Musk's Terafab courts TSMC for Texas chipmaking months after naming Intel its first partner

Two deal structures are under discussion, and they affect Intel differently. In the first, TSMC would run the Texas facility itself and Terafab would be its anchor customer ^[16]. In the second, TSMC would contribute manufacturing expertise while other investors hold majority ownership ^[17].

The first structure is the costly one for Intel. A TSMC-run plant anchored by Terafab would give the Musk companies a second source for chips Intel was named to help make with 14A ^[16][2]. I'd expect the second structure to leave more room for the partner already in place. The third outcome is no deal at all, in which case Intel holders paid for a conversation ^[3].

I think the market priced a real risk. A partner targeting about 1 terawatt a year of output ^[10] is worth less to Intel once TSMC is also in the conversation. The counter-case comes from the same reporting. Cryptobriefing, which says the Musk companies appear to be shopping for a second foundry ^[13], also notes that nothing in Musk's comments says Intel is being replaced ^[7]. On that reading, the drop priced in a replacement Musk has not described. My view is wrong if the talks end with TSMC as a minority adviser and Intel's 14A still in Terafab's plans. In that case Intel keeps the work it was named for in April, and TSMC sits in an advisory role.

The scale figures explain why two foundries would want the work. The first phase in Grimes County carries about $16.8 billion of investment and at least 3,000 jobs ^[4], which puts capital at no more than about $5.6 million per job ^[14]. The footprint could eventually exceed 100 million square feet ^[18], and the chips are meant for Optimus robots, Cybercab vehicles, SpaceX data centers and xAI ^[9]. The reporting does not say whether a TSMC-run facility would be that first phase or a separate plant, or who the majority investors in the second structure would be.