On September 11, news broke that Besxar, a startup founded by former OpenAI employees, is attempting to move semiconductor manufacturing into space to leverage the natural vacuum environment of space as a replacement for the costly cleanroom infrastructure on Earth. Recently, Besxar has begun verification testing of its orbital semiconductor manufacturing prototype via SpaceX's Falcon 9.
Core Logic: Going Where the Laws of Physics Are Already "Ready"
Ashley Pilipiszyn, Founder and CEO of Besxar and a former OpenAI employee, stated that the fundamental reason ground-based fabs are "extremely massive and complex infrastructure projects" is the necessity to build positive-pressure cleanrooms to block even the tiniest dust and contaminants—a single speck of dust landing on a wafer can ruin an entire chip.
"We are at a point in time where going to an environment where the laws of physics are already on our side is actually more cost-effective than doing it on Earth," Pilipiszyn stated. "Don't fight the laws of physics on Earth."
In Besxar's view, the vacuum environment of space naturally possesses a level of purity that is unattainable by ground-based cleanrooms, which is essential for protecting sensitive materials from contamination. Rather than spending billions of dollars to replicate a "near-vacuum" environment on the ground, it is better to directly utilize space itself as a cleanroom.
Furthermore, the deeper value of space manufacturing lies in the fact that microgravity environments can eliminate thermal convection and buoyancy effects. On the ground, thermal convection during material melting acts like a "blender," disrupting atomic arrangements and leading to crystal defects. In a microgravity environment, atoms can arrange themselves more quietly, resulting in more perfect crystal structures for grown compound semiconductors, with conduction efficiency potentially 10 to 100 times higher than ground-based products.
Josh Western, CEO and Founder of Space Forge, another startup also targeting space-based semiconductor manufacturing, previously noted that the vacuum and microgravity conditions in space can enable more effective development of entirely new semiconductor materials. "Growing compound semiconductors is a very intense and slow process; they essentially grow from the atomic level." "Therefore, gravity has a profound impact on them, fundamentally altering the bonds between these atoms. In space, you can overcome this hurdle because there is virtually no gravity."
Western added that space provides a better manufacturing foundation for almost any material. "Through the combination of microgravity and high-purity vacuum (without the need for multi-stage pumps), going to space can enable approximately a billion new alloy combinations and access to extreme temperatures ranging from -260°C to 260°C, allowing researchers to manufacture semiconductors that are '10 to 100 times more efficient than semiconductors on Earth.'"
Maiden Flight Verification: Wafer Samples Cleaner Than Those on the Ground
In October 2025, Besxar reached an agreement with SpaceX to verify its orbital semiconductor manufacturing prototype across approximately12 Falcon 9 flights, aiming to validate the concept of utilizing the natural vacuum of space to replace the costly cleanroom infrastructure on Earth.
On July 5, 2026, Besxar's maiden flight mission, "Mission Asimov," launched aboard a SpaceX Falcon 9 booster. Two V1 Fabships (manufacturing chambers) completed a suborbital flight with the booster, carrying compound semiconductor wafer samples such as GaAs and AlInAsSb from the University of Texas at Austin and the University of Virginia. It is important to note that these two microwave-sized "Clipper-class" containers are modular manufacturing units that utilize space properties (such as vacuum, microgravity, and thermal environments) as process conditions. They were mounted on the Falcon 9 booster and did not enter orbit; instead, they returned to Earth with the booster after a few minutes of suborbital flight.
The objective of this test was to verify whether the containers could withstand launch loads, protect the wafer samples from contamination, and expose them to the space vacuum. According to Compound Semiconductor, post-flight inspections showed that the container structures remained intact, and the wafer samples had no cracks, no warpage, and no visible damage.
Pilipiszyn further pointed out, "The samples that experienced the flight are the cleanest, with the lowest particulate content, compared to ground-based wafers that have never been to space... This is highly ideal for our subsequent scale-up." Independent material-level analysis of the compound semiconductor wafer samples is currently ongoing.
