Fairly confident. If it keeps rising for superficial reasons, we lean toward adding more.
| Date | Ticker | Action | Shares | Total | Execution Price | Weight Before | Weight After |
|---|---|---|---|---|---|---|---|
| 2026-10-05 | SPCX | Short | -372.00 | 62,754.88 | 168.70 | -13.93% | -17.86% |
2026-10-05 Current Position:
| Ticker | Total Shares | Average Cost | Current Price | Gain | Return | Exposure | Weight |
|---|---|---|---|---|---|---|---|
| SPCX | -1,692.00 | 178.21 | 168.42 | 16,560.37 | 5.49% | -284,966.6 | -17.86% |
2026-10-05 Trade Record:
| Date | Ticker | Action | Shares | Total | Execution Price | Weight Before | Weight After |
|---|---|---|---|---|---|---|---|
| 2026-06-15 | SPCX | Short | -729.00 | 135,941.11 | 186.48 | -10.03% | |
| 2026-06-16 | SPCX | Short | -167.00 | 35,102.67 | 210.20 | -11.30% | -13.88% |
| 2026-06-30 | SPCX | Short | -159.00 | 27,258.39 | 171.44 | -11.04% | -13.11% |
| 2026-07-07 | SPCX | Short | -265.00 | 40,469.96 | 152.72 | -11.19% | -14.01% |
| 2026-10-05 | SPCX | Short | -372.00 | 62,754.88 | 168.70 | -13.93% | -17.86% |
2026-09-01 Technology route: Rocket launch, satellite internet, frontier AI model development, space data centers.
- Rocket launch: One of a kind in the world. Genuinely impressive. Reading through the road Musk and SpaceX have walked and what they have pulled off gets us fired up and full of admiration. | Very high-end, very grand, very future. But the overall market is still small for now, though it will keep getting bigger in the foreseeable future.
- Satellite internet: Commercially, this is smart and beautiful. The rocket launch market is small? Fine, I will go make it big myself.
- Frontier AI model development: Twitter is a mess, xAI is a mess. Merging and acquiring the two amounts to using every SpaceX shareholder's interests to clean up Musk's own mess. The Cybertruck order given to Tesla is the same kind of thing. | The 60B for Cursor feels overpaid, but 1 - the purchase itself was made with overvalued SpaceX stock, and 2 - Cursor's AI user entry point really is what xAI needs most right now. | Grok is actually decent at this point.
- Space data centers: Analyzed separately below.
2026-09-01 Space data centers: Does Musk really believe space data centers can be commercialized at scale in the short-to-medium term? Three possibilities: 1 - We are wrong. 2 - He is wrong. 3 - He is not being sincere. We think the odds run 3>1>2. | Just like the second-generation Roadster back then—there was never a plan to seriously build it, maybe just a vague idea, but that did not stop it from being a nice setup that helped with the cash flow problem. | Same thing here: float a story, help push the valuation up, raise money, and go spend it on rocket launch and AI. They do not actually want to build space data centers. | If he really believes it and pours serious resources into space data centers, that would be an excellent long-term bear case.
- Advantages: 1 - Five to eight times the solar output for the same panel area. 2 - Skips the grid interconnection queue. 3 - Land cost. (A false premise. Suitable space—dawn-dusk sun-synchronous orbit—is itself a scarce resource. )
- Difficulties: 1 - Heat rejection: anyone who has played Oxygen Not Included and built in its space environment understands that dumping heat in a vacuum is a big problem. 2 - Operations and repair: launch Musk himself up there to fix it? 3 - Latency: our understanding here is limited, but we feel the interconnect will have a hard time reaching ground data center levels. 4 - Launch capacity: the cost has to be cheap enough, and the supply has to be plentiful enough. 5 - Hardware refresh: chips iterate fast. Long term we would estimate five years, after which even if they still work, the efficiency will have fallen behind.
2026-09-01 Danger signals: # Signals that should make us re-examine our view and the risks.
- Grok succeeds and takes meaningful market share.
- Starlink revenue keeps growing fast.
- Space data centers actually make progress, with evidence that supports their commercial viability.
