On the eve of the lock-up, it plunged 13.6%. Is SPCX expensive?

NEXT PICK · Market Insights

On the eve of the lock-up, it plunged 13.6%. Is SPCX expensive?

Wednesday, August 5, 2026

A company that just delivered a 92% year-on-year revenue growth saw its stock plunge 13.61% in the same week, closing at $108.27, with trading volume expanding to 1.8 times the three-month daily average—not a performance crash, but the market preparing for tomorrow.

On August 6, 911.5 million shares of SPCX held by employees and early investors will be unlocked, with a market capitalization of approximately $116 billion at current prices; At present, the freely tradable shares on the float number less than 280.1 million.

In other words, the number of shares that could flood the market tomorrow is more than three times the current circulating stock. As a result, the impressive figures in the financial report—$7.81 billion in revenue and nearly half of the loss—were temporarily put underground.

So here's the question

Is this drop a misjudgment caused by liquidity events, or is it the market's true pricing for a financial report that "doubled revenue but burned $18.37 billion in a single quarter"?

Latest price $108.27 ▼ -13.61%

Q2 revenue $7.81 billion ▲ Year-on-year +92%

Quarterly capital expenditure 18.37 billion USD ▲ This is more than six times year-on-year

A company that integrates rockets, broadband, and AI into the same financial statement

Founded in 2002 and headquartered in Starbase, Texas, SPCX will go public on Nasdaq on June 12, 2026, at $135 per share, raising about $75 billion, making it the largest IPO ever.

Today's revenue is built on three business segments. The Space division is responsible for designing, manufacturing, and launching reusable launch vehicles—Falcon 9, Falcon Heavy, and Starship—serving commercial satellite operators as well as Department of Defense and NASA.

The Connectivity division operates the Starlink low-earth orbit broadband constellation, covering consumer, enterprise, and government clients. This segment grew 66% year-on-year this quarter, contributing more than half of the company's total revenue, with over 12 million subscribers—doubling year-on-year and up 17% quarter-on-quarter.

The third part is the AI department, which packages the Grok large model, X real-time information platform, and AI computing infrastructure, forming a closed loop from data acquisition and model training to inference deployment.

This structure is almost unheard of in the US market

One company is simultaneously a launch service provider, telecom operator, and cloud computing power supplier, and these three factors leverage each other—their own rockets send their own satellites into space, and their satellites become their own AI computing nodes.

This is where irreplaceability also grows. Falcon 9 has completed hundreds of successful recoveries, with a single booster reused more than 20 times; Starship is the largest thrust carrier currently under human development; Constellation scale is something competitors spend tens of billions of dollars and five to seven years to catch up.

Rather than calling it an aerospace company, it's more accurate to say it's building a toll road from the ground to the track—and once the road is done, the first fleet to hit the road is its own.

When the P/E ratio fails, what do you use to price 1.4 trillion yuan?

If you open the SPCX valuation snapshot, you'll find that the wrenches in traditional toolboxes are basically unusable: Q2 loss per share is $0.09, with no P/E ratio to talk about; Free cash flow is negative, and common anchors like free cash flow yield and enterprise value multiples also fail to produce satisfactory figures.

The only thing left to calculate is revenue multiples. With quarterly revenue of $7.81 billion, it reached about $31.2 billion, corresponding to a market value of $1.43 trillion, with a price-to-sales ratio (P/S) of about 46 times. This figure isn't unreasonable for a company whose revenue is doubling, but it's definitely not cheap.

Scenario simulation can provide a sense of boundaries. Assuming revenue growth slows to 50% over the next 12 months, the market is only willing to give a 20x price-to-sales ratio, implied market value of about $936 billion, a 35% discount from the current price; If the growth rate stays at 80% and 25 times, the implied market value will be basically the same as today; If growth reaches 100%, a 35-fold increase, the implied market value would be about $2.18 trillion, up 50% from the current level.

It's also interesting to do the sum of segments from another angle

Taking the launch business based on peer multiples, converting Starlink assuming 12 million users and about $66 per month ARPU, and assuming the AI segment at a 20 to 30 times price-to-sales ratio based on revenue, the combined value of the three segments is roughly between $430 billion and $620 billion.

There is a huge gap between this and the actual market capitalization of $1.43 trillion. What is inside the gap? It's Starship's full commercialization, orbital hashrate, exclusive shares of Nvidia Rubin GPUs—all forward options that haven't yet generated revenue.

