SpaceX Aims to Reshape the Global AI Race with Massive Compute Expansion
SpaceX is set to dramatically scale up its ambitions in the global race for artificial intelligence. Elon Musk aims to boost the company’s computing capacity to approximately 13 gigawatts (GW) by the end of 2027. According to estimates from the research firm SemiAnalysis, this would represent a computing power comparable to what Google has developed over nearly 27 years. This particularly ambitious goal highlights the accelerating pace of investment in the infrastructure required for AI.
Rapid Growth in Computing Capacity
During the presentation of its Q3 2026 earnings, the company’s leadership indicated that SpaceX is expected to have 2.5 to 3 GW of computing capacity by the end of this year, before adding approximately 10 GW in 2027. As of September 30, the company had already reached 1.7 GW, up from 1.4 GW at the end of the second quarter.
This rapid expansion is largely driven by the development of its Colossus data centers, inherited from the acquisition of xAI, which was finalized in February 2026. However, the stakes go beyond simply building data centers. SpaceX is positioning itself to become a major provider of computing power for companies developing AI models. Contracts signed with key industry players illustrate this new strategic direction:
- Google has committed to paying $920 million per month to access approximately 110,000 Nvidia graphics processing units (GPUs), starting in October 2026.
- Anthropic is also among the clients utilizing SpaceX’s infrastructure.
In September, SpaceX announced a fourth computing capacity lease agreement, estimated at $13.3 billion annually. According to information provided by its Chief Financial Officer, these four contracts now represent $41.1 billion in potential annual revenue. It is important to note that these are projected contractual amounts, not yet realized revenue.
Financial Impact and the "Super Intelligence" Segment
This diversification is beginning to significantly impact financial results. In Q3 2026, SpaceX’s Super Intelligence segment generated approximately $7.6 billion in revenue, driven primarily by computing service contracts and the development of its AI-related activities.
The Energy Challenge and Regulatory Hurdles
However, this expansion raises a central question: the energy required to power these massive infrastructures. Data centers consume significant amounts of electricity and water to power and cool their equipment.
At Colossus 2 in Memphis, xAI faces particular controversy regarding the use of gas turbines for electricity generation. In July 2026, Reuters reported that 59 turbines had been installed without the required federal permits, based on exchanges with regulatory authorities.
The controversy has now extended beyond environmental concerns. The U.S. midterm elections on November 3, 2026, have strengthened the positions of political officials opposed to the unchecked expansion of data centers. Several candidates favoring restrictions won key races, while local initiatives aimed at limiting such installations also gained significant support. This political shift could complicate permitting processes and slow down certain infrastructure projects.
Beyond Earth: Orbital Computing
Meanwhile, the race for computing power is no longer limited to Earth. SpaceX is planning to launch experimental satellites dedicated to AI computing in orbit starting at the end of 2027, with a potential scale-up in 2028. Google is also pursuing this frontier with its Suncatcher project. Its first prototype satellite, launched on October 1, 2026, is already testing the functionality of its specialized processors in space.
Conclusion: A New Strategic Battleground
For SpaceX, the challenge is twofold: building fast enough to meet growing global demand for computing power while securing the necessary energy supply and infrastructure. In this new technological battle, the ability to develop AI models is no longer sufficient. Mastery of data centers, electricity, and computing capabilities has itself become a strategic competitive advantage.