Complete Notes
Created the Note
Trade Record:
| Date | Ticker | Action | Shares | Total | Execution Price | Weight Before | Weight After |
|---|---|---|---|---|---|---|---|
| 2026-06-15 | SPCX | Short | -729.00 | 135,941.11 | 186.48 | -10.03% | |
| 2026-06-16 | SPCX | Short | -167.00 | 35,102.67 | 210.20 | -11.30% | -13.88% |
| 2026-06-30 | SPCX | Short | -159.00 | 27,258.39 | 171.44 | -11.04% | -13.11% |
| 2026-07-07 | SPCX | Short | -265.00 | 40,469.96 | 152.72 | -11.19% | -14.01% |
Current Position:
| Ticker | Total Shares | Average Cost | Current Price | Gain | Return | Exposure | Weight |
|---|---|---|---|---|---|---|---|
| SPCX | -1,320.00 | 180.89 | 143.23 | 49,711.20 | 20.82% | -189,063.60 | -12.17% |
Company-Specific Analysis and Research
Technology route: Rocket launch, satellite internet, frontier AI model development, space data centers.
- Rocket launch: One of a kind in the world. Genuinely impressive. Reading through the road Musk and SpaceX have walked and what they have pulled off gets us fired up and full of admiration. | Very high-end, very grand, very future. But the overall market is still small for now, though it will keep getting bigger in the foreseeable future.
- Satellite internet: Commercially, this is smart and beautiful. The rocket launch market is small? Fine, I will go make it big myself.
- Frontier AI model development: Twitter is a mess, xAI is a mess. Merging and acquiring the two amounts to using every SpaceX shareholder's interests to clean up Musk's own mess. The Cybertruck order given to Tesla is the same kind of thing. | The 60B for Cursor feels overpaid, but 1 - the purchase itself was made with overvalued SpaceX stock, and 2 - Cursor's AI user entry point really is what xAI needs most right now. | Grok is actually decent at this point.
- Space data centers: Analyzed separately below.
Space data centers: Does Musk really believe space data centers can be commercialized at scale in the short-to-medium term? Three possibilities: 1 - We are wrong. 2 - He is wrong. 3 - He is not being sincere. We think the odds run 3>1>2. | Just like the second-generation Roadster back then—there was never a plan to seriously build it, maybe just a vague idea, but that did not stop it from being a nice setup that helped with the cash flow problem. | Same thing here: float a story, help push the valuation up, raise money, and go spend it on rocket launch and AI. They do not actually want to build space data centers. | If he really believes it and pours serious resources into space data centers, that would be an excellent long-term bear case.
- Advantages: 1 - Five to eight times the solar output for the same panel area. 2 - Skips the grid interconnection queue. 3 - Land cost. (A false premise. Suitable space—dawn-dusk sun-synchronous orbit—is itself a scarce resource. )
- Difficulties: 1 - Heat rejection: anyone who has played Oxygen Not Included and built in its space environment understands that dumping heat in a vacuum is a big problem. 2 - Operations and repair: launch Musk himself up there to fix it? 3 - Latency: our understanding here is limited, but we feel the interconnect will have a hard time reaching ground data center levels. 4 - Launch capacity: the cost has to be cheap enough, and the supply has to be plentiful enough. 5 - Hardware refresh: chips iterate fast. Long term we would estimate five years, after which even if they still work, the efficiency will have fallen behind.
Stock community:
Current view: Phase 1: SpaceX is going to IPO at a 1.78T market cap, and we think that market cap is clearly overvalued. But because of the way the IPO is structured, plus Musk's reputation, this already-overvalued 1.78T could get pushed even further into ridiculous territory. We think there is room to profit from a short in there. | Phase 2: The company's market cap is severely overvalued, so there is room to profit from a short, and it also helps hedge overall market and AI risk.
Working title: The space data center is a story told without sincerity
Danger signals: # Signals that should make us re-examine our view and the risks.
- Grok succeeds and takes meaningful market share.
- Starlink revenue keeps growing fast.
- Space data centers actually make progress, with evidence that supports their commercial viability.
Current decision: Hold the position at -12.17%. Fairly confident. If it keeps rising for superficial reasons, we lean toward adding more.
Considering Adding
Price rise: The stock has been running very well lately, up 55.26% from the earlier low of $108.27 to $168.
Current decision: Increasing the short from -14.02% to -18%. Fairly confident. If it keeps rising for superficial reasons, we lean toward adding more.
| Date | Ticker | Action | Shares | Total | Execution Price | Weight Before | Weight After |
|---|---|---|---|---|---|---|---|
| 2026-10-05 | SPCX | Short | -372.00 | 62,754.88 | 168.70 | -13.93% | -17.86% |
Current Position:
| Ticker | Total Shares | Average Cost | Current Price | Gain | Return | Exposure | Weight |
|---|---|---|---|---|---|---|---|
| SPCX | -1,692.00 | 178.21 | 168.42 | 16,560.37 | 5.49% | -284,966.6 | -17.86% |