SpaceX Quarterly Upends Wall Street's Valuation Framework. The explosive expansion of AI and computing power businesses is driving this companyaerospaceThe company has accelerated its transformation into a technology super platform. Major Wall Street banks have raised their price targets.
According to the wind-chasing trading desk,JPMorgan ChaseRaised the SpaceX price target from $225 to $240 in the latest research report, maintaining an "overweight" rating.
The core driving forces are:The company expects annualized recurring revenue (ARR) to exceed $100 billion in December 2026 — significantly surpassing JPMorgan Chase's previous forecast of $75 billion in full-year 2027 revenue — while moving forward the time node for achieving $1 trillion in annual revenue from 2031 to 2030.
According to a JPMorgan report, the bank raised its 2026 revenue forecast by 14% to $47 billion and its 2027 revenue forecast by a significant 64% to $123 billion.
The core catalyst of this valuation reshaping is the unexpected acceleration of the monetization process of computing power. SpaceX plans to expand its computing power deployment from more than 2 GW in 2026 to 5 to 10 GW in 2027, much higher than JPMorgan Chase's previous estimate of 4.2 GW.
At the same time, the company newly disclosed a US$6.7 billion six-month cloud service contract from an anonymous customer, superimposing major customers such as Anthropic, Google and Reflection AI that had previously signed contracts. The AI cloud business has formed an intensive commercial cash-out rhythm, which has a direct impact on the market valuation logic.
Previously,Goldman SachsThe report raised the target price of SpaceX to $220, while Citi saw it above $900 in the long run.
SpaceX closed down 13.6% at $108.27 on Wednesday.
AI cloud contract accelerates landing, mysterious big customers trigger market speculation
The commercialization of SpaceX's AI computing power is advancing at a rate that the market didn't expect.
At present, the company has signed AI cloud computing power contracts with a number of head technology institutions:$920 million per month for Anthropic, $1.25 billion per month for Google (effective from October), $150 million per month for Reflection AI, and a combined $6.7 billion for an anonymous customer over six months from October (equivalent to about $1.12 billion per month).
The identity of the mystery customer has sparked widespread speculation. The range of potential buyers with an average monthly spending of more than $1 billion is extremely limited and could includeMetaOpenAI,MicrosoftSoftbank,NVIDIAOr other agencies. The large scale of the contract further confirms the structural tension of the current imbalance between supply and demand in the computing power market.
On the pricing side, Musk believes that the pricing of Vera Rubin architecture GPUs has the potential to reach $30 to $50 per watt, which is much higher than the $11 to $16 per watt range adopted by JPMorgan Chase in previous modeling.
Higher computing power scale and higher realization pricing make SpaceX expected to realize the more than $100 billion AI revenue originally expected to be realized by 2028 to 2027.
Expansion of computing power accelerated, and capital expenditure was significantly advanced
The acceleration of the computing power expansion plan directly drives the scale of capital expenditure to a new height.
SpaceX is expected to spend nearly $200 billion in capital expenditure in both 2027 and 2028 — with the 2027 forecast raised to $196 billion from the previous $128 billion — to support the construction demand of 3 to 8 GW of additional computing power in 2027, corresponding to an investment of about $30 billion to $40 billion per GW.
Huge capital investment will continue to suppress free cash flow, with a projected free cash flow shortfall of approximately $108.2 billion in 2027. This trend is highly consistent with the capital cycle of hyperscale cloud computing vendors.
It is worth noting that SpaceX has currently achieved a recovery cycle of about one year or even less in new computing power investment, and JPMorgan Chase believes that this efficiency is significantly ahead of the industry.
SpaceX's competitive advantage in computing power construction lies in its migration and application of engineering capabilities accumulated by rocket and satellite businesses to data center construction, and reduction of energy costs by self-built power supply (BTM) instead of completely relying on the power grid, thus achieving more efficient capacity delivery than its peers.
CEO Musk's planned total installed capacity target of ground computing power is 15 to 20 GW, but GPU supply is the main bottleneck at present.