On the cash flow side, the numbers are finally no longer blurred. In the first half of 2026, the company's operating activities generated about $3.47 billion in cash, with capital expenditures reaching $28.48 billion during the same period, resulting in negative free cash flow depth. Looking at the second quarter alone, capital expenditure was $18.37 billion, of which $15.83 billion was allocated to AI infrastructure.

The good news is that there are plenty of ammunition

At the end of the quarter, cash and securities on hand approached $100 billion, a sharp jump from $24.7 billion at the end of Q1—the $75 billion raised from the IPO was the main source of this money. At the current pace of burning, the runway is measured by year rather than quarter.

The bad news is that the nature of this money means it can only be burned once. So the real question is: with $15.8 billion in AI investment per quarter, how long and at what scale of revenue will it take to prove itself?

Just one step away from breaking even, but the market dealt a slap in the face first

Putting SPCX into the company's lifecycle means it has already moved far beyond the concept and government validation stages: annualized revenue approaches $31.2 billion, Starlink users exceed tens of millions, and Department of Defense contracts like NSSL Phase 3 and Lane 2 continue to be implemented.

But it hasn't reached its breakthrough period yet. GAAP is still at a loss, with a net loss of $541 million this quarter; Free cash flow depth is negative; The AI sector is in a period of large-scale investment, with operating losses reaching $6.35 billion for the full year of 2025.

What truly deserves its "turning point eve" label is the profit margin slope: operating losses narrowed from $970 million in the same period last year to $143 million, and operating margins improved from about -23.8% to about -1.8%. One more step and breaking even.

However, the technical side sends the opposite signal. Since it has only been listed for two months, the 50-day and 200-day moving averages are currently the same number at $144.53, while the stock price is already 25% lower than that. Fundamentals are rising, prices are falling—when these two lines will realign remains the core suspense in the coming quarters.

One-on-one duels are tough for every opponent, but no one can play three matches at once

On the launch service side, the most direct competitors are ULA (a joint venture between Boeing and Lockheed Martin), Bezos's Blue Origin, and Rocket Lab. ULA's Vulcan Centaur has already made its maiden flight but hasn't yet achieved high-frequency multiplexing, with annual launches far below Falcon 9.

Blue Origin's New Glenn is still in the early stages of commercialization; Rocket Lab's Electron and Neutron focus on small to medium-sized payloads, with limited overlap with SpaceX's in the medium and heavy-duty market. In terms of cost and frequency, SPCX currently leads by a single step, not half a size.

On the satellite broadband side, competitors are Amazon's Project Kuiper, Eutelsat OneWeb, and Iridium. Kuiper plans to deploy about 3,200 satellites, but will not enter large-scale commercial service by mid-2026; OneWeb focuses on enterprise and government back-and-forth, with limited consumer penetration. Starlink has over 12 million subscriptions, leading its peers by an order of magnitude.

AI is actually its weakest battleground. Microsoft has OpenAI behind it, Alphabet has Gemini, Meta has Llama, and each has computing power budgets and engineering reserves that rival theirs.

But the differentiation is precisely hidden in the intersection

When it comes to bringing AI computing power into orbit, no other company can build rockets, satellites, or connect with Nvidia chips on its own. This moat is not particularly deep in any one dimension, but rather a four-layer layer—technological heritage, economies of scale, government trust, spectrum, and regulatory licensing.

The switching costs on the government side are especially high. Multi-year framework contracts like NSSL require suppliers to undergo rigorous security reviews and infrastructure certifications, while new entrants may take years to meet standards; Once Starlink is embedded in the communication architecture of enterprises or the military, the risk of migration disruption is almost unbearable.

The ceiling on three tables with a 3% penetration rate

Combine the addressable markets of three business segments

Global commercial launch services are expected to reach a scale of $100 billion to $150 billion by 2030; The satellite broadband and connectivity market is in the $200 to $300 billion range; The AI infrastructure and computing power market is expected to surpass $500 billion or even $1 trillion by 2030. Together, these three fall between $800 billion and $1.4 trillion.

But the service market is much more realistic

Launches are expected to be about $50 to $80 billion, Starlink about $80 billion to $120 billion, and AI computing power and platform services at the initial stage of about $50 billion to $100 billion. Based on $31.2 billion in annualized revenue, SPCX's current overall penetration rate is only about 3% to 4%.

