SpaceX spends 55 billion to build Terafab AI chip factory: computing power infrastructure enters national competition

According to court documents, SpaceX plans to invest at least $55 billion to build the Terafab chip factory in Austin, Texas, and the total investment for the full expansion could reach $119 billion. The factory is jointly operated by SpaceX and Tesla, with Intel providing design and manufacturing support. The target annual output can support 200GW of computing power.

Core conclusion

On May 7, 2026, according to public hearing records disclosed in court documents, SpaceX is planning to build an AI chip manufacturing factory called Terafab in Austin, Texas, with an initial investment of at least 55 billion, and if all expansion is completed, the total investment can reach 119 billion. This super chip factory, jointly operated by SpaceX and Tesla and assisted by Intel in its design and manufacturing, marks the official upgrade of the competition for AI computing infrastructure from the enterprise level to the national level.

Key Points

Event Time: 2026-05-07 (court documents disclosed)
Affected objects: AI development teams, computing power service providers, chip industry
Core changes: AI chip manufacturing industry welcomes heavyweight new players, and the computing power supply pattern may be rewritten

Background: From space exploration to chip manufacturing

Elon Musk first unveiled Project Terafab in March this year. This chip factory in Austin is Musk’s most radical layout extending from space exploration to AI infrastructure. Unlike SpaceX, which focused on rockets and satellites in the past, Terafab directly targets the core bottleneck of the AI era - computing power is in short supply.

According to reports from the New York Times and CNBC, the factory is jointly operated by SpaceX and Tesla, and the chips produced will serve three major scenarios: AI training reasoning, robot control, and space data centers. Intel also announced last month that it would provide chip design and packaging support for Terafab. Musk has publicly stated that the factory's goal is to produce enough chips to support 200 gigawatts (GW) of computing power per year - equivalent to the power output of dozens of nuclear power plants for AI computing.

Key impact analysis

Dimensions	Changes	Impact on AI ecosystem	Recommended actions
Computing power supply	Target annual output of 200GW of computing power	Significantly alleviate the shortage of AI computing power and may reduce API call costs	Pay attention to Terafab's production timeline and adjust computing power procurement strategy
Chip landscape	Joint entry of SpaceX + Tesla + Intel	Breaking NVIDIA’s monopoly in AI training chips	Consider multi-architecture compatibility and not be bound to a single ecosystem
Investment scale	55 billion → 119 billion US dollars	The same level as TSMC’s Arizona factory, a national strategic investment	Assessing supply chain security and geopolitical risks
Space computing power	Support for space data centers	Earth-orbit hybrid computing architecture may change the way AI is deployed	Focus on architectural evolution in low-latency demand scenarios

Why this matters for AI practitioners

1. AI computing power costs may drop significantly

Currently, the main component of AI training costs is GPU computing power rental. The long-term shortage of NVIDIA H100/B200 clusters has pushed up the cost of using APIs such as OpenAI and Claude. If Terafab achieves its goal of producing 200GW of computing power per year, it will significantly increase market supply, thereby reducing inference and training costs.

For teams that use OpenClaw, Claude Code, n8n and other tools to automate content production, the reduction in computing power costs means that more parallel agents can be run and more complex tasks can be processed.

2. The strategic turning point of chip diversification

Intel’s participation in the Terafab project means the return of the x86 architecture in the field of AI chips. At present, the AI training market is mainly occupied by NVIDIA (CUDA ecosystem). If Terafab succeeds in mass production, it will form a tripartite structure of NVIDIA, AMD, and Intel-Terafab. Developers need to take precautions to ensure the multi-architecture compatibility of the tool chain.

3. New imagination of space computing

Terafab is uniquely positioned to power space-based data centers – delivering low-latency AI inference services around the world using solar power and no need for cooling systems. This is still in its early stages, but a breakthrough could revolutionize the geographic distribution of AI infrastructure.

Related extended information

Tool entry

This article covers the following AI tools and related platforms: OpenAI, ChatGPT, Claude Code, n8n, OpenClaw, NVIDIA, Intel, Claude. Tool brand names that naturally appear in the text will automatically trigger the tool floating entry of waytoclawearn.com.

Internal link guidance

🤖 Want to know how to reduce AI computing power costs? Watch the tutorial: DeepSeek V4 vs Claude Code: 90% Cheaper, Same Quality
💼 The democratization of AI infrastructure allows independent developers to create miracles: 18-Year-Old Built a $5,000/mo SaaS With AI Agents — Zero Hand-Written Code
📱 Want to build your own AI automated workflow? Reference tutorial: AI Agent-Driven Content Automation: n8n MCP Building Guide from Scratch