SpaceX发首份财报的前夜，马斯克焦虑到熬夜玩法环

Original Author: 加六 (Jia Liu)

The latest gossip circulating in Silicon Valley circles, besides Oracle founder Larry Ellison strolling through Newport Beach, Los Angeles with his sixth wife, a Chinese-American, is that former world's richest man Elon Musk has once again become engrossed in *Elden Ring* after SpaceX's stock price decline.

This is not the first time Musk has immersed himself in gaming during periods of high stress. In 2022, after deciding to acquire Twitter, he played *Elden Ring* until 5:30 AM in a Vancouver hotel room. This was recounted by Musk's ex-girlfriend and later included in Walter Isaacson's biography *Elon Musk*.

This time, Musk's anxiety stems from SpaceX's plunging stock price.

On its first trading day, June 12th, SpaceX closed at $161, reaching a market capitalization of $2.1 trillion, instantly ranking among the top six U.S. stocks. Four days later, on June 16th, it hit an intraday all-time high of $225.64, briefly surpassing Microsoft with a market cap of $2.94 trillion. Musk's personal net worth that day reached $1.45 trillion, marking the wealthiest moment in recorded human history.

Now, seven weeks later, the stock has fallen from its $225.64 peak to $114, nearly halved. The market capitalization has eroded by approximately $1.2 trillion, and Musk's personal paper wealth has decreased by over $750 billion.

People close to Musk say he has recently become extremely addicted to playing *Elden Ring*, even asking employees to play for him, and has been playing games during meetings. Cursing at employees during meetings has become commonplace. He often shows up late or doesn't appear for evening meetings at all, with 8 PM meetings sometimes being postponed to 2 AM while people still wait for him.

While we cannot confirm the veracity of these reports, this week, Musk's anxiety has indeed reached a peak because SpaceX's first earnings report on Tuesday coincides with the first major lock-up expiry on Thursday.

Who Bears the Burden of the $1.2 Trillion Wipeout?

On June 16th, just four days after its listing, SpaceX's market cap briefly touched $2.94 trillion.

But the good times didn't last. With low float and high valuation, SpaceX experienced a 51-day decline. As of today, its market cap has evaporated by roughly $1.2 trillion.

The person most affected is undoubtedly Musk himself. He holds approximately 4.8 billion SpaceX shares, along with 350 million options with a strike price of $8.40, and about 700 million Tesla shares.

On June 22nd, when SPCX dropped 16.4% in a single day, he lost $152 billion in one day. On July 1st, his net worth fell below $1 trillion, officially losing his trillionaire status. On July 16th, the Starship launch was aborted, costing him another $45 billion in a day. By July 27th, his net worth had dropped to $695.7 billion. Over five weeks, Musk's fortune evaporated by approximately $750 billion—a sum greater than the combined net worth of the world's second and third richest people.

And those who are most "unable to voice their grievances" are the millions of American retirement account holders who never placed a single buy order for SpaceX.

On July 7th, SpaceX was added to the Nasdaq-100 Index. Under the old rules, newly listed companies had to wait for a three-month "seasoning period" to qualify. Nasdaq specifically modified the rules for SpaceX: companies with a total market cap exceeding existing constituents can skip the waiting period. SpaceX was added to the index just 15 trading days after its IPO, the fastest inclusion in Nasdaq-100 history.

JPMorgan estimated that just the Invesco QQQ Trust, the largest ETF tracking the Nasdaq-100 with assets under management of about $480 billion, generated approximately $4.3 billion in passive buying demand. When all products tracking the Nasdaq-100 are combined—over 200 funds managing roughly $800 billion in assets—total passive flows are estimated between $22 billion and $27 billion. The vast majority of this money entered around the close on July 6th and the open on July 7th, when SPCX was priced between $157 and $161. By July 22nd, QQQ held 39.7 million shares of SPCX, valued at $4.57 billion, representing a 0.98% portfolio weight.

In other words, millions of 401(k) holders became SpaceX shareholders at around $160 per share. Their asset allocation now includes a stock that posted a net loss of $4.9 billion in 2025, trades at a price-to-sales ratio exceeding 115 times, and has been public for less than a month. Based on the closing price of $114.53 on August 3rd, the unrealized loss is approximately 28%.

And this buying decision wasn't even made by them.

What Exactly is SpaceX Today?

On February 2, 2026, SpaceX acquired Musk's AI company xAI in an all-stock deal, valuing xAI at $250 billion, with the combined entity valued at $1.25 trillion. In May, Musk announced that xAI would no longer exist as a standalone company. On July 6th, it was officially renamed SpaceXAI. The trademark filing describes: satellite data centers, orbital computing services, and AI software. Musk's rationale is that global AI electricity demand "simply cannot be met by terrestrial solutions," and moving data centers into space is "the only logical solution." SpaceX has applied to the U.S. Federal Communications Commission to deploy up to 1 million AI satellites.

