Originally published by Zach Jennings on LinkedIn, September 7, 2026. Expanded here with source references.
The biggest IPO in history filed its first quarterly report as a public company last month. Almost nobody has opened it.
I did. Here’s what’s in there.
Start with the segments
Three months ended June 30, 2026. Total revenue $7,814 million. Three reportable segments.
Space, which is Falcon, Dragon and Starship: $962 million. That’s 12% of the company.
Connectivity, which is Starlink: $4,291 million. 55%.
AI: $2,561 million. 33%.
The AI business is now nearly three times the size of the rocket business. If you still picture SpaceX as a launch company, the launch company is the smallest thing in the building.
And that segment barely existed six months ago.
SpaceX acquired xAI on February 2, 2026. Grok, the Colossus data centers, and X, all of it folded in. Because Musk controlled both sides, it’s a common control transaction, and the prospectus says the financials were retrospectively recast for all periods presented to include it.
So the entity trading under SPCX is SpaceX plus Starlink plus Grok plus X. Every historical comparison you’ve read treats that as one continuous company, because under the accounting rules it is.
Watch the AI line move. First quarter of 2026: $818 million. Second quarter: $2,561 million. It tripled in three months.
What the headlines got wrong
The stock has had a rough few months, and the story attached to that has mostly been about disappointment. The filing says something else.
Revenue for the quarter was $7,814 million against $4,071 million a year earlier. That’s 92% growth.
Loss from operations for the quarter: $(143) million. Same quarter last year: $(970) million.
For the half, $12,508 million of revenue against $8,138 million, and the operating loss for the six months is almost entirely the first quarter’s. The second quarter nearly broke even on that line.
Net loss is a different story. $(541) million for the quarter, $(4,817) million for the half. The operating line improved. The bottom line did not, and R&D nearly doubled to $3,548 million in the quarter, so this is not a cost-discipline story either.
I’m not going to tell you what that’s worth. I don’t do price targets and this company isn’t a client. But whatever you think of the valuation, “the numbers are deteriorating” is not what the 10-Q says.
The most interesting page in the document
Then you get to the cover page. As of July 28, 2026:
Class A common stock outstanding: 7,696,293,669 shares.
Class B common stock outstanding: 5,485,486,276 shares.
Class B is 41.6% of the shares. Each one carries ten votes.
Do the arithmetic and Class B controls 87.7% of the voting power. Class A, which is every share sold in that record-breaking offering and every share bought since, is 58.4% of the company and 12.3% of the vote.
That is on the cover of a public document. It takes about ninety seconds with a calculator.
Which is why it’s a controlled company
More than half the voting power sits with one group, so under Nasdaq’s rules SpaceX qualifies as a controlled company and can rely on exemptions from a set of governance requirements. It can skip a majority-independent board. It can skip an independent compensation committee. It can skip independent oversight of who gets nominated as a director.
The audit committee is the one thing it can’t skip, and only because that requirement comes from SEC Rule 10A-3 rather than from the exchange’s discretion. Everything the exchange had discretion over, the exchange gave away.
How it got here
SpaceX submitted its draft registration statement confidentially in the spring. Nobody outside the company and its banks could read a word of it until May 20.
Confidential submission came out of the JOBS Act in 2012, and it was built for emerging growth companies, issuers under roughly $1.2 billion of revenue, on the theory that a failed offering conducted in public is fatal when you’re small. In July 2017 the SEC opened it to everybody.
So the largest IPO in history used the on-ramp written for the smallest companies in the market, and then arrived and switched off every governance requirement it was permitted to switch off.
None of that is a criticism of SpaceX. They used rules that exist, and any competent banker would have told them to. It’s a comment on where accommodations end up.
The other rulebook
On July 22 the SEC approved a Nasdaq standard that delists a company whose market value of listed securities sits under $5 million for 30 consecutive business days. No cure period. A hearing request does not stay the suspension. Compare the bid price rule, where a company gets 180 calendar days and often another 180 before anyone issues a determination.
It was stayed on July 29, seven days later, after Cemtrex and the Small Public Company Coalition filed notices of intent to petition for review. It’s still stayed, with no deadline on the Commission.
The Commission’s own analysis, printed in its own approval order: 91 issuers would have failed that standard in 2025. And Craig Lewis, the SEC’s former Chief Economist, studied the 816 companies that spent 30 days under $5 million between 2006 and 2025. 78% climbed back above the line.
Two rulebooks on the same exchange
At the top, a company can use the small-issuer on-ramp to go public, sell 638.9 million shares at $135, raise $85.7 billion, close its first day up 19%, and operate without a majority-independent board.
At the bottom, a stretch of bad tape becomes a delisting with no cure period, and the exchange’s case for it is contradicted by the regulator’s own economist.
Nobody sat in a room and designed that. It’s what happens when every accommodation ends up available to whoever can afford to claim it, and every protection ends up written against whoever can’t afford to fight it.
The cover page of a 10-Q is free. So is the approval order. Most of what’s argued about on this platform is sitting in documents nobody opens.
Disclosure
Atlas Signal, Inc. publishes both paid sponsored reports and uncompensated reports on US-listed micro, small and mid-cap companies. This article is uncompensated commentary. SpaceX is not an Atlas Signal client, and Atlas Signal has received no compensation from any company mentioned in this article.
Atlas Signal, Inc. does not, and will not, own, hold, or take any position — long or short — in the securities of any company it features or covers.
This article is for general informational purposes only and is not individualized investment advice or an offer or solicitation to buy or sell securities. Atlas Signal is not registered as a broker-dealer or investment adviser and is not a member of FINRA or SIPC. Figures cited are drawn from public SEC filings and SEC releases as of the publication date and may change without notice. Readers should independently verify material information and consult their own professional advisers.