A-Portfolio

The company in one breath

Space Exploration Technologies, ticker SPCX on the Nasdaq, sells rocket launches, Starlink internet, Grok artificial intelligence, computing capacity, and advertising on X.

That sentence sounds like somebody merged an aerospace contractor, a phone company, an AI lab, and a social network after losing a bet. That is basically what happened. SpaceX bought xAI in February, and xAI already owned X. Public buyers now get three businesses under one roof: rockets, global connectivity, and AI.

The crown jewel today is Starlink. It beams internet from low orbit to homes, ships, aircraft, businesses, and governments. The rocket operation makes the satellites and launches them. The AI operation sells Grok and computing capacity while consuming money like a bonfire with a corporate card.

Why now

The stock closed Thursday at $134.00 and the latest overnight trade I can verify is $134.95 at 3:59:44 a.m. Eastern on August 21, from Alpaca's broker feed. SpaceX sold its June IPO at $135.00, so the public stock has completed a ten-week, multi-trillion-dollar round trip and landed five cents below the launchpad.

The timing matters for two reasons.

First, SpaceX just reported its first full public quarter. Second-quarter revenue reached $7.814 billion, up 91.9% from a year earlier. That means sales almost doubled in twelve months. For a company already producing billions each quarter, anything above 30% would look strong. Nearly 92% is freakish.

Second, another roughly 319 million shares became eligible for sale Thursday when a post-IPO lockup expired. A lockup blocks insiders and early holders from selling for a set period. The expiration does not force them to sell, but it opens another lane onto the highway. More unlocks arrive September 9, September 24, October 9, October 24, and finally December 8.

That supply pressure knocked the stock down 3.9% Thursday, from $139.65 to $134.00. I like the company more than I like the stock. That distinction pays the bar tab.

The numbers, translated

$7.814 billion of second-quarter revenue, up 91.9%. Revenue is the money customers paid before expenses. It nearly doubled. Good for a business this large would be 25% to 30% growth, so SpaceX cleared that bar with room for a booster.

$4.291 billion of connectivity revenue, up 65.8%. Connectivity is mostly Starlink service and hardware. It supplied 54.9% of total sales, which means the recurring internet business now carries more weight than rocket launches. Good would be above 30% growth with customer economics holding steady.

12.0 million Starlink subscriptions, up from 6.0 million. SpaceX doubled the number of service lines in one year. That is category-leader growth. Good would be at least 40% growth without giving away the service.

$66 of monthly Starlink revenue per subscription, down from $85. The company calls this ARPU, average revenue per user. SpaceX gained customers fast but collected 22.4% less per line. Good would be stable or rising ARPU while subscriptions grow. Doubling customers is beautiful. Discounting the drinks until the bar fills is less beautiful.

$2.561 billion of AI revenue, up 247.5%. Grok, X advertising, AI services, and computing infrastructure more than tripled their combined sales. Good would be growth above 50% with losses shrinking. Sales passed. The second test did not.

A $143 million operating loss, equal to 1.8% of revenue. Operating profit measures what the business keeps after normal operating costs but before interest and taxes. SpaceX lost about two cents for every sales dollar, much better than the $970 million loss a year earlier. Good would be a positive margin above 10% because this monster needs to fund expensive hardware.

A $541 million net loss, equal to 6.9% of revenue. Net income is the bottom line after interest and other costs. The loss improved from $1.008 billion a year earlier. Good would be repeatable profit, not merely a smaller crater.

$3.466 billion of operating cash in the first half, against $28.476 billion of equipment spending. Operating cash is money the business produced before buying rockets, satellites, chips, and data centers. Capital spending is what it paid for those long-lived assets. The gap was negative $25.010 billion. Good would be positive free cash flow, the cash left after both numbers meet.

$93.522 billion of cash and $39.364 billion of debt and finance leases. The IPO filled the tank, so near-term survival is not the issue. Good would be using that cash to create profitable capacity before another giant financing becomes necessary.

