1 hour ago
1
Masayoshi Son wants to build a trillion-dollar city in the Arizona desert. Not a metaphorical trillion dollars. An actual $1 trillion industrial complex dedicated to manufacturing AI-powered robots and advanced technology components on American soil.
The SoftBank founder pitched the proposal, internally called “Project Crystal Land,” to potential partners including TSMC and Samsung on June 20, 2025.
What Project Crystal Land actually looks like
The Arizona complex would house production lines for AI-powered industrial robots alongside a dedicated free-trade zone designed to lure tenants with tax incentives. Think of it as an attempt to replicate China’s Shenzhen, the manufacturing powerhouse that churns out a staggering share of the world’s consumer electronics, except planted in the Sonoran Desert and focused squarely on AI and robotics.
Son isn’t planning to bankroll the whole thing himself. The financing structure relies on project-finance models, meaning individual components of the complex would be funded independently, limiting SoftBank’s direct capital exposure.
The outreach to TSMC and Samsung is telling. TSMC already operates fabrication plants in Arizona, and Samsung has been expanding its US semiconductor footprint.
He’s also tapping into SoftBank’s own portfolio. The Vision Fund has backed numerous robotics startups, and firms like Agile Robots are reportedly part of the conversation. The idea is to create a vertically integrated hub where chips get made, robots get assembled, and AI software gets deployed, all within driving distance of each other.
The $30 billion AI bet paying off
SoftBank has invested over $30 billion in OpenAI, targeting roughly 13% ownership of the company. SoftBank reported profits exceeding $45 billion tied to OpenAI in a single fiscal year.
Project Crystal Land also doesn’t exist in isolation. Son chairs the Stargate project, a separate $500 billion initiative aimed at dramatically expanding US AI data center capacity and the power infrastructure needed to run it. Between Stargate and Crystal Land, Son is essentially proposing $1.5 trillion in AI-related infrastructure spending on American soil.
Stargate addresses the compute bottleneck, building the data centers and energy capacity that AI models need to train and run. Crystal Land addresses the hardware bottleneck, manufacturing the physical robots and components that turn AI software into real-world products.
Why Arizona, and why now
Arizona has become a magnet for semiconductor investment, driven by federal incentives and state-level tax breaks designed to reshore advanced manufacturing. TSMC’s existing Arizona operations have already created a gravitational pull for suppliers and adjacent industries.
The timing aligns with broader US policy priorities. Washington has been aggressively pushing to reduce dependence on foreign supply chains for critical technologies, particularly semiconductors and AI hardware. The CHIPS Act funneled billions into domestic chip production.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) · 