Former Atomic Semi rebrands as 'Fab2' to mass-produce modular chip fabs

The semiconductor industry’s reliance on massive, centralized fabrication plants may soon face a radical alternative. Fab2, formerly known as Atomic Semi, is pioneering a modular approach to chipmaking with its "fab fab" concept—a factory that manufactures smaller, software-defined semiconductor fabrication plants. Founded by industry veteran Jim Keller and DIY fabrication pioneer Sam Zeloof in 2022, the company has now rebranded, relocated its headquarters to Texas, and is scaling up operations to produce its vision.

Building the "Fab Fab": A New Model for Chip Production

Fab2’s core innovation lies in its ability to design, assemble, and mass-produce complete semiconductor fabs in-house. Unlike traditional foundries that rely on sprawling 300mm wafer lines, Fab2’s approach focuses on compact, replicable facilities tailored for rapid prototyping and low-volume production. The company fabricates every critical component internally—from pumps and gas lines to lithography systems and vacuum chambers—before integrating them into functional machines. These machines are then assembled into full-fledged fabs, which can be deployed as standalone facilities.

Key to this model is Studio, an in-browser electronic design automation (EDA) toolkit that enables collaborative schematic layout, simulation, and verification. Previously branded as Atomic Studio, the software streamlines the chip design process, allowing teams to iterate quickly without the constraints of traditional design tools. This integration of hardware and software positions Fab2 as a one-stop solution for emerging chipmakers.

From Garage to Factory: Proving the Concept

The idea of small-scale, software-defined fabs isn’t entirely new. Sam Zeloof, Fab2’s co-founder, demonstrated its feasibility years ago by fabricating lithographic chips in his parents’ garage, achieving feature sizes around 300 nanometers. While this achievement highlighted the potential of decentralized chipmaking, it also underscored a critical limitation: throughput. Electron-beam lithography, the technique Zeloof used, writes patterns directly onto silicon rather than projecting them through a mask, making it significantly slower than state-of-the-art methods like EUV scanners. A single patterning step for a small chip can take hours—far longer than an EUV scanner needs to expose an entire 300mm wafer.

This tradeoff means Fab2’s model is best suited for prototyping, research, and low-volume production rather than high-volume commercial foundries. However, the company argues that its approach aligns with the growing demand for agile, distributed chipmaking—particularly in sectors like AI accelerators, quantum computing, and specialized sensors where rapid iteration is paramount.

Expansion and Strategic Shift to Texas

Fab2’s relocation to Texas marks a strategic pivot from its original California base. The company now operates three facilities:

• A 120,000 square foot headquarters in Austin, serving as the hub for research and production.
• A 30,000 square foot "fab fab" in Lockhart, dedicated to assembling and mass-producing smaller fabs.
• The original 25,000 square foot "garage fab" in San Francisco, which remains operational for legacy projects and low-volume work.

The move to Texas reflects Fab2’s focus on scaling its workforce and operations in a region rapidly becoming a tech manufacturing hub. As of May 2026, the company employs approximately 84 people, according to data from Tracxn. Fab2’s growth trajectory received a significant boost in 2023 with a $15 million seed round led by the OpenAI Startup Fund, valuing the company at around $100 million. Notable angel investors, including Naval Ravikant, Nat Friedman, and Fred Ehrsam, also contributed to the round.

Competing Visions for the Future of Semiconductor Manufacturing

Fab2’s modular approach contrasts sharply with recent high-profile projects like Tesla’s Terafab, a proposed $119 billion megafab in Austin designed to deliver a terawatt of annual compute capacity. While Terafab aims for mass production at an unprecedented scale, Fab2 prioritizes flexibility, speed, and decentralization. These two models represent fundamentally different answers to the same question: How should the U.S. expand its chipmaking capacity?

The semiconductor industry’s traditional reliance on gigantic fabs has proven effective but inflexible, vulnerable to supply chain disruptions and geographic concentration risks. Fab2’s vision offers a complementary path—one that could democratize access to chipmaking while accelerating innovation in emerging technologies. As the demand for specialized chips grows, the success of companies like Fab2 may redefine the industry’s future, proving that sometimes, smaller is not just better—it’s smarter.