Eliza Ward: Two weeks into SpaceX being a public company and I already feel like I need a spreadsheet just to have a conversation about it — how are you holding up?
Brian Reed: Let me see — yeah, same. The numbers keep getting bigger and the explanations keep getting shorter.
Eliza Ward: Okay, so — August 8th. SpaceX's first earnings call, ever, as a publicly traded company on the Nasdaq. Musk gets on and projects one trillion dollars in annual revenue by 2030. And I want to ask you to react to that, but actually the more I think about it — the number isn't even the most interesting part.
Eliza Ward: The venue. This wasn't a tweet. It wasn't an X post. An earnings call is a regulatory document — analysts, shareholders, the SEC, all simultaneously. Whatever he says there carries a materially different weight.
Brian Reed: So the question isn't really whether he believes it — it's what it means that he said it there, in that room, for the first time to a public shareholder base with no historical baseline for SpaceX.
Eliza Ward: Right — and the target itself moved. It was 2031 before this call. He pulled it to 2030 and mentioned a non-zero chance of 2029. That's the thing that made me actually stop.
Brian Reed: He accelerated it on the record. On day one.
Eliza Ward: And accelerating it on record is — because the math from that baseline is not aggressive. It's structurally implausible.
Brian Reed: So let me just put a picture on this. Imagine a lemonade stand made eighteen dollars last summer. Someone tells you it needs to make a thousand dollars by next Thursday. Except — yeah, Thursday is five years away. That's the gap. Eighteen point six seven billion in 2025 revenue, one trillion by 2030. The compound annual growth rate you need to get there is a hundred and twenty percent. Every single year. For five consecutive years.
Eliza Ward: That's a 54x increase.
Brian Reed: Fifty-four times. In five years. And — hang on, the part I don't get is who's actually checking this. Like, who has standing to push back on that number?
Eliza Ward: That's exactly what I want to know. And the answer is — the investment banks that underwrote the IPO. They sat in rooms with SpaceX's actual books. They saw the cost structure, the subscriber economics, the forward contracts. Their published forecasts are nowhere near a trillion.
Brian Reed: Right — and they're conflicted, because they were paid to underwrite the deal. So their silence on this, the gap between what they'll put in a research note versus what Musk said on that call — that silence is actually the loudest signal in the room.
Eliza Ward: Wait — so Steven Silver at Argus Research upgraded SpaceX to Buy on August 7th. A hundred and sixty dollar price target. That's the day before the earnings call.
Brian Reed: Before the call. So the AI investment thesis was already being priced on narrative before Musk said a word publicly. And then you've got Cathie Wood — ARK Invest bought over twenty million in SpaceX shares during a prior dip. ARK's own 2026 model has a two point five trillion valuation scenario for Starlink. But that's a model. It's not audited. It's a high-adoption assumption run through a spreadsheet.
Eliza Ward: So the only people who've seen the real numbers think the target is too high. And the believers are working off models, not books.
Brian Reed: But that's actually where the Q2 numbers get strange. Because on the surface — Starlink looks like the good news. Four point three billion in revenue, up 66% year-over-year. Subscriber base doubled to 12 million. It's the only profitable segment SpaceX has.
Eliza Ward: Hold on — doubled to 12 million and revenue per subscriber went down.
Brian Reed: By about twenty dollars. Year-over-year. So they added, I don't know, millions of customers, and somehow made less money per person. That doesn't add up.
Eliza Ward: It's the lower-tier plans. Growth is happening downmarket. Like — okay, picture this. A family in rural Montana signed up in 2023 on a higher-rate plan, maybe the only option at the time. Now their neighbors are joining on cheaper entry-level tiers. Same satellite overhead. Less revenue per household. Multiply that across six million new subscribers and that's the ARPU compression, just... playing out roof by roof.
Brian Reed: And 12 million subscribers is — wait, what percentage of actual internet users is that?
Eliza Ward: Roughly 0.2%. Of the world's six billion internet users.
Brian Reed: So Musk said on that same August 8th call that Starlink could eventually carry a majority of internet traffic in its markets. Against a 0.2% baseline. I mean — actually, no, that's not even the contradiction. The contradiction is that the only engine built to get to a trillion is simultaneously growing volume and losing pricing power at the same time.
Eliza Ward: That's — yeah. And we don't know if that reverses. It might. But at the current trajectory the math gets harder, not easier. And there's a piece of this we haven't touched yet — why that trillion-dollar number might matter less as a forecast and more as something else entirely. That part shifts how you read everything Musk said on that call.
Brian Reed: The only profitable segment is the one with the pressure on its margins. That's a real problem for the model.
Eliza Ward: Which makes the permission structure reading almost more unsettling. Because two days before that call — August 6th — the first post-IPO lockup expiration cleared. No crash. Insiders didn't dump. That was the big fear and it just... didn't happen.
Brian Reed: Right — and then forty-eight hours later Musk is on the earnings call telling a shareholder base that has literally never owned SpaceX before — no historical baseline, no prior quarters to anchor against — that the company will hit a trillion by 2030. Maybe 2029. I mean, the timing is — actually, that's not a coincidence. That's sequencing.
Eliza Ward: Used how, though? Like what does the number actually do?
Brian Reed: It tells the organization — and the market — that the spending coming is justified. Think about it: if you're a SpaceX engineer and Musk says we'll hit a trillion by 2030, that's not a forecast you're supposed to verify. That's a mandate. It determines what gets funded. And the moon manufacturing stuff — robots scaling factories, a mass accelerator, solar production — he literally called it 'super sci-fi right now.' On the record. And then said it was inevitable anyway.
Eliza Ward: Wait — he said 'super sci-fi' himself?
Brian Reed: His words. And yet it's sitting in the same earnings call as the CFO saying some AI compute investments have a less than one-year payback. Which is — okay, those are two very different confidence levels in the same document.
Eliza Ward: That CFO number is the one I can't get past. Less than a year payback on some AI compute deals — but there's no AI compute revenue line. It was never broken out. Never independently validated. It's a claim inside an earnings call with nothing underneath it.
Brian Reed: And the deadline migration makes it harder to treat any of it as data-driven. 2031 to 2030 to maybe 2029 — does each move reflect something real that changed operationally, or is the ceiling just rising to meet a newly public audience that doesn't know what the old ceiling was?
Eliza Ward: And the Argus upgrade the day before — Steven Silver sets a $160 target, Rocket Lab and Intuitive Machines both rally with SpaceX. That's the market pricing orbital infrastructure broadly. Not the trillion. Not 2029. Just — the category.
Brian Reed: The $100 billion milestone. That's the checkpoint, right? Management's own framing is that by end of 2026 SpaceX should be around $100 billion annualized. If that lands — actually, wait, even if it lands, that's still ten times from there in three and a half years. The test isn't whether they hit $100 billion. The test is whether Starlink's ARPU stops falling and whether an AI compute revenue line actually shows up on a filing, as a real number, before the capital spending gets so large it requires dilutive raises.
Eliza Ward: And the people positioned to flag it first — the underwriters, the same banks who already don't believe the trillion — they're the ones watching those two variables most closely. ARPU trend, AI revenue line. If neither moves by 2026, they'll say so. Quietly, in a revised model.
Brian Reed: We started this thing saying you needed a spreadsheet just to have the conversation. I'm not sure the spreadsheet actually helps.
Eliza Ward: No — it just makes the gaps more legible.
Brian Reed: Yeah. That's maybe the most honest place this landed.