Where will you be in three years? In a conservative scenario, launch business is expected to grow 10% to 15% year-on-year, Starlink users reach 25 million, AI revenue grows 80% to 100%, and total revenue could reach $60 billion to $80 billion; Neutral scenarios are about $80 billion to $110 billion; The optimistic scenario—Starship's full commercialization and orbital computing power beginning to contribute revenue—could reach $120 billion to $160 billion.

These figures assume that all three assumptions hold true

The return cycle for AI capital expenditure is two to three years, Starlink has an annual churn rate below 5%, and there are no major geopolitical or regulatory black swan events. If any one misses, the curve will noticeably flatten.

The cards held by the bulls are two facts already confirmed by the financial reports

To be bullish on SPCX, you don't need to believe any Mars stories; you only need to trust the two sets of numbers already printed on the financial reports and their slopes:

▲ Bull Case

The bear's hand is a threefold pressure falling simultaneously

And the reasons for being bearish don't need to deny any of the above; just point out: the good news is about three years from now, the bad news is about tomorrow morning:

▼ Bear Case

In the next 12 months, four points will change the narrative

In the coming period, SPCX's stock price will likely be event-driven rather than performance-driven; whoever pays first will gain pricing power for the next round:

•	August 6: 911.5 million shares unlocked took effect, causing a short-term liquidity shock and a negative direction
•	Starting in October: The $6.7 billion in signed cloud service revenue began to climb, providing the first revenue validation for the AI segment, with a positive direction
•	Early 2027: Starmind AI-1 prototype satellite orbital testing, Nvidia Vera Rubin NVL72 system runs in orbit for the first time, with a positive direction
•	End of 2027: Starmind satellites enter mass production, with Nvidia Rubin GPU supply shares fulfilled, indicating a positive direction
•	Time to be determined: Taiwan has relaxed restrictions on foreign telecom holdings, opening new markets for Starlink. The direction is positive but the scale is limited

Less than two months after listing, the fundamentals and stock price have taken two opposite paths

SPCX's trading history is so short it can be summed up in one sentence: priced at $135 on June 12, closed at $108.27 on August 5, a drop of about 20% in less than two months.

During the same period, however, fundamentals improved rapidly—revenue jumped from $4.07 billion to $7.81 billion, net loss was halved, and operating losses narrowed by more than 80%. This divergence is not mysterious.

First is the post-IPO valuation digestion: the $1.77 trillion pricing itself carries a very high expectation premium, and the subsequent pullback feels more like a mean correction than a fundamental deterioration.

Second, there is the mechanical suppression of the unban event

The market began pricing in a supply shock of 911.5 million shares three weeks in advance, and the 13.61% single-day drop on August 5 was a concentrated release of this expectation.

Finally, there is schedule anxiety caused by capital expenditures

A financial report with revenue far exceeding expectations saw the stock price fall about 8% in after-hours trading, and sellers' reactions were completely divided—some institutions raised their target prices, while others did so on the same day. 23 companies offered buy, 6 hold, and 2 selled, with the gap between the highest and lowest target prices exceeding tenfold.

When a company's seller consensus is so divided, it means its value depends almost entirely on your assumptions for the next three years, rather than your interpretation of the previous quarter.

⚠️ Risk Notice • The unlocking of 911.5 million shares will take effect tomorrow, with a scale of about $116 billion, and the free float may expand several times over • If the $18.37 billion capital expenditure for the quarter cannot be converted into revenue, the earnings schedule will be significantly delayed • The AI sector is expected to lose $6.35 billion in 2025 operating losses, with a payback cycle lacking hard data verification • Restricted supply of power, GPUs, and key components is a high-level risk the company reported in Q10 • If spectrum licensing and national licensing are blocked, Starlink's expansion in emerging markets will slow down

🟡 Neutral A fundamental turning point has appeared, but lock-up unlocking and capital expenditure have not yet been fully priced in.

💬 Discussion No need to rush ahead before the wave of restrictions passes. Its real-time buy and sell positions can be viewed via SPCX via the official account.

Data source

•	Source: NextPick real-time market snapshot + SpaceX Q2 2026 earnings report and conference call + SEC filing documents + mainstream financial media public reports, data as of the US East Coast close on August 5, 2026.