Atomic Semi no longer exists under that name. The semiconductor startup founded by renowned chip architect Jim Keller and DIY fabrication pioneer Sam Zeloof has rebranded as Fab2. It has shifted its main operations to Texas. The change signals more than cosmetic adjustment. It underscores an ambitious plan to manufacture not just chips but the small fabs that produce them.
Fab2 describes itself as a “fab fab.” That term captures the core idea. The company aims to mass-produce compact, software-defined semiconductor fabrication facilities along with all the tools they require. Pumps. Valves. Gas lines. Lithography systems. Vacuum chambers. The list runs long because Fab2 designs and builds nearly every component in-house. Then it assembles those pieces into machines. Those machines form complete small fabs. And those fabs, the plan goes, will be produced at scale inside a dedicated factory in Lockhart, Texas.
Tom’s Hardware first reported the rebrand and relocation. The company’s new site at fab2.com makes the vision plain. “fab2 prints chips fab2 prints fabs fab2 prints masks fab2 prints transistors fab2 prints chambers fab2 prints pumps.” The repetition drives the point. This outfit wants to control the entire stack. From raw components to finished production tools. Speed matters above all. Making chips should be fast, the site declares. Prototypes in hours rather than weeks or months.
The roots trace to a garage in New Jersey. As a teenager Sam Zeloof built a functional semiconductor fabrication setup in his parents’ garage. He achieved features down to 300 nanometers using basic lithography. That proof of concept still inspires the current effort. Yet scaling from one garage to industrial output demands far more. Keller, whose resume includes major processor designs at AMD, Intel, Tesla and Tenstorrent, brings the architectural heft. Together they raised a $15 million seed round in 2023 led by the OpenAI Startup Fund. Valuation at the time reached $100 million. Angels included Naval Ravikant, Nat Friedman and Fred Ehrsam.
Now the physical footprint has expanded and moved. Fab2 maintains three sites. A 120,000-square-foot facility in Austin serves as headquarters for research, development and chip production. Another 30,000-square-foot site in nearby Lockhart houses the growing “fab fab” itself. The original 25,000-square-foot garage-style lab remains in San Francisco but hiring has pivoted heavily toward Texas. Some 38 roles stand open at present. Many focus on mechanical engineering, precision systems, process development, lithography research and manufacturing operations. The company lists roughly 84 employees as of recent data.
But why Texas? Several factors converge. Lower costs. Business-friendly policies. Proximity to major semiconductor investments already reshaping the state. Samsung continues work on its massive Taylor campus. Other suppliers and research efforts cluster around Austin and beyond. Texas Governor Greg Abbott’s administration has distributed grants through the Texas Semiconductor Innovation Fund to multiple players. Recent awards went to Arm, Xycarb Technology and FormFactor among others. Fab2 arrives without fanfare of nine-figure subsidies yet. Its bet rests on technical differentiation rather than sheer capital scale.
That differentiation looks radical next to conventional fabs. Traditional facilities stretch across hundreds of thousands of square feet. They cost billions. They require years to build and ramp. Output focuses on high-volume runs of advanced nodes. Fab2 pursues the opposite. Small. Fast. Replicable. Targeted at prototyping, low-volume specialty parts, research and rapid iteration. Its tools emphasize electron-beam lithography for precision on small areas rather than full wafers. Throughput stays modest by design. The payoff comes in turnaround time. Hours instead of months. Software plays an equal role. The company developed Studio, an in-browser collaborative EDA platform for schematic capture, layout and simulation. It matches the speed of the physical fab, executives believe.
And the team can build anything. We design all the hardware and software needed to make chips and we make the fabs, tools, and components ourselves. Then we use automation to drive novel process engineering. We’re looking for exceptional, hands-on people who can iterate to the limits of what’s possible. Those words from the Fab2 site capture the culture. Hands-on iteration. Automation layered on top. No reliance on vendors for every subcomponent. That vertical integration carries risk. It also offers freedom from supply chain bottlenecks that have plagued the industry.
Observers draw parallels with other bold Texas projects. Elon Musk’s Tesla and SpaceX have pursued Terafab concepts aimed at terawatt-scale computing infrastructure. Estimates for one such effort run toward $20 billion or more. The contrast stands stark. Megafabs versus networks of small ones. Centralized production versus distributed, replicable facilities. Both approaches seek to expand American chipmaking capacity. Both respond to concerns over reliance on Asian foundries. Yet they answer the question differently. Fab2’s model could enable faster innovation cycles for startups, researchers and defense applications. It might not displace TSMC or Intel at leading-edge logic. Then again, it doesn’t claim to.
Recent coverage reinforces the momentum. RuntimeWire noted the shift from garage chips to factory-scale prototyping. Social media buzz on X highlights the departure from California after four years. One post framed it as another high-tech move drawn by Texas policy and lower friction. Broader semiconductor news from the past week shows the state continuing to attract investment. FormFactor received a $24.2 million grant for its Farmers Branch expansion. Xycarb secured funds to grow its Georgetown components plant. These moves add to Samsung’s Taylor project, now eyeing a second fab there amid surging demand.
Challenges remain. Building reliable semiconductor tools from scratch tests even seasoned engineers. Achieving consistent yields at small scale brings its own complexities. Regulatory approvals, cleanroom certification and talent acquisition in a hot market will demand focus. The company has not disclosed detailed timelines for its first replicated fabs or target process nodes beyond the prototyping emphasis. Yet the rebrand itself suggests confidence. Operations have consolidated around the Texas sites. The website now leads with fab2.com. Job postings flow through AtomicSemi domains but target the new locations.
Industry veterans watch closely. Keller’s track record lends credibility. Zeloof’s hands-on origins provide a distinct flavor rare among venture-backed hardware startups. Their combination points toward a hybrid. Garage ingenuity scaled by software, automation and serious capital. Success could open new paths for chip design and manufacturing. Faster feedback loops. Lower barriers for experimentation. More resilient supply options. Failure would underscore how difficult true vertical integration remains even at modest scale.
Either outcome carries weight. The United States has poured tens of billions into onshoring semiconductor capacity through the CHIPS Act and state incentives. Most funds support conventional large fabs. Fab2 represents a parallel experiment. Smaller. Nimbler. Potentially faster to replicate across regions. Its Texas footprint puts it squarely inside the state’s growing semiconductor corridor. Austin. Lockhart. Nearby research at UT Austin. Suppliers expanding in the area. The pieces align even if the model diverges.
So the company prints. Chips. Fabs. Transistors. The supporting infrastructure. All under one conceptual roof. The rebrand crystallizes that focus. The move to Texas supplies the ground. What emerges over the next few years will test whether this approach can deliver on its bold promises. For now the vision stands clear. Make chipmaking fast. Build the tools that build the tools. And do it in a state eager for the next wave of advanced manufacturing.
WebProNews is an iEntry Publication