$47.461 billion of backlog. Backlog is contracted work the company still owes customers. It equals about one and a half years of second-quarter sales at the current pace. Good would be backlog growing while work turns into revenue and cash on schedule.

About $1.78 trillion of stock market value at $134.95. Multiply the price by roughly 13.18 billion shares, and that is what buyers value the equity at. Annualizing the second quarter gives about $31.3 billion of sales, so the stock costs roughly 56.9 times sales. Good for even a world-class, hardware-heavy grower would be closer to 15 to 25 times sales. At 57, the future already has a penthouse and valet parking.

What I like

The moat is physical. SpaceX designs rockets, builds engines, runs launch sites, makes satellites, operates a global network, and increasingly uses its own launches for its own businesses. A competitor cannot download that stack over a weekend. Rocket Lab (RKLB) and Amazon (AMZN) can attack pieces. Nobody currently matches the whole machine.

Starlink is becoming the economic center. Subscriber lines doubled. Consumer connectivity revenue grew 44.4%, and enterprise and government connectivity revenue grew 108.3%. Selling to households is good. Selling to airlines, navies, telecom carriers, and governments usually brings longer contracts and harder switching costs.

The operating trend bent the right way. Revenue nearly doubled while the operating loss fell from $970 million to $143 million. That is operating leverage, sales growing faster than the costs required to run the business. Good operating leverage eventually turns growth into profit.

The catalyst list is absurd. Starship reuse, direct-to-device service, Indian expansion, more government work, AI infrastructure, and a federal goal of more than 1,000 annual launches and reentries by 2030 can all expand the market. Polymarket (private) traders put a 97% chance on SpaceX catching a Starship flight with its launch-tower arms, but timelines remain the dangerous part.

What worries me

The valuation asks for domination across several industries at once. At roughly 57 times annualized sales, being excellent is not enough. SpaceX must remain the launch leader, win global satellite internet, make Starship reusable, and turn AI infrastructure into profit. One miracle is hard. The stock wants the combo platter.

AI swallowed the cash story. First-half capital spending reached $28.476 billion, four times the prior year's $6.965 billion. Research and development doubled to $7.062 billion. Much of the increase came from AI data centers and Starship. These investments may build enormous value. They may also build very expensive rooms full of equipment that ages faster than milk.

The clean businesses subsidize the speculative one. Connectivity grew 65.8%. Space revenue grew 29.0%. AI revenue grew 247.5%, but the AI operation still carries heavy losses and spending. I do not want Starlink's cash machine chained to every ambitious project management can name.

Control belongs to Elon Musk. The IPO filing said he held about 82.4% of voting power after the offering. Public shareholders own economic exposure with almost no steering power. Musk built the machine. He can also merge assets, set priorities, and move capital with limited resistance.

Price discovery is unfinished. The stock traded near $225 after the IPO and near $105 in August. Another five lockup tranches can add sellers through December. A famous company does not repeal supply and demand. It just gets better T-shirts.

What would make this a buy

I need proof and price, not another Mars speech.

The proof is three consecutive quarters with revenue growth above 40%, positive operating income, and annualized capital spending falling below annualized revenue. That would show the growth can fund more of the machine instead of leaning on the IPO cash pile.

I also want Starlink subscriptions above 15 million with monthly revenue per line stabilizing above $65. Subscriber growth without stable revenue per line can hide discounting and weaker customer mix.

The price zone that makes me do serious work is $75 to $90, not because the business is weak, but because the present valuation leaves no room for ordinary failure. At $90, the same share count implies about $1.19 trillion of equity value and roughly 38 times annualized second-quarter sales. That would still be expensive. It would merely stop being performance art.

This is research, not a buy signal. If the proof arrives first, my acceptable price rises. If the price falls first because the business cracks, I do nothing.

What makes me walk away

I walk if Starlink subscriptions grow while monthly revenue per line falls below $55 for two quarters. That would say growth is getting cheaper in the bad way.