Then there's Cursor. On June 16th—the day the stock peaked—SpaceX announced it would exercise previously obtained options to acquire Anysphere, the parent company of AI coding tool Cursor, in an all-stock deal valued at $60 billion. This is the largest acquisition of a venture-backed startup on record. The purpose is to integrate Cursor's programming data into the training pipeline of the Grok large language model, while also giving Cursor access to xAI's Colossus supercomputer cluster.

So, a rocket company is using stock gained from its recent IPO to buy an AI coding tool, in order to train a large language model, running on data centers it plans to launch into low-Earth orbit.

SpaceX will now need to finance not only rockets, satellites, and ground networks, but may also need to provide capital for Musk's broader technological ambitions within his ecosystem. The problem is, its balance sheet is not lightweight.

According to figures disclosed in the IPO prospectus: 2025 full-year revenue was $18.7 billion, with a net loss of about $4.9 billion. Q1 2026 revenue was $4.69 billion, with a net loss of approximately $4.3 billion and capital expenditures approaching $10 billion. AI-related spending as a share of capital expenditures was 61% in 2025, rising to approximately 76% by early 2026. Full-year 2025 capital expenditures were about $21 billion—spending more than it earned. Looking ahead, analyst consensus expects capital expenditures to reach $48.7 billion in 2026 and further surge to $118.4 billion in fiscal 2028, with debt projected to grow more than fivefold from $41.7 billion to $218 billion over the same period.

So what exactly is SpaceX today?

It could be argued that this is no longer a rocket and satellite internet company, but an AI infrastructure company. Musk is using Starlink's subscription cash flow and his own stock to pay for AI computing power bills.

And the situation inside the AI division is far uglier than the rumors suggest.

Bloomberg Businessweek published an investigative report on July 16th based on interviews with over a dozen insiders and internal documents. This spring, Michael Nichols, a systems-oriented operations executive who had long been responsible for Starlink engineering, was appointed President of xAI to take over the division. Musk gave him only one task: catch up to Anthropic's Claude. Every time Claude updates, Musk demands Grok keep pace. Multiple internal projects are directly named after Claude, and several Slack channel names bear the competitor's product name. Nichols' initial memo to the team stated: the near-term goal is to match Claude's performance and make Grok "maximally useful."

But what Nichols inherited was a wreck. Dozens of employees left after the merger, including multiple co-founders. In March, plans were made to cut up to 30% of staff, resulting in situations where people were let go without anyone informing them. Hiring was paralyzed by chaos, with candidates waiting indefinitely after interviews because HR was too understaffed to complete the processes. The pre-training team was at one point reduced to fewer than five people. The departure list also included the head of post-training and a co-founder from French AI company Mistral who had joined within the year. Some employees began to question Musk's judgment, believing that someone who built his career on cars and rockets is fundamentally an amateur when it comes to large language models.

Follow-up reporting from U.S. tech investigative outlet The Information proved even more decisive: all 11 original co-founders of xAI have now left the company. More than 50 researchers and engineers have moved to Meta and the emerging AI research company Thinking Machines Lab. After the mass exodus of co-founders, the company's structure has been in a state of near-constant flux for an extended period, with Musk at one point personally managing dozens of direct reports.

A company planning to spend $118.4 billion in capital expenditures by 2028 has had its core AI research team hollowed out, its CEO is directly managing a few dozen people, and that same CEO is simultaneously managing rockets, satellites, an EV company whose cash flow has just turned negative, and a social media platform.

First Earnings Report Collides with First Major Lock-up Expiry

On Tuesday, August 4th, after market close, SpaceX released its first quarterly earnings report since going public. This is the first time Wall Street will see the company's complete operating data.

Let's first look at institutional expectations for SpaceX this quarter.

Visible Alpha's consensus estimate, under S&P Global, is total revenue of $6.9 billion; Bloomberg's consensus is $6.81 billion; Zacks' consensus is $6.72 billion; Koyfin data shows $6.82 billion. Roughly landing around $6.8 billion, implying about 15% year-over-year growth and a significant sequential jump from Q1's $4.69 billion. That sounds decent, but the market isn't waiting for growth itself—it's waiting to see if growth can accelerate.

On losses, expectations range from $0.22 to $0.26 loss per share. Koyfin expects adjusted EBITDA of approximately $2.05 billion. Full-year 2026 consensus is a loss of $0.64 per share, with profitability only expected to turn positive in 2027 at $0.63 earnings per share.

More important than the aggregate numbers is the segment breakdown. SpaceX currently has three major business segments: Starlink satellite internet, Falcon and Starship launch services, and AI computing with the Grok large language model. The market expects Starlink's operating margin to reach 35.9% this quarter, with its profits offsetting operating losses from the launch and AI segments. Analysts also expect Starlink revenue to grow over 50% year-over-year to $4.7 billion in Q3.

In other words, a company with a $1.4 trillion market cap has its entire valuation story resting solely on the Starlink business.

But Starlink's own story has cracks too. As of the end of Q1, Starlink had approximately 10.3 million subscribers across 164 countries, doubling year-over-year—very fast growth. The problem is that average revenue per user per month has dropped from $86 a year ago to $66. Management says this number will continue to decline as Starlink expands into overseas and lower-priced markets.

