SpaceX Targets 10GW as Infrastructure Speed Reshapes AI Race
The artificial-intelligence infrastructure race is shifting from securing GPUs to delivering power and data-center capacity, according to SemiAnalysis. Chips cannot generate revenue until facilities are built and energized, making grid connections, construction schedules and power availability critical constraints. Operators that compress those timelines could reshape the economics, pricing and revenue ramp of AI computing services.
SpaceX is pursuing on-site power generation at its facilities to reduce dependence on lengthy utility-grid upgrades and connection queues. The company is targeting 10 gigawatts of computing capacity by 2027, SemiAnalysis said. The approach is recasting competitive advantage around execution speed: not simply how many accelerators a company can acquire, but how quickly it can energize them and turn installed hardware into revenue-producing capacity.
All Coverage
1 original reportsThe Backstory
The history behind this eventMusk Plans 10GW SpaceX AI Buildout With Microsoft as Top Buyer
Elon Musk is seeking to expand SpaceX beyond rockets and satellite communications into a major provider of artificial-intelligence infrastructure. The push comes as frontier models require rapidly growing computing and power capacity for inference as well as training, making multi-gigawatt data-center projects strategically important to chip suppliers, utilities, financiers and cloud-service companies.
SpaceX plans to add as much as 6 to 10 gigawatts of AI computing capacity in 2027, according to a SemiAnalysis report, with estimated capital spending of $300 billion to $500 billion. The analysis assessed inference economics for frontier models from OpenAI and Anthropic, and said Microsoft could sign a large contract that would make it the project’s biggest computing customer.
SpaceX AI Cloud Revenue More Than Triples to $2.6 Billion
SpaceX is expanding beyond rockets and Starlink by selling artificial-intelligence computing capacity to customers including Anthropic and Google, positioning itself in the fast-growing “neocloud” market. The business gives the company another source of recurring revenue and is increasingly viewed as a central pillar of its potential public-market valuation, even as the capital-intensive operation remains far from profitable.
The AI division generated $2.6 billion in revenue in SpaceX’s latest reported quarter, more than triple the previous level, according to the company’s quarterly financial results. Demand from Anthropic, Google and other customers drove the increase. The unit nevertheless recorded a $1.5 billion loss in the same quarter, underscoring the heavy spending required to build computing capacity and the uncertainty over when rapid sales growth will translate into earnings.
xAI Renamed SpaceXAI as Musk Targets Space-Based AI Computing Market
Elon Musk’s artificial intelligence company xAI has reportedly been folded into SpaceX and integrated as SpaceXAI, bringing rocket launches, satellite networks, AI models and computing services under one organization. If completed, the move would give Musk a direct foothold in the space-based data center market and offer an alternative to the power, cooling and land constraints facing terrestrial facilities.
The latest information says SpaceXAI plans to deploy AI computing satellites in 2028 and has secured computing orders from Anthropic and Google. However, as of July 20, 2026, the event data did not specify when the merger and renaming were completed. Nor did it disclose the value of the two companies’ orders, the number of satellites or the timetable for the first launches.
Subscribe to Mark Radar Weekly
Every Friday, the week's strongest signals in your inbox. Unsubscribe anytime.
If you search news on Google, you can set Mark Radar as a preferred source—our coverage will show up more often in your results. Set as preferred source on Google →