Report: SpaceX Plans In-House GPU Production, Warns of 'Significant Capital Expenditure'
On the eve of a $1.75 trillion IPO, SpaceX disclosed plans to independently manufacture GPUs through the Terafab project, a joint venture with Tesla and xAI, classifying it as a major capital expenditure. This move aims to break chip supply bottlenecks and strengthen its positions in space and AI, but due to extremely high technical barriers and substantial costs, it is flagged as a growth risk and has become a key variable affecting the company's valuation
As SpaceX prepares for an IPO potentially valued at $1.75 trillion, it disclosed plans to manufacture Graphics Processing Units (GPUs) in-house, simultaneously warning potential investors about risks related to chip shortages and high capital expenditures, bringing the company's ambitions and potential bottlenecks in the AI sector into sharp focus.
Excerpts from S-1 registration documents obtained by media show that SpaceX lists "in-house GPU manufacturing" as one of the "major capital expenditure" projects currently underway and warns that chip supply shortages could constrain the company's growth. The specific scale of this expenditure remains undetermined; SpaceX did not respond to requests for comment from the media.
The aforementioned plan is closely linked to the Terafab project, jointly advanced by multiple companies under Elon Musk's umbrella. Terafab is an advanced AI chip manufacturing complex planned by SpaceX, xAI, and Tesla to be built collaboratively in Austin, Texas, though key details regarding chip types and manufacturing technologies have yet to be disclosed.
GPU manufacturing has traditionally been one of the highest-barrier fields in the semiconductor industry, typically relying heavily on specialized foundries such as TSMC. SpaceX's disclosure on the eve of its IPO has sparked external interest in the feasibility and timeline of its AI chip strategy, adding new variables to valuation assessments by potential investors.
Key Risk Disclosure Ahead of IPO
SpaceX is expected to complete its listing this summer, with a valuation of approximately $1.75 trillion. In its S-1 registration file submitted to the U.S. Securities and Exchange Commission (SEC), the company lists in-house GPU manufacturing as a major capital expenditure project while flagging insufficient chip supplies as a potential growth bottleneck in its risk factors section.
SpaceX's GPU manufacturing plan is tightly tied to the Terafab project. Developed jointly by SpaceX, its subsidiary xAI, and Tesla, and spearheaded by Musk, Terafab is located in Austin, Texas, and is positioned as an advanced AI chip manufacturing complex.
Musk previously stated that Terafab's goals cover chips required for automobiles, humanoid robots, and space data centers, but it remains unknown which specific type of AI chip (including GPUs) will be produced. Furthermore, it is unclear which party—SpaceX, the Terafab development team, or its partner Intel—will assume the manufacturing technology role within the factory.
At Tesla's analyst meeting this Wednesday, Musk stated that by the time Terafab expands capacity, Intel's next-generation 14A process technology "may be quite mature or ready for mass production," calling this move "seemingly the right choice."
GPU manufacturing is widely recognized as a high-complexity challenge. Taking industry leader NVIDIA as an example, its GPUs are renowned for their versatility and suitability for large-scale data computation. Meanwhile, Alphabet-owned Google has taken a different path, adopting Tensor Processing Unit (TPU) routes optimized for specific functions to build AI models and run chatbots like Anthropic's Claude.
