Tesla and SpaceX Terafab plan nears construction in Texas, with Elon Musk citing split compute for Optimus and AI spacecraft

Tesla is linked to a major manufacturing and computing buildout underway through its SpaceX partnership, with a reported $16.8 billion “Terafab” project approaching construction in Texas. The facility is described as being built in Grimes County, about an hour northwest of Houston, placing it within reach of a large regional industrial corridor.

The market report characterizes the Terafab as a large-scale effort that will be supported by a major compute strategy. In the same update, Elon Musk is cited as discussing how computing capacity would be allocated, with the report framing the split as 25% for Tesla Optimus and roughly 75% for AI spacecraft.

The portion aimed at Tesla Optimus refers to the company’s humanoid robotics program, which Tesla has said is designed to perform useful tasks. A dedicated compute share for Optimus implies the facility is intended not just for physical manufacturing, but also for running the data processing and training workloads needed to develop and refine robotic behaviors.

The larger portion earmarked for “AI spacecraft” points to an effort centered on SpaceX’s use of artificial intelligence in space operations. In practical terms, that allocation suggests the Terafab compute will be used heavily for training and optimization tasks tied to spacecraft autonomy, mission planning, or onboard decision-making systems, even though the report does not spell out the specific applications.

Tesla and SpaceX’s Terafab concept sits at the intersection of industrial production and advanced computing. “Terafab” is commonly used as a shorthand for a factory that would manufacture large numbers of compute-focused components at high volume. The report’s framing indicates the companies are tying fabrication throughput to algorithm development and model training demands, rather than treating computing capacity as a separate, downstream step.

While the market post provides key directional details on location, scale, and compute allocation, it does not include additional specifics on permitting milestones, engineering timelines, or any contract structure for the buildout. It also does not clarify what portion of the $16.8 billion budget is for the physical plant versus equipment, power infrastructure, or data center systems.

Notably, the report does not offer a breakdown of the expected production output once operating. It also does not state whether the compute allocation is measured by raw throughput, training time, or another metric. Those omissions matter because they affect how to interpret the operational focus between robotics development and spacecraft-related AI workloads.

For investors and watchers, the next practical indicators to monitor would be any formal construction-start announcements, equipment purchase disclosures, and updates tied to major permitting or utility interconnection. Additional detail on what “25%” and “~75%” mean in operational terms could also provide clues about the companies’ near-term priorities between humanoid robotics and AI-driven aerospace systems.