"electron beam direct write" for Innovator Small Run?
Innovators Need a Foundry Too: Fixing the U.S. Semiconductor Bottleneck
Pete Singer
4 days ago
Lex Keen, Founder and CEO of SecureFoundry
There’s no shortage of significant investments in American semiconductor manufacturing right now. Yet, while the U.S. races to rebuild its semiconductor leadership, one critical truth keeps getting lost: we can’t outpace global competition if the only companies with access to chip manufacturing are the ones that already won. Today’s system is built for scale, but innovation doesn’t start at scale. It begins at the edge, in startups, national labs, and research programs that need only a few wafers to prove a concept or advance a defense capability. Yet, those innovators face a near-impossible barrier to entry: access to domestic, flexible-volume semiconductor manufacturing. The reason is simple: the U.S. chip ecosystem overwhelmingly serves volume, not vision.
Today’s dominant foundries are optimized for massive throughput. To make the economics work, they need large volume commitments, expensive mask sets, and long lead times. For many innovators, that model is simply out of reach. It’s not because their ideas aren’t feasible; it’s because the infrastructure isn’t built for them. That’s the bottleneck. It’s not just one of supply or capacity. It’s institutional.
I’ve seen firsthand the institutional void between a promising design and a finished product. The current ecosystem is so consolidated and optimized for volume that big tech companies require committed production runs, often in the hundreds of thousands, even to entertain a project. That works fine for smartphones. But it doesn’t work for defense systems needing a single validated component or startups building AI chips to optimize power efficiency rather than market share.
And then there’s the cost. Traditional semiconductor manufacturing still relies on photomasks, custom stencils used to pattern designs onto silicon wafers. A single advanced mask set comes with a hefty price tag and takes six to twelve months to fabricate. For small companies or low-volume production needs, that’s a non-starter.
Maskless lithography using electron beam direct write technology offers one way around this. Instead of using masks, electron beams “draw” the design directly onto the wafer. For decades, single-beam systems were too slow to be practical. However, with advances in multi-beam electron tools, throughput is improving to a level where this approach becomes viable not just for R&D but for limited production runs and rapid prototyping.
The flexibility goes beyond cost and time. When you’re not tied to masks, you can make design changes on the fly, produce highly customized chips, and even embed unique identifiers or security features at the die level, capabilities that are increasingly important for both commercial and defense technology.
Beyond innovation, there’s another layer: obsolescence. Critical defense systems still rely on components fabricated at mature nodes, often 180nm or larger, that are no longer supported by mainstream fabs. It’s not that manufacturers don’t want to make these parts; in many cases, they simply can’t. The fabs and tooling is gone, and sometimes even the design files are missing. Reverse engineering becomes the only path to restoring designs, and that’s not just about building a chip; it’s about reconstructing the intent behind its design from fragments of documentation and aging test hardware.
That takes time, talent, and infrastructure we can’t afford to lose.
The CHIPS Act was passed to revitalize American semiconductor manufacturing, but in its current form, it’s missing this piece of the puzzle. To enable innovation, we need flexible-volume manufacturing, infrastructure, tools, and pathways for new and legacy technologies to move into production without massive overhead. The next big breakthrough isn’t waiting for more funding. It’s waiting for a way in the door.
PS
shills and "experts" call opportunities walls.
LOL. |
