Transitioning computing power services provider to generate revenue: Musk's xAI plans to provide large-scale computing support to programming startup Cursor.

Against the backdrop of a continued strong demand for computing power, Elon Musk's artificial intelligence company xAI plans to provide large-scale computing support to the programming tool startup Cursor. According to reports, Cursor may use "thousands of pieces" of GPUs obtained from suppliers such as NVIDIA Corporation (NVDA.US) to train its new generation code model Composer 2.5. The report pointed out that this cooperation signifies that xAI is beginning to export its self-built large-scale computing resources externally, extending from a single model development company to a role that also serves as a "computing service provider", becoming a potential alternative choice for mega-scale cloud service providers such as Microsoft Corporation (MSFT.US), Alphabet Inc. Class C (GOOGL.US) and new cloud service providers like CoreWeave (CRWV.US), Nebius (NBIS.US). For xAI, renting out computing power externally can alleviate the high cost pressure of its data center construction and operation, as well as strengthen its connection with the developer ecosystem. Especially for companies like Cursor that have a large amount of developer data, in the increasingly competitive AI programming race, they possess potential strategic value. This move is also expected to increase xAI's sources of income, and alleviate concerns for some investors during the process of Elon Musk's commercial space giant SpaceX moving towards its first initial public offering (IPO). It is reported that SpaceX's acquisition of xAI before the IPO is the most controversial variable in this upcoming listing. As a newcomer to the large model race, xAI is currently in a period of high research and development investment, operating losses exceeding $300 million per month, with projected net losses exceeding $4 billion by 2025, and no profitability in the short term in sight. The market's general concern is that after this acquisition, xAI's massive cash burn will be included in SpaceX's financial statements, diluting Starlink's profitability potential, and even leading to a situation where the public company pays for Musk's personal entrepreneurial projects, harming the interests of public shareholders, making it one of the biggest uncertainties for SpaceX's IPO. Computing resources are becoming one of the core variables in the current competition in AI. Musk has previously stated that xAI's future competitive advantage will come from a larger supply of computing power. Over the past two years, xAI has continued to advance its data center project named "Colossus," deploying approximately 200,000 GPUs and planning to expand to a scale of 1 million. However, there is still room for improvement in operational efficiency for xAI. The company has disclosed internally that the GPU utilization rate in its model training is approximately 11%, significantly lower than the industry's common range of 35% to 45%. Providing computing power externally is also seen as a realistic path to improve resource utilization. Musk had apologized in March for not "correctly" building xAI, and stated that they are undergoing a reconstruction process. It is worth noting that this is not the first collaboration between Cursor and xAI. In March of this year, xAI hired two former Cursor product engineering managers, Andrew Milich and Jason Ginsburg. According to previous reports, they will be responsible for xAI's product team and report directly to Musk and xAI President Michael Nicolls. Cursor's annual recurring revenue last month has reached $2 billion. Insiders say that the company's revenue run rate (Revenue Run Rate, an estimate of projected revenue for 12 months based on current sales) has doubled compared to three months ago. Around 60% of the revenue comes from enterprise customers, including both companies using Cursor for the first time and existing customers increasing seat numbers. Established in less than five years, Cursor has become one of the fastest-growing startups in history. Its products have now integrated into the daily work of software programmers, with its client base ranging from tech companies like OpenAI to organizations like Anheuser-Busch InBev SA/NV Sponsored ADR (AB InBev) that are not typically associated with software. In November of last year, the startup was valued at $29.3 billion in a funding round led by venture capital firms Accel and Coatue, making it one of the highest-valued AI startups in the United States. Last month, reports surfaced that Cursor was in talks for financing at a valuation of approximately $50 billion. But as large artificial intelligence startups like Anthropic and OpenAI actively enter the code assistant field, Cursor also faces intense competition pressure.