On August 6, the official announcement was made regarding the selection of a site for the ambitious Terafab project — a chip manufacturing plant that will be located in Grimes County, Texas. Initial investments from SpaceX and Tesla will total $16.8 billion. This decision, announced by Elon Musk, marks a new stage in the vertical integration of both companies.
Scale and Specialization
The production complex will occupy a colossal 9.3 million square meters. A key feature is the consolidation of all stages — from wafer production to packaging and testing — under one roof. This will optimize logistics and accelerate product rollout. The plant will provide at least 3,000 jobs for residents of Grimes and Brazos counties.
The product range will be clearly divided into two directions:
- Chips for edge computing and inference, intended for Tesla Optimus robots and Cybercab robotaxis.
- High-performance solutions for SpaceX orbital data centers.
Interestingly, the site was not chosen by chance: the plant will be located near Gibbons Creek Reservoir, which previously served to cool a coal-fired power plant closed in 2018. The company has committed to using exclusively this reservoir, without affecting the region's underground water sources.
Deal Structure and Uncertainties
Despite the scale, the project is characterized as a "general framework" in SpaceX's IPO prospectus. There are no binding legal agreements between SpaceX and Tesla: intellectual property rights have not been allocated, and neither party has firm obligations for further participation. Earlier, in April, Intel announced its intention to participate in the plant's construction, but its role and investment volume remain unspecified. According to some reports, Intel may become the key manufacturing partner.
The financial picture is also uneven. Initially, in March, Musk cited a figure of $25 billion. The current $16.8 billion is only the first-phase investment. At the same time, SpaceX previously acknowledged that total project costs could reach $119 billion. This suggests that we are witnessing only the beginning of a long-term and highly capital-intensive strategy.
Context: The Race for Computing Power
The Terafab announcement coincided with a sharp surge in SpaceX's spending on AI infrastructure. The company's capital expenditures in the second quarter grew more than sixfold year-over-year, reaching $18.37 billion, of which $15.83 billion was allocated to the AI segment. Musk himself forecasts an increase in own capacity to ~10 GW by the end of 2027.
Notably, for immediate tasks, the company continues to purchase chips from third-party manufacturers. Thus, on August 4, a partnership with Nvidia was announced: Starmind AI1 satellites will be equipped with Rubin GPUs and Vera CPUs. Musk directly stated that AI infrastructure will be built exclusively on Nvidia equipment. This underscores the strategy: use the best solutions available on the market now, while simultaneously developing own capabilities for future needs.
Tax Incentives and Public Tension
Such a large-scale project could not have avoided complex negotiations with local authorities. The Grimes County Commissioners Court approved a full exemption from county property taxes until 2036 in exchange for fixed payments: $10 million as a one-time payment and $20 million annually. The school portion of the incentives was approved by the governor in July under the JETI program.
However, disputes arose around the deal. After documents were published on June 6, a discrepancy emerged: SpaceX publicly stated investments of $55 billion, while the contract mentioned $5 billion by 2030 and 1,800 jobs by 2035. It also turned out that the company could exit the project at any time with 30 days' notice. Residents expressed concerns about the strain on the power grid and water consumption.
The company's lawyers proposed an alternative calculation, according to which the effective tax discount would be about 78%, and the total payments to the county over 35 years would reach $710 million. As environmental guarantees, SpaceX committed to building its own gas-fired power plants and operating on a closed-loop water cycle. Initial earthworks will begin in the near future.
My view: Terafab is not just a plant, but a strategic bet on full-cycle production of critically important components. However, the vagueness of commitments between SpaceX and Tesla and the dependence on external contractors such as Intel create significant risks. Given SpaceX's warnings in its IPO filing about the potential commercial inviability of orbital data centers, this project looks like a long game with high stakes, where success will depend not only on technology but also on the ability to build a sustainable economic model.