Nvidia and SpaceX have formed a "match made in heaven"! Rubin's computing power is heading to space orbit, while Musk anchors SpaceX's valuation to a Type II Kardashev civilization.

Just after "AI chip superpower" NVIDIA Corporation (NVDA.US) announced an upward adjustment in global AI capital expenditure expectations and a strong earnings report amid a new bull market for the AI computing industry chain, Elon Musk attempted to push the expansion of AI infrastructure from the ground into space orbithis company SpaceX (SPCX.US), which he founded and leads, plans to launch the first batch of AI data center satellites powered by the next-generation NVIDIA Corporation computing clusterVera Rubin architecture AI GPU clusters in the fourth quarter of 2027, aiming to achieve "significant scale" by 2028. This does not mean that ground-based data centers will be rapidly replaced; rather, it bets on orbital computing power being able to bypass significant bottlenecks such as ground electrical grids, land, and water resources, thus becoming a new layer of supply for AI computing power. For NVIDIA Corporation, this represents an extension of the Vera Rubin potential market from "ground AI factories" to "orbital AI factories"; for SpaceX, it integrates Starship launches, orbital energy, satellite networks, and AI cloud computing into a vertically integrated platform. However, Wall Street financial institutions like Evercore predict that substantial revenue data will not arrive until the 2029 fiscal year, with heat dissipation, radiation, collisions, and regulation remaining key constraints on commercialization. NVIDIA Corporation has forged a "match made in heaven" with SpaceX, planning to send Vera Rubin into orbit. In some parallel universe, NVIDIA Corporation CEO Jensen Huang and "Mr. Everything" and the world's richest man Elon Musk might be a pair of cosmic superheroes. But in this universe, in this world, they are still planning what seems like an inconceivable space AI super blueprintNVIDIA Corporation provides the core computing power, and SpaceX offers launch and orbital infrastructure, together expanding the physical boundaries of AI computing. Vera Rubin and Starship form a closed-loop space hardware and software system, but the current valuation mainly reflects long-term options rather than recent cash flow. This project is part of the increasingly close collaboration between NVIDIA Corporation and SpaceX. During NVIDIA Corporation's earnings call on August 26, CFO Colette Kress stated that the Vera Rubin system has been fully put into production, with core shipment targets including AI cloud computing and computing power leasing leaders such as Oracle Corporation (ORCL.US), Amazon.com, Inc. (AMZN.US) AWS, and SpaceX (SPCX.US). "We believe that NVIDIA Corporation clearly positions SpaceX AI as one of its key partners, further demonstrating the depth of the relationship and proving that SpaceX AI is among the earliest Vera Rubin customers," stated a team of analysts from Evercore ISI in an investor report. "This relationship is also mutual: NVIDIA Corporation disclosed it holds 122.8 million shares of SpaceX stock, making NVIDIA Corporation both an important supplier and a significant shareholder of SpaceX." NVIDIA Corporation's latest earnings report shows revenue of $96.22 billion for the second quarter of FY2027, an increase of 106% year-on-year; data center revenue was $89 billion, up 117%; adjusted earnings per share were $2.22. The company anticipates third-quarter revenue to be approximately $108 billion, with a fluctuation of 2%, and has unusually forecasted a revenue growth of about 70% for the 2028 fiscal year, significantly exceeding Wall Street's previous expectation of 44%; NVIDIA Corporation management stated during the earnings meeting that Vera Rubin has been fully put into production and started delivery, which largely indicates that the growth constraints for NVIDIA Corporation mainly come from memory capacity and bottlenecks in advanced manufacturing capacity at Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR, rather than insufficient orders. After NVIDIA Corporation announced its earnings, Wall Street raised its target prices across the board. As of last Friday, NVIDIA Corporation's stock closed at $217.55. Citigroup raised its target price from $300 to $315, maintaining a "buy" rating; Goldman Sachs Group, Inc. increased its target price from $285 to $300, and Morgan Stanley raised its target price from $288 to $300. Their unanimous bullish judgment is based on the continued strong demand for AI computing power, the ramp-up of Vera Rubin, and the strengthening of NVIDIA Corporation's comprehensive hardware and software platform advantages. Musk stated that the data center designed by SpaceX would completely rely on NVIDIA Corporation AI GPU accelerators. Additionally, SpaceX completed a stock-only acquisition of Cursor valued at approximately $60 billion in August. Jensen Huang recently stated in a PowerPoint presentation that Cursor has been utilized across NVIDIA Corporation. The latest third-party statistics indicate that this AI programming assistant has an annual revenue run rate exceeding $1 billion. Regarding the commercialization of the space AI data centers that Musk and Huang focus on together, Evercore estimates that any significant contribution from this project to SpaceX will not materialize until at least the 2029 fiscal year. "Importantly, we do not account for any orbital computing revenue data for 2026 or 2027, and we expect the first 1 gigawatt of space AI orbital