I walk if annual capital spending stays above $50 billion while operating cash remains below $10 billion. The IPO bought time, not immunity.

I walk if Connectivity operating profit falls while AI spending rises. Starlink must not become the family ATM.

I walk if Starship reuse slips deep into 2027 without a successful upper-stage catch and reflight. Musk has moved this deadline before. Being early is being wrong when each delay burns billions.

I walk if management completes another major related-party merger without clear independent review and public economics. Public shareholders already handed over the wheel. They should not also close their eyes.

Earnings On Deck

QMLS, August 25 after the close. QumulusAI (QMLS) needs to turn contracted computing demand into revenue without wrecking cash. A good report shows higher utilization, tighter losses, and funded expansion. A bad report shows idle capacity and another financing need.

STDN, August 26 after the close. Standard Nuclear (STDN) needs to show that enriched-uranium plans are becoming customer commitments and milestones, not nuclear fan fiction. A good report adds firm backlog and protects its cash schedule. A bad report pushes production dates while costs climb.

The verdict

Score: 69 out of 100. Watch.

SpaceX earned every available point for leadership. Nobody else owns the same combination of launch cadence, reusable hardware, satellite scale, and distribution.

The near-doubling of revenue and Starlink's subscriber growth earned the full growth score. Backlog and fresh federal support added most of the catalyst points.

The company lost points because it still reports losses, spent $25.010 billion more on equipment than operations produced in cash during six months, and tied healthy assets to an AI buildout with uncertain returns. Musk's execution record earned credit. His voting control and related-party structure took some back.

The valuation lost almost everything available. A roughly $1.78 trillion company at about 57 times annualized sales offers no obvious cushion. The stock sits near its IPO price, but five more lockup dates keep entry quality weak.

This lands in watch, not candidate. The company may be the best industrial platform of this generation. The stock can still be a lousy transaction.

The DBTech read

A lockup is a temporary rule that stops insiders and early investors from selling after an IPO. When it expires, those holders gain permission to sell, but nobody forces them. The important number is the newly eligible share count compared with the shares already trading freely, traders call that the float. A large unlock can pressure price even when the business stays strong because supply suddenly grows. Treat the calendar like weather, prepare for it, but do not pretend it guarantees rain.

Sources and as-of date

Research completed August 21, 2026 at 7:10 a.m. Eastern.

• Price: $134.95, latest verified overnight trade at 3:59:44 a.m. Eastern on August 21, 2026. Source: Alpaca (private) broker feed, Blue Ocean overnight venue. Thursday's $134.00 close and daily volume came from Alpaca's SIP historical bars.
• Share count, market value, and sales multiple: DBTech45 calculations using the $134.95 price, 7,696,293,669 Class A shares and 5,485,486,276 Class B shares outstanding on July 28, and second-quarter revenue.
• Revenue, segment sales, losses, Starlink subscriptions and monthly revenue per line, backlog, cash, debt, operating cash, capital spending, launches, and research expense: SpaceX Form 10-Q for the quarter ended June 30, 2026, filed August 4, 2026, SEC accession 0001628280-26-052535.
• IPO price, offering terms, intended use of proceeds, voting control, and lockup framework: SpaceX final prospectus filed June 12, 2026, SEC accession 0001628280-26-042639.
• Lockup schedule and Thursday price move: Benzinga (private) coverage dated August 20, 2026, checked against the prospectus and Alpaca price history.
• Launch policy, India filing, Starship timing, and competitive context: White House national space transportation policy and company statements reported August 20 and 21, 2026.
• IPO calendar: Nasdaq IPO calendar checked August 21, 2026. It showed no scheduled upcoming deal with both an expected date and price range, so IPO Watch is omitted.
• Earnings dates: Nasdaq earnings calendar checked August 21, 2026.
• Recent public discussion: Reddit, YouTube, Hacker News, GitHub, and Polymarket scan covering July 22 through August 21, 2026. Social discussion informed risk questions, not the score.