Technology Roadmap: Gradual Upgrades from Heating to Deposition
Besxar's iterative path is clear and gradual. The company plans to progressively increase the complexity of its experiments over the next two years:
Phase 1: Heat wafers to verify on-orbit thermal management capabilities.
Phase 2: Deposit a single material to achieve basic thin-film growth.
Phase 3: Expand to two or more materials, moving closer to true semiconductor manufacturing processes.
Besxar's ultimate goal is to mount larger-scale manufacturing modules on SpaceX's Starship, which is still under development. When Pilipiszyn first approached SpaceX three years ago, the initial discussions were about utilizing Starship launch services. However, they ultimately chose to first carry payloads via Falcon 9 boosters to reduce technical risks at a controllable cost.
Once capable of producing qualified samples, Besxar plans to supply wafers to leading chip manufacturers for producing advanced chips responsible for power regulation in data centers, robots, and NEVs (New Energy Vehicles). This market positioning avoids direct competition with TSMC (Taiwan Semiconductor Manufacturing Company) and Samsung in logic processes, focusing instead on the power semiconductor sector, which is highly sensitive to material purity.
Although the maiden flight verified basic feasibility, there is a huge gap between "engineering experiments" and "commercial manufacturing." The suborbital flight of the Falcon 9 booster only provides a vacuum exposure window of a few minutes, whereas true semiconductor thin-film deposition may require controlled environments lasting hours or even longer. When Starship will achieve normalized, low-cost operations remains unknown.
Multiple Companies Have Begun to Position Themselves
To date, Besxar has raised nearly USD 14 million, including a USD 9 million seed round led by Dauntless Ventures and Overture VC, with participation from Keymaker VC, 645 Ventures, and other institutions. The company has also received early support from the NVIDIA Inception program and has initiated a U.S. Department of Defense contract focusing on the R&D of defense-grade materials and radiation-hardened components.
Besxar is not the only player in the space semiconductor manufacturing track. United Semiconductors and Space Forge are also exploring the use of the space environment to produce higher-quality semiconductors. However, all players face a common bottleneck: how to bring a sufficient quantity of products back to Earth. Pilipiszyn expects that the output of a single factory will gradually expand from hundreds to thousands of wafers in the future, but this path is highly dependent on the maturation of low-cost rockets—whether it is SpaceX's Starship or the new-generation rockets from competitors like Rocket Lab and Stoke Space.
Previously, a research project funded by the U.S. CHIPS Act and NASA's In Space Production Applications has also been studying semiconductor manufacturing in space. The principal investigator of this project is Curtis Hill, a senior materials engineer at NASA, who leads a team in leveraging the weightless environment of space to simplify the etching steps in the semiconductor manufacturing process.
△ A researcher setting up experimental equipment in space to understand the effects of microgravity on brazing alloys.
The president of Japanese wafer foundry Rapidus recently stated in a speech in the U.S. that they plan to build a semiconductor factory on the lunar surface around 2040.
If companies like Besxar are focusing on "manufacturing wafers in space," then Elon Musk's Terafab plan, released in March 2026, targets deploying computing power in space.
Terafab is jointly built by Tesla, SpaceX, and xAI. Described by Tesla as "the largest chip manufacturing factory ever," its goal is to produce 1 terawatt of AI computing chips annually, equivalent to 50 times the current global annual production of AI computing power. The project will break the existing global division of labor model in chip manufacturing, concentrating the entire chain of lithography masks, chip manufacturing, and packaging and testing within a single facility, achieving an ultra-fast iterative closed loop of "mask making—chip manufacturing—testing—mask optimization—remanufacturing."
Musk explicitly stated that the layout focus of Terafab is placed on space, fundamentally stemming from the natural limitations of Earth's energy and computing power. The space-optimized high-power custom chips it produces will be specifically adapted to the extreme space environment and deployed in SpaceX's orbital AI data center network.
Editor's Note:
Manufacturing chips in space is indeed a good idea. However, there are still many critical issues that need to be resolved.
