Musk Accepts $1 Trillion Price Tag for Tesla-SpaceX Terafab

Elon Musk agreed on November 6, 2026 that Terafab, the joint chip venture of Tesla (NASDAQ:TSLA) and SpaceX (NASDAQ:SPCX), could cost $1 trillion over time. The number came from investor Ron Baron at the Baron Investment Conference, and Musk, speaking by video, accepted it — with the caveat that the money will arrive in phases. Transcripts of the livestream provide the account.

The only official cost figure on record is far smaller. On August 6, 2026, Texas Governor Greg Abbott's office estimated the first phase, at Gibbons Creek in Grimes County, at more than $16.8 billion, on a 100-million-square-foot site. An earlier published estimate put the project's total cost above $100 billion. The trillion-dollar figure, in other words, is an aspiration, not a budget.

The market reaction was immediate, if unattributed. Tesla rose 4.22% to $358.20. SpaceX gained 8.53% to $134.16, and Intel (NASDAQ:INTC) climbed 7.75% to $56.30. No source ties those gains to the remarks, which followed a post-election selloff in all three stocks.

What scale is Musk actually claiming?

Musk said Terafab aims to produce a terawatt of AI chips per year — a goal he called "terrifyingly large." He estimated global AI chip output at roughly 20 to 25 gigawatts in 2026 and 35 to 45 gigawatts in 2027. A terawatt would be about 30 times 2027 production by his arithmetic. His own range implies 22 to 29 times, so the figure is approximate at best.

The production roadmap is similarly provisional. On SpaceX's October 20 earnings call, Musk targeted a first SRAM test chip — a simple memory chip used to validate the line — around May or June 2027. He set volume production for the third quarter of 2028 and called those dates "very aggressive."

Terafab makes no chips today. Meanwhile, SpaceX buys Nvidia (NASDAQ:NVDA) GB300s at a pace of roughly 200,000 a month. Nvidia, worth $4.51 trillion, reports earnings on November 17.

How does the Austin research fab fit?

The plan pairs a research fab at Gigafactory Texas in Austin with the Grimes County site. Austin tests chip processes; whatever works moves to Grimes County. Musk said this could compress a process cycle from about a year to about a month, and he acknowledged the plan could fail.

A process cycle is the time needed to run a change through the line and see whether it lifts yield — the share of working chips per wafer. Shorter cycles let a fab fix defects and cut costs faster.

Who pays for the later phases?

The financing question remains open. Tesla's Q3 free cash flow fell 61.63% while capex rose 110.5%. SpaceX earned $4.40 billion in Q3 but guided Q4 capex to $35 billion to $40 billion. If either company raises equity, existing shareholders are diluted.

Intel said in April it would join Terafab, giving its foundry a high-profile customer. That means a unit that lost $1.85 billion in Q3 helps build a rival.

Tesla trades at a P/E near 373x after Q3 operating income fell 55.79%. Terafab adds an open-ended claim on its capital. Special committees discussing a Tesla-SpaceX combination have not agreed on terms, so it remains unclear which company's shareholders will fund the project.

Until the first SRAM chip runs in 2027, or does not, the trillion-dollar figure stands as a vision attached to a $16.8 billion down payment.