computing power to come online in FY2029; by the end of that fiscal year, the orbital computing capacity will reach 8 gigawatts, while the ground computing capacity will be 10 gigawatts," Evercore said. "If the originally scheduled orbital infrastructure for FY2029 is advanced to FY2028, it will significantly release demand for AI computing resources at the inference end; if calculated based on Musk's discussed economics of $30 to $50 per watt, it could significantly increase our forecast." Evercore has given SpaceX an "outperform" rating, with a target price set at $230. As of Friday's market close, SpaceX's stock price was $141.50. "The key question is what Musk's so-called 'significant scale' in 2028 actually meansit represents the scale of computing power capable of generating revenue data, or is it a broader technological validation phase prior to commercialization," Evercore added. However, according to a recent report from the Brookings Institution, space-based data centers face significant technical barriers. "Ground data centers use both air-cooling and liquid-cooling systems to dissipate heat through conduction and convection," said the Washington, D.C.-based think tank. "They first transfer heat from the chips to the liquid coolant, which then dissipates the heat to the surrounding atmosphere. For data centers floating in the vacuum of space, this cooling strategy is almost impossible to implement since there is no atmosphere to absorb heat. In fact, thermal radiation may be the only way to dissipate heat generated by orbital data centers, but some scientists have calculated that properly dissipating the heat generated by a single orbital data center would require an astonishing 2.15 million square feet of radiators," the Brookings Institution stated in a report. Other issues include equipment damage from continuous exposure to the most direct and powerful solar UV radiation, as well as an increased probability of orbital collisions due to the growing number of satellites. Regulatory approvals could also pose a problem. Are AI data centers about to enter a new dimension? The AI super factories are gearing up to move from the ground to orbit. SpaceX's prospectus clearly defines the "construction of a continuously expanding space civilization, ultimately progressing towards a Type II civilization capable of harnessing all the energy of the sun" as a long-term paradigm shift; Musk himself has stated that lunar satellite factories, mass drivers, and AI hardware deployments exceeding 100 terawatts annually will drive humanity toward "non-trivial progress" toward a Type II civilization. Thus, orbital AI data centers are not isolated projects but the first layer of infrastructure connecting "earth-constrained computingsolar system-level energymulti-planetary civilization." A Kardashev Type II Civilization, as discussed, can utilize all the energy of its star, while a Type III civilization would dominate the energy of the entire galaxy, far exceeding SpaceX's current narrative. The so-called $28.5 trillion Total Addressable Market (TAM) that SpaceX claims further capitalizes this civilization vision: traditional space business is only $370 billion, the high-speed connectivity business led by Starlink is $1.6 trillion, while the AI comprehensive business market size reaches $26.5 trillion, accounting for approximately 93% of the total size, including $2.4 trillion for AI comprehensive computing infrastructure, $760 billion in consumer subscriptions, $600 billion in digital advertising, and $22.7 trillion in enterprise applications. The underlying judgment of Musk and other top management at SpaceX is that global data center computing power demand could reach 235 gigawatts by 2030, with 70% used for AI, and that the terrestrial electrical grid, land, approvals, and environmental capacities cannot support terawatt-level expansion; the sun accounts for about 99.8% of the energy in the solar system. Therefore, SpaceX plans to begin commercializing modular orbital computing by the end of this decade, with a long-term goal of deploying 100 gigawatts of AI computing power to orbit annuallyif operated continuously throughout the year, its energy consumption would be approximately one-fifth of the total U.S. electricity generation in 2025. However, from an investment perspective, the $28.5 trillion is the theoretical market boundary, certainly not a revenue or profit forecast; orbital AI is essentially a long-term, highly convex technology option. Achieving the annual deployment of 100 gigawatts requires Starship to complete thousands of launches each year, sending about 1 million tons of equipment into orbit and resolving issues like radiation dissipation, irreparable chip failure, satellite mass manufacturing, and regulatory challenges. At the same time, SpaceX's AI business had capital expenditures of $12.727 billion in 2025, with an adjusted EBITDA loss of $1.237 billion, and further capital expenditures of $7.723 billion in the first quarter of 2026 with an adjusted EBITDA loss of $609 million, with the company acknowledging that sustained profitability will require years of investment cycles. Therefore, recent valuations should be grounded in cash flows from launches and Starlink, while orbital AI contributes long-term upward elasticity; what investors truly need to track are the unit cost of Starship at orbital insertion, launch frequency, computing power per ton, heat dissipation reliability, and computing power utilization rate, rather than directly endorsing current valuations with the $28.5 trillion figure.