Terafab

Name

Terafab

Legal name

Terafab

Website

https://terafab.ai

Company type

Private

Founded year

2026

Operating status

Operating

Ownership category

akta.pro rank

Product category

Semiconductor Manufacturing

NAICS

Semiconductor and Related Device Manufacturing (334413), Semiconductor and Other Electronic Component Manufacturing (33441)

SIC

Semiconductors & Related Devices (3674), Guided Missiles & Space Vehicles & Parts (3760)

akta.pro primary industry

Logic Foundries (Advanced Nodes) (HDAHACAA)

akta.pro secondary industries

Package Substrates & Interposers (ABF/BT, Silicon/Glass) (HDAHADAE), Assembly & Packaging Services (Wirebond/Flip-Chip/FCBGA) (HDAHADAA), Spacecraft Integration, Assembly & Test (IAT) (IMABACAK)

Keywords

• Semiconductor manufacturing
• AI chip fabrication
• Custom silicon design
• Advanced node foundry
• Orbital AI compute

GTM type

B2B

Offering type

Hardware or Manufacturing

Cost components

Infrastructure, Technology or R&D, Supply Chain, Personnel, Operations

Revenue model

1. Internal Chip Production: Terafab produces AI chips primarily for internal consumption by its founding companies: Tesla (vehicles, Optimus robots), SpaceX (AI satellites, orbital data centers), and xAI (AI compute infrastructure). The company aims to produce 1 terawatt of compute per year to meet internal demand, reducing dependence on external suppliers like TSMC.
2. Technology Licensing & Foundry Services: Intel provides technology licensing for the 14A process node to Terafab, with Terafab serving as Intel's first major customer for the node. The arrangement represents a technology licensing deal that could bring revenue to Intel while providing Terafab access to established fabrication technology.

Go-to-market motion2 records

Distribution channels2 records

Marketing channels4 records

Core offering

Terafab is a vertically integrated semiconductor fabrication joint venture that designs, manufactures, packages, and tests custom AI chips (AI5, AI6, D3) under one roof for its founding companies Tesla, SpaceX, and xAI. Using Intel-licensed 14A process technology and advanced packaging such as EMIB and glass substrates, the fab targets 1 terawatt of annual AI compute output for applications including Tesla Full Self-Driving, the Optimus humanoid robot, and orbital AI compute satellites.

Product overview

Terafab is a multi-phase semiconductor manufacturing initiative launched by Tesla, SpaceX, and xAI to build the world's largest chip fabrication complex in Austin, Texas. The platform combines chip design, manufacturing, packaging, and testing operations producing AI inference and processor chips including AI5 (FSD/Optimus), AI6 (Optimus), and D3 (space applications). The operation targets 1 terawatt of annual compute capacity—50 times current global AI computing—with orbital compute infrastructure to overcome terrestrial power constraints. Intel serves as a technology partner providing 14A process technology licensing and process support, while ASML and Applied Materials supply lithography and chip-making equipment.

Differentiator

Problem solved

Functional benefit

Products and services

• AI5 Chip Custom AI inference chip designed to power Tesla Full Self-Driving (FSD) camera-only unsupervised autonomous vehicles and Tesla Optimus humanoid robots, manufactured using Intel-licensed 14A process technology within Terafab's Austin fab. The chip is allocated internally to Tesla for vehicle and robotics applications.
• AI6 Chip Next-generation custom AI processor chip designed to power Tesla Optimus humanoid robots for large-scale deployment, produced in Terafab's Austin fab and allocated internally to Tesla.
• D3 Chip Space-focused AI chip designed for orbital compute and AI satellite applications, powering SpaceX's space-based AI data centers and orbital compute satellites. Produced in Terafab's Austin fab and allocated to SpaceX.
• Orbital AI Compute Infrastructure Space-based AI data center infrastructure using satellites equipped with Terafab-manufactured AI chips, deployed via SpaceX launch to provide scalable AI compute outside Earth power grid and cooling constraints. Aimed at supporting the Musk ecosystem's AI compute demand at terawatt scale.

Quantifiable outcome

• Targeting 1 terawatt of AI compute per year - 50x current global AI computing capacity
• +2 more outcomes

Named customers4 records

Segments1 record

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

AI capability3 records

Feature7 records

Partnerships

Four partnerships are on record, tiered core.

