Pat Gelsinger wants to save Moore’s Law, with a little help from the Feds – RocketNews

A year after being pushed out of Intel, Pat Gelsinger is still up before dawn and still immersed in the semiconductor fight—just from a new command post. Now a general partner at venture firm Playground Global, he works closely with a slate of startups. One of them, xLight, has seized most of his attention after announcing a preliminary deal for up to $150 million from the U.S. Commerce Department—an arrangement that would also make the government a meaningful shareholder.

For Gelsinger, who spent 35 years across two stints at Intel before the board lost confidence in his turnaround plans, the deal is a public validation of a new thesis: that the most stubborn bottleneck in chipmaking—lithography—can be broken open with fresh physics and deep pockets. It’s also fueling wider debate in Silicon Valley about a subtle but significant shift in industrial policy: Washington taking equity stakes in strategically vital companies.

“What the hell happened to free enterprise?” California Governor Gavin Newsom asked this week, giving voice to the unease among those who prefer markets to handle capital allocation without government appearing on the cap table.

Gelsinger seems unbothered by that philosophical argument. As executive chairman of xLight, he’s betting that the startup’s approach—massive free-electron lasers powered by particle accelerators—can redefine how chips are patterned. Lithography, the painstaking process of etching microscopic circuits onto silicon wafers, has long set the pace (and limits) for Moore’s Law. Shrink the features, pack in more transistors, and you get faster, more efficient computing. Fail to advance lithography, and the cadence slows.

“I have this long-term mission to continue to see Moore’s Law in the semiconductor industry,” Gelsinger said. “We think this is the technology that will wake up Moore’s Law.”

Public money, public stakes

xLight’s preliminary award is an early, high-profile example of a broader push to rebuild U.S. chip leadership through the Chips and Science Act. While grants and loans are familiar tools, taking equity positions marks a different posture—one that blends national security priorities with the potential for financial upside if the technologies succeed.

Critics argue that government shareholding risks distorting incentives and picking winners. Supporters counter that semiconductor manufacturing is a strategic asset—too crucial to leave entirely to market cycles and offshore supply chains. Either way, the message is clear: the federal government is becoming a more active partner in the next generation of chip technologies.

Why lithography is the bottleneck

Modern chips demand unimaginably fine features, driving the industry to ever-shorter wavelengths of light and ever-more-precise optics. Today’s leading-edge factories rely on extreme ultraviolet systems measured in tens of tons and assembled from thousands of painstakingly engineered parts. They are technological marvels, but they are also expensive, scarce, and approaching practical limits in throughput and resolution.

xLight’s premise is to generate powerful, tunable light using free-electron lasers (FELs). In theory, FELs could deliver high-intensity, coherent light suitable for patterning at smaller scales, potentially improving throughput or enabling new process windows. The catch is profound: building particle-accelerator-class equipment that operates reliably on a fab floor, integrates with existing photoresist chemistries and masks, and runs cost-effectively, is an enormous engineering challenge.

If the physics scales, the payoff is huge

Should xLight’s technology prove viable at manufacturing scale, the implications could be far-reaching:

• Extend Moore’s Law: Smaller, cleaner features help pack more transistors into the same area, driving performance-per-watt gains.
• Lower cost per transistor: Higher throughput and yield improvements can bend cost curves back in the right direction.
• Resilience and capacity: New lithography options could diversify supply chains and reduce bottlenecks concentrated in a few tools or regions.

But there are equally substantial hurdles:

• System complexity: Particle accelerators and FELs must be tamed into industrial equipment with uptime measured in the 90th percentile.
• Process integration: Photoresists, masks, and metrology must all align to new light sources and exposure conditions.
• Power and footprint: Any solution must fit into fabs without blowing up energy budgets or floor plans.
• Economics: Even if it works, total cost of ownership has to beat—or at least match—the incumbent path.

Gelsinger’s second act

Gelsinger’s pivot from running one of the world’s largest chip companies to backing frontier hardware may look abrupt, but it tracks with his long-standing belief that the industry can and must keep pushing physical limits. As a general partner at Playground Global, he is working with roughly 10 startups across the deep-tech landscape, with xLight as the standout swing.

His role blends operator grit and investor patience: the early mornings, the systems-level skepticism, and the drive to get from lab demonstrations to production reality. If xLight’s approach matures, it could become the kind of platform technology that resets the trajectory of the entire sector. If it stalls, the lessons will still inform the next wave of attempts to crack the lithography problem.

The bigger picture

The pairing of bold physics and government backing is becoming a defining feature of the chip renaissance. The United States wants leading-edge manufacturing and the supply-chain security that comes with it; entrepreneurs want to compress decades of R&D into a few ferocious years. Those goals are aligned—for now.

Ultimately, the market will judge xLight on performance, reliability, and cost. The coming milestones are predictable: demonstration systems, pilot lines, customer trials. If the startup can clear those hurdles, the payoff could be transformative—not just for one company or one investor, but for the arc of Moore’s Law itself.