First, there are currently no equipment suitable for manufacturing semiconductors in space. If existing semiconductor equipment is launched directly into space, it would not only be highly costly but also not necessarily meet the requirements. Additionally, equipment installation is a formidable task. For example, the most critical lithography machines required for semiconductor manufacturing are inherently massive, weighing tens of tons, and take months to install and debug on the ground. Unless equipment manufacturers like ASML customize space-compatible versions, which would also incur extremely high costs.
However, teams are currently advancing the verification of compact space manufacturing equipment. For instance, Nitride Global, together with United Semiconductors and Axiom Space, was selected for NASA's SBIR program to develop a physical vapor deposition reactor for growing aluminum nitride (AlN) crystals in a microgravity environment. Its proof-of-concept reactor consumes only 250-400 watts of power, weighs less than 700 grams, can operate at high temperatures of 2800-3200°C, and meets the integration constraints of the mid-deck stowage lockers on the International Space Station.
Additionally, the compound semiconductor material growth focused on by Besxar is inherently more suitable for miniaturization than the lithography process for logic chips. Of course, to realize a complete "space fab," equipment adaptation for processes such as deposition, etching, and metrology remains a massive engineering challenge.
Second, even when the production lines for space-based semiconductor manufacturing are ready, they will still face the same awkward situation: whether the raw materials required for semiconductor manufacturing can be supplied continuously and stably. If relying solely on ground launches for supply, both costs and logistics time will be problematic, considering that semiconductor production lines require 24/7 uninterrupted operation.
Furthermore, after the products are manufactured, they must queue for return capsule space. The Dragon cargo spacecraft can only bring back approximately 3,000 kg of downmass per trip, and the schedule is determined by the mission rotation of the International Space Station, meaning manufacturers cannot "call a ride" according to their own production pace. Industry analysis shows that the bottleneck in return logistics has already "reshaped the economics of the entire emerging industry." The return services market size in 2026 is approximately USD 1.5 billion to USD 1.8 billion, and is expected to grow to USD 3.5 billion by 2030.
Third, space radiation is also an issue that must be addressed. High-energy particles in space can alter the crystal lattice structure of chips, leaving "vacancy defects" or "interstitial defects," leading to performance degradation or even failure.
However, studies indicate that for a 200mm wafer, the effective radiation damage received under no-electric-field conditions is less than 10 rads, which can be repaired through a simple annealing process. More compelling counter-evidence comes from Google's Project Suncatcher experiment: its Trillium AI chip successfully survived testing equivalent to the radiation dose of a 5-year low Earth orbit mission. This means that the threat of radiation to "wafers under manufacturing" may be far less than the threat to the "long-term operation of finished chips." The core challenges of space manufacturing remain focused on thermal management and logistics, rather than radiation.
Fourth, the ultimate form of space-based semiconductor manufacturing may not be "transporting back to Earth" but "using on-site," meaning "space factories" directly serve "space infrastructure," forming a closed loop independent of Earth.
Although Musk's Terafab plan is primarily ground-based, its product positioning explicitly targets SpaceX's orbital AI data center network. If orbital data centers, satellite constellations, and space manufacturing platforms themselves achieve scale in the future, then chips, motherboards, and cooling systems manufactured on-orbit can be assembled and deployed on-site, completely bypassing the bottleneck of "downmass transportation." Of course, this requires supporting capabilities in space chip manufacturing, motherboard manufacturing, and packaging and testing to also go to space simultaneously, with a difficulty no less than chip manufacturing itself. But at least logically, this is more self-consistent than "manufacturing and then transporting back to Earth for packaging and testing."
Besxar's current positioning is to supply wafers to ground-based chip manufacturers, which is still the path of "manufactured in space, used on Earth." But a true closed loop requires the space-ification of the entire industry chain.
In summary, the logic of space-based semiconductor manufacturing holds, but its bottleneck lies not in "whether it can be manufactured," but in the "need for supporting space manufacturing equipment," stable material supply, and competitive cost-effectiveness. Therefore, the most viable path in the short term remains the growth of high-value, small-volume compound semiconductor materials, rather than complete logic chip manufacturing.
Editor: Xinzhixun - Langkejian