• IntelcoreTechnology or Integration · 7 April 2026Intel joined Terafab as the first external partner, providing chip design, fabrication, and packaging capabilities. Intel will license its upcoming 14A process technology to Terafab and serve as process consultant for the facility. Terafab is Intel's first major customer for the 14A node, representing a significant credibility boost for Intel's foundry business.
• TeslacoreStrategic or Co-development Partner · 21 March 2026Tesla is one of three founding companies of Terafab, launching the 'most epic chip-building effort ever' alongside SpaceX and xAI. Tesla will build and operate a $3 billion research fab at its Texas campus capable of processing a few thousand wafers per month for R&D. Tesla contributes vehicle AI expertise and Optimus robot applications.
• SpaceXcoreStrategic or Co-development Partner · 21 March 2026SpaceX is one of three founding companies of Terafab, contributing rocket launch capabilities and satellite manufacturing. SpaceX will handle high-volume chip manufacturing for the facility. The company operates the largest PCB manufacturing site in the U.S. and flies the world's most powerful rocket, enabling unique orbital AI compute deployment.
• xAIcoreStrategic or Co-development Partner · 21 March 2026xAI is one of three founding companies of Terafab, contributing AI training cluster expertise. xAI has built the first GW-scale AI training cluster and the largest coherent supercomputer. The company is the only entity capable of building orbital AI compute at scale, providing critical AI infrastructure expertise.

Scale indicators8 records

Recent moves6 records

Expansion highlights7 records

Direct peers

• TSMC: TSMC is the world's largest contract logic foundry and Terafab's most important comparable. Both build leading-edge custom silicon for major customers (NVIDIA, Apple, AMD for TSMC; Tesla/SpaceX/xAI for Terafab). TSMC's N2 (2nm) node directly competes with the Intel 14A process that Terafab is licensing.
• Intel Foundry: Intel Foundry Services is the most direct peer to Terafab: a leading-edge logic foundry targeting advanced nodes. Intel is also Terafab's process licensor and first external partner, making it both competitor and supplier. Terafab's 14A-based manufacturing puts the two in direct competition for external foundry customers going forward.
• GlobalFoundries: GlobalFoundries is a major contract foundry serving automotive, industrial, and communications markets with differentiated processes. While GF focuses on trailing-edge/mature nodes rather than leading-edge, its automotive/industrial customer base overlaps directly with Tesla's needs, making it a credible alternative supplier.
• Samsung Foundry: Samsung Foundry is the second-largest contract foundry, offering advanced nodes (3nm/2nm GAA) and competing for the same custom AI chip design wins that anchor Terafab's value proposition. Both target hyperscale AI and HPC customers with leading-edge logic processes.

Others

• ASML: ASML is the sole supplier of EUV lithography systems required for leading-edge semiconductor manufacturing. ASML has publicly confirmed collaboration with Terafab and is a critical equipment enabler — without ASML's EUV tools, Terafab's 14A process node would not be possible.
• Applied Materials: Applied Materials is a leading supplier of semiconductor deposition, etch, and inspection equipment used in fab production. Terafab is in discussions with Applied Materials for advanced AI and next-generation logic chip production equipment, making it a key infrastructure partner.

Broad incumbents

• NVIDIA: NVIDIA dominates the AI accelerator market with its GPUs (H100, B200) — the benchmark Terafab claims a 10x cost advantage against. NVIDIA's designs run on TSMC's leading-edge nodes, putting it in indirect competition with any vertically integrated AI compute effort, including Terafab's custom AI5/AI6/D3 chips.
• AMD: AMD is a fabless designer of CPUs, GPUs, and AI accelerators (MI300/MI325) manufactured at TSMC. AMD competes with Terafab's AI inference ambitions in the data-center and automotive AI markets, and represents the kind of external foundry customer Terafab could eventually serve.

Emerging players

• Cerebras Systems: Cerebras builds wafer-scale AI accelerators for inference and training, with chips fabricated at TSMC. While not a foundry competitor, Cerebras represents an emerging vertically-integrated AI compute model (chip design + system + software) that parallels Terafab's integrated approach to AI silicon.
• Broadcom: Broadcom is a leading fabless designer of custom AI ASICs (Google TPU, Meta MTIA) manufactured at TSMC. As a specialist in custom silicon for hyperscalers, Broadcom represents both a potential customer and competitor for Terafab's custom AI chip design capability.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights6 records

Customer concentration

Revenue estimate

Valuation estimate

Number of profiles

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Terafab