Terafab

One-line summary: Tesla / xAI / SpaceX consortium chip-fab program announced by Musk on March 21, 2026; Intel joined April 7 as foundry partner; $55B initial / $119B total prototype investment in Austin, TX, targeting 1 TW/year of AI compute capacity.

What it is

Vertically integrated semiconductor fabrication program — chip design, lithography, memory, advanced packaging, and testing all under one roof. Prototype targets 2nm; full-scale facility uses Intel 14A. Long-term goal: 1 million wafer starts per month. Located in Austin near Tesla's Gigafactory Texas. Full-scale facility costs estimated $5–13 trillion (multi-decade horizon).

Why it matters to stock-market

Terafab is the largest single capex commitment the project tracks: $119B for prototype phases is roughly 3x Intel's annual capex. Two implications: (1) it locks in Intel as Musk's foundry of choice for the foreseeable future, validating IFS at the highest scale; (2) it creates direct demand for ASML High-NA EUV (the only tooling that supports Intel 14A simplification) — see asml.

Key facts

Announced: March 21, 2026, by Elon Musk
Partners: Tesla, xAI, SpaceX, Intel
Intel joined: April 7, 2026, as foundry partner
Investment figure: $25B per 24/7 Wall St. (April 7, 2026) — earlier reporting cited $55B initial / $119B total; the discrepancy is unresolved and likely reflects different program phases or reporting windows. Treat as uncertain pending official disclosure.
Scope (from source transcript, attributed to Musk in Dwarkesh Patel interview Feb 5, 2026): "That's the plan with Terrafab. Millions of wafers a month of advanced process nodes." Further: "I think the Terrafab's got to do memory, it's got to do logic, memory and packaging." Target: "I want a million wafers a month in 2030."
Location: Austin, Texas (near Tesla Gigafactory)
Process node: 2nm prototype; Intel 14A full-scale
Throughput goal: >1 TW/year AI compute capacity; 1M+ wafer starts/month by 2030
Intel's quoted statement: "Our ability to design, fabricate, and package ultra-high-performance chips at scale will help accelerate Terafab's aim to produce 1 TW/year of compute to power future advances in AI and robotics."
Musk on node choice: Intel's 14A "will be probably fairly mature or ready for prime time" when Terafab scales up

Strengths (from a thesis-input perspective)

Anchor scale — Intel 14A's most ambitious customer commitment to date
Multi-billion-dollar Musk-aligned commitment is hard to walk back
Validates Intel 14A at production scale

Weaknesses (from a thesis-input perspective)

Multi-year horizon — prototype timing not yet specified; full-scale is decadal
Capital intensity is unprecedented; financing structure unclear
Musk-aligned megaprojects historically face execution risk and timeline slippage

Distinction: Terafab vs. Tesla $3B Austin Intel-14A R&D Fab

Terafab is the SpaceX/Tesla/xAI/Intel consortium program at $25B+ scale targeting millions of wafers/month. Separately, Tesla is building its own $3B semiconductor R&D fab in Austin on Intel 14A at pilot/R&D scale ("a few thousand wafers per month"), per Tesla Q1 2026 earnings (April 22, 2026). These are two distinct investments that both use Intel 14A as the foundry node:

Terafab: consortium-scale production, $25B+, millions of wafers/month target
Tesla Austin R&D fab: Tesla-only, $3B, pilot scale, chip design/prototyping function

Both reinforce Intel 14A as the chosen node for Musk-ecosystem silicon, but the Texas R&D fab represents an earlier and smaller validation — operational R&D scale with Intel 14A ahead of Terafab's production ramp.

Terafab

Terafab

What it is

Why it matters to stock-market

Key facts

Strengths (from a thesis-input perspective)

Weaknesses (from a thesis-input perspective)

Distinction: Terafab vs. Tesla $3B Austin Intel-14A R&D Fab

Sources

Related