Exclusive binding to NVIDIA, "Xingzhi" satellite opens orbital computing power layout
On the chip strategy,SpaceX has decided to adopt Nvidia GPUs exclusively within its data centersThe core reason is that the Vera Rubin architecture has significant technical advantages-the NVL72 rack configuration is described as a "disruptive simplification" of the previous-generation GB NVL72 solution.
As a consideration for deepening cooperation, SpaceX is expected to receive an "extremely considerable proportion" of Nvidia's GPU shipments in 2027.
The cooperation between the two sides also extends to the field of orbital computing power. SpaceX and Nvidia jointly launched the first AI satellite project, named "Starmind", which also uses Vera Rubin architecture as computing hardware, and is scheduled to start launching and deployment in 2027.
This marks the official extension of SpaceX's AI computing power map from the ground to space, laying a hardware foundation for its long-term construction of orbital computing power network.
Musk predicts that as time goes by, the proportion of computing power used to train Grok model will drop to about 10% of the total computing power of SpaceX AI, and the rest of the capacity will be turned to inference services and external leasing, further opening up the commercialization space.
Starship Technology Breakthrough is Imminent, Connection Business Pattern Expands
In terms of the progress of Starship, the management has expressed relatively clear technical confidence.
After the 13th test flight, SpaceX has a great grasp of the solution to the "thermal protection problem"-which was previously regarded as the core technical challenge of Starship's re-entry phase.The company plans to attempt to capture the parent rocket for the first time in its 14th launch at the end of this month, but it still needs to complete several preconditions, including regulatory approvals.
At the same time, the management said that the goal is to achieve the daily launch frequency of Starship in one year, and it plans to build five launch towers.
In the connectivity business, President and COO Gwynne Shotwell confirmed that SpaceX plans to launch its V2 mobile satellite next year and launch its domestic Starlink mobile service in the U.S. by the end of 2027.This relies on the 65MHz spectrum assets that the company acquired from EchoStar with FCC approval in May 2026, but the related transaction is not expected to be completed until November 2027.
Shotwell said Starlink's mobile service is on track to win a "significant number" of subscribers from the three major U.S. telecom operators, but JPMorgan maintained its judgment that Starlink still needs to be deployedAcerStation, can be fully compared with traditional operators in user experience.
On the government and enterprise business side, SpaceX won more than $6 billion in multi-year U.S. government contracts in the second quarter, mainly from two major projects of the Space Force's Starshield security satellite network, pushing the total backlog to $47.5 billion at the end of the second quarter, of which about $20 billion was added in this quarter alone.
Shotwell said that the company has not lost any corporate customers so far, and expects that the revenue of corporate and government businesses will surpass that of consumer businesses in the future. JPMorgan Chase expects this intersection to occur in 2029.
The pressure of lifting the ban is approaching, and the risk factors cannot be ignored
Although the fundamental picture is quite bright, an important technical pressure in the short term cannot be ignored.
The lock-up period of 911.5 million shares will expire on Thursday, August 6, and the potential release is equivalent to 143% of the current 639 million shares in circulation. JPMorgan Chase believes that the market has made considerable pre-position adjustment before the lifting of the ban, but there are still many batches of lifting of the ban in the following months.
From the medium-and long-term risk dimension, JPMorgan Chase listed six key downside risks:
Once Starship encounters technical delays or regulatory obstacles, it will implicate multiple business lines; The structural shortage of GPU and power supply may restrict the pace of computing power expansion; The cumulative negative free cash flow expected to exceed US$300 billion between 2026 and 2030 depends on continuous external financing support; Regulatory uncertainties arising from export controls, spectrum allocation and environmental reviews; And the highly concentrated governance risk created by Musk's holding of about 82% of voting rights.
Based on 2028 GAAP earnings of $7.02 per share and a price-to-earnings ratio of about 34x, JPMorgan arrives at a price target of $240, which represents a significant premium to the median valuation of similar large-cap tech companies of about 20x to reflect SpaceX'saerospaceMarket leadership in the three major tracks of launch, satellite connection and AI computing power.
