What Is Terafab?

Terafab is Elon Musk's plan to build the largest semiconductor fabrication facility ever constructed — by an enormous margin. The project is a joint venture between three of Musk's companies: Tesla, SpaceX (which recently acquired xAI in an all-stock deal), and xAI. The goal is to produce one terawatt (1 trillion watts) of AI computing power per year — a figure that dwarfs all existing chip output on Earth combined.

Musk framed the project bluntly at the launch event: "We either build the Terafab, or we don't have the chips. And we need the chips, so we build the Terafab." He noted that the combined output of every semiconductor factory currently operating on Earth provides only about 2% of the compute his companies will need across all projects.

Key Facts at a Glance

What Chips Will Terafab Produce?

Terafab will manufacture two distinct categories of chips:

Why Build in Austin?

Terafab will be sited at or near the North Campus of Giga Texas, Tesla's existing gigafactory in Austin. Musk thanked Texas Governor Greg Abbott, who was in attendance at the launch event, for the state's support. On X the following day, Musk clarified that the initial Austin facility will focus on chip design and advanced testing — a proof-of-concept before the main Terafab facility, which will require thousands of acres and is still evaluating multiple locations.

The Supply Chain Crisis Behind the Announcement

Musk first flagged the chip supply problem at Tesla's Q4 2025 earnings call, warning investors that external capacity from TSMC, Samsung, and Micron would hit a ceiling within three to four years. He acknowledged his existing suppliers warmly — "We're very grateful to Samsung, TSMC, Micron and others" — before adding that their expansion pace simply isn't fast enough for his ambitions.

Current global AI compute output sits at roughly 20 gigawatts per year, according to Musk. Terafab's 1-terawatt target would be 50× the current global total. Even skeptics acknowledge that the scale of compute needed for billions of Optimus robots, full autonomy across a Tesla fleet, and orbital AI satellites is genuinely unprecedented.

AI in Space: The Most Ambitious Part

The most striking element of the Terafab vision is that approximately 80% of its compute output is destined for space. Musk argues that orbit is actually a superior environment for running AI:

• "It's always sunny" — satellites in orbit receive constant solar power, with no weather or night cycles interrupting output
• Solar panels in orbit generate roughly 5× more power than equivalent panels on Earth's surface
• As Starship reduces launch costs, orbital compute becomes cheaper than terrestrial data centers
• SpaceX is targeting 10 million tons to orbit per year to support this infrastructure

D3 chips produced at Terafab would power these orbital AI satellites, launched by SpaceX's Starship rockets. Musk showed animations of SpaceX launching satellites from the surface of the Moon during the presentation.

Terafab vs TSMC: Can It Compete?

For context: at full capacity, Terafab would scale to roughly 70% of TSMC's current global output — making it the most significant new fab ever announced. Terafab is targeting 2-nanometer process technology, the same cutting-edge node TSMC is currently ramping for Apple and Nvidia. This is extraordinarily ambitious for a company with no history in semiconductor fabrication.

The Skeptics' View

Not everyone is impressed. Electrek and others have pointed to parallels with Tesla's 2020 Battery Day announcement, where Musk promised a revolution in battery manufacturing with the 4680 cell — targets that remain largely unmet five years later. Critics note:

• Musk has no background in semiconductor production
• Tesla's auto business declined for the second consecutive year in 2025
• The full $20–25B cost is not yet incorporated into Tesla's 2026 capex plan
• AI5 chip was already delayed to mid-2027 before this announcement
• Putting 80% of compute in orbit has "essentially zero connection to near-term business reality," according to critics

That said, Musk's track record of being directionally right on big bets — even when timelines slip — means the announcement cannot be dismissed outright.

What This Means for AI in 2026 and Beyond

Regardless of timeline, Terafab signals that the AI compute race has entered a new phase. The major cloud providers (Microsoft, Google, Amazon) are already spending $50B+ each on AI infrastructure in 2026. Musk is betting that vertical integration — owning the chip design, fabrication, and deployment stack — is the only way to remain competitive at the scale he envisions.

For the AI industry broadly, Terafab adds another potential source of domestic US chip supply at a moment when geopolitical tensions around TSMC and Taiwan remain a strategic concern for the US government.