Ahead of the first earnings report, brokerage views on SpaceX have diverged to an almost absurd degree. After a company goes public, underwriters observe a quiet period (this time until July 7th), after which analysts can officially issue coverage. Price targets are based on analysts' own valuation models projecting stock prices 12 months out—not predictions of earnings, but judgments on long-term value. On July 7th, over a dozen banks simultaneously published their initiation reports. When the results came out, the market discovered just how ridiculous the divergence was.

U.S. investment bank Raymond James' analyst Brian Gesuale gave the highest target of $800, calling SpaceX "one of the most defining industrial infrastructure companies of the 21st century," with the core logic being that Starship can drive launch costs to extremely low levels. Morgan Stanley's lead analyst Adam Jonas gave $300, based on a long-term outlook of $3.3 trillion in revenue by 2040. Goldman Sachs analyst Eric Sheridan gave $205. In between were JPMorgan at $225, Bank of America at $235, Wells Fargo at $230, UBS at $210, Citi at $200, Macquarie at $250, and RBC at $225.

On the bearish end: HSBC initiated coverage with a "Hold" rating and a $115 price target, below the $135 IPO price. CFRA, S&P's independent research arm, issued a "Sell" rating at $115, citing "an extremely aggressive growth strategy, excessively high valuation expectations, and significant capital intensity." Morningstar, one of the world's largest independent investment research firms, gave an even lower fair value estimate of just $62—less than half the IPO price.

A total of 23 banks participated in the offering, and 18 of them issued price targets. The median was $225, with an average of approximately $237. From the lowest of $62 to the highest of $800, there's a 13-fold difference. And the stock is currently trading around $114, still well below the sell-side median.

As you can see, this group of people who know SpaceX best can't even agree on a range for its fair valuation.

As for Tuesday's SpaceX earnings call:

Morgan Stanley's lead analyst Adam Jonas listed in his latest report the things he wants to hear: whether SpaceX plans to add more than 2 gigawatts of computing power next year; whether there are new major cloud computing partnerships; and the usage trends of Grok on Cursor and Cursor's annual recurring revenue growth rate. He also flagged three risk points: capital expenditure guidance significantly exceeding ~$50 billion, another round of financing before year-end, and slowing Starlink subscriber growth. The first one is almost certain to happen.

SpaceX has also launched a special question-submission page built by its own AI chatbot Grok for this call, where the public can submit and vote on questions—a model borrowed from Tesla's previous approach of using Say Technologies to solicit shareholder questions. Looking at Reddit, the largest U.S. forum community, and the retail investment community Stocktwits, the two most common retail demands are: whether management will provide detailed segment-level data for the first time, and whether there will be any form of reassurance regarding Thursday's lock-up expiry.

First Shareholders Can Finally Sell

On Thursday, August 6th, SpaceX's first lock-up expiry arrives.

Unlike the traditional single 180-day unlock, SpaceX designed a tiered, rolling release schedule to spread supply over several months and avoid a cliff-style crash on any single day. The trigger for the first tranche is the second full trading day after the Q2 earnings release. It applies to employees and some early investors, representing 20% of their restricted shares, up to a maximum of 911.5 million shares.

That number is nearly half again larger than the total shares SpaceX sold at its IPO, including the over-allotment, which was approximately 629 million shares.

Another more tangible comparison: the current total market value of freely tradable SPCX shares is about $86 billion. At $114.53 per share, the shares released on this day would be worth over $104 billion. In other words, the sellable shares flooding the market on this day will be larger than the entire existing float. $104 billion needs to find counterparties in a pool of $86 billion.

Regarding holder composition: employee equity will be released in five tranches of 7% each on days 70, 90, 105, 120, and 135 after listing. Musk himself, along with several undisclosed large institutional holders, have signed 366-calendar-day lock-up agreements, running through June 12, 2027.

An engineer who joined in 2019 with a single-digit strike price would still realize dozens of times their investment selling at $114. Their considerations would be house down payments, tax planning, and concentrated position risk. Some wealth management firms have already set up services specifically for SpaceX employees, reminding them that the estimated tax payment deadline for August sales is September 15th.

One firm explicitly stating it won't sell is Coatue, the major U.S. tech-focused investment fund. One of their investors said "I'm holding" in a CNBC interview, adding a memorable point: his biggest mistakes all came from optimizing for the short term. He cited Nvidia as an example, saying selling it back then was one of his bigger errors.

ARK founder Cathie Wood, known as "Woodie," has been buying the dip in SpaceX at a position where it has lost an entire Tesla's market cap.

Duan Yongping, one of the most well-known value investors in the Chinese investment community and founder of BBK Electronics, has also made a move at this level. On July 23rd, he posted his trade records on Xueqiu: selling SpaceX put options with a $115 strike price, collecting a premium of $23.26. This means if SpaceX falls below $115 at expiry, he'll take delivery at $115, and after accounting for the premium already received, his actual cost is $91.74 per share—he doesn't lose money as long as it stays above $92. If the stock remains above $115 at expiry, he keeps the premium as pure profit—a