Introduction

If you’ve been following tech news this year, you might have noticed something unusual happening. The U.S. government—which often moves slowly—is making billion-dollar bets on technologies that sound like science fiction. And venture capital firms that built their reputations on software investments are suddenly pouring money into companies that build physical machines you can actually touch .

This isn’t random. Something real is shifting in American technology policy and investment. In July 2026, the Commerce Department announced it would invest $874 million in semiconductor research targeting the bottlenecks holding back AI progress . In August, the White House released a new National Security Science and Technology Strategy that treats technological advantage as central to national security . And in the first half of 2026 alone, venture capitalists poured $47.4 billion into what they’re calling “Physical AI”—machines that can operate in the real world .

What’s different this time isn’t just the scale. It’s the coordination. The U.S. isn’t simply funding research and hoping for the best. It’s building the physical, intellectual, and institutional infrastructure to maintain technological leadership for decades to come..

The $874 Million Bet on AI Computing Bottlenecks

On July 29, 2026, the Commerce Department signed letters of intent with seven companies to provide up to $874 million in federal incentives under the CHIPS and Science Act .

Here’s the important detail: This funding is not for building more chip factories. That’s a different part of the CHIPS Act. This $874 million is specifically for research and development—targeting the specific bottlenecks that are slowing down AI progress .

The Three Largest Investments

GlobalFoundries – Up to $300 million for co-packaged optics

This is the largest single award. The funding aims to accelerate the development of silicon photonics technology—using light instead of electricity to move data between chips .

Why does this matter? As AI models get bigger, the biggest bottleneck isn’t always computation—it’s moving data from one chip to another. Copper interconnects have limits. They generate heat, consume power, and can’t keep up with growing bandwidth needs. Silicon photonics places optical components right next to AI processors, shortening the electrical signal path and improving both speed and energy efficiency .

The Commerce Department believes this investment could advance U.S. leadership in AI infrastructure by two to three years .

Kepler Computing – Up to $245 million for ferroelectric memory

This is the second-largest award. Kepler is developing a new class of AI memory that combines 3D integration with ferroelectric technology—a material that retains data without continuous power .

The real advantage is architectural. Traditional memory requires moving data across a high-bandwidth bus, which creates bottlenecks. Kepler’s approach allows certain operations to happen directly on the memory chip, reducing data movement and improving performance .

Multibeam Corporation – Up to $140 million for advanced packaging

Multibeam is developing technology that assembles and stacks multiple chips with thousands of interconnections—crucial for “Chiplet” designs where chips are built as smaller components and combined into a single package .

What makes Multibeam interesting is its technology: multi-electron beam direct-write lithography. Instead of using traditional photomasks, they pattern chips directly using multiple micro electron beams .

The Other Investments

• Extropic – $75 million for thermodynamic sampling units that use natural thermal fluctuations to solve complex problems with less energy
• Thintronics – $50 million for ultra-low-loss dielectric materials for next-generation interconnects
• OBSIDIA Semiconductors – $34 million for counterfeit detection technology to secure AI supply chains
• Aeluma – $30 million for substrate technology for photodetectors and lasers used in AI optical interconnects 

One Unusual Detail: Equity Stakes

The Commerce Department will receive minority, non-controlling equity stakes in each company as part of the funding agreements . That means U.S. taxpayers could see returns if these technologies succeed—a departure from conventional one-way subsidies. This approach has been described as a new “government as investor” model, also seen in recent quantum computing and critical minerals investments .

The National Security Science and Technology Strategy

In August 2026, the White House unveiled a new National Security Science and Technology Strategy that provides the framework for this investment .

The strategy organizes the U.S. approach around four priorities:

1. Focused – Directing technology competition toward areas where the U.S. has advantages
2. Resilient – Reducing vulnerabilities in critical supply chains
3. Agile – Accelerating innovation by removing regulatory hurdles
4. Secure – Preventing foreign exploitation of U.S. intellectual property 

Priority Technology Areas

The strategy identifies specific technology areas that matter most: AI and autonomy, space, undersea systems, advanced manufacturing, semiconductors, and nuclear energy .

It also highlights several key enabling areas: communications and networking, directed energy, future computing, hypersonics, cybersecurity, and sensing technologies .

In military terms, the strategy prioritizes submarine and anti-submarine warfare, space capabilities from low Earth orbit to cislunar space, and AI for military decision-making. It calls for a mix of high-end platforms and lower-cost autonomous systems to create asymmetrical costs for adversaries .

A Shift on Talent

One notable change in the new strategy: It states that the U.S. “will further strengthen its workforce by attracting and retaining top-tier global talent in critical national security S&T fields” .

This is a departure from the previous year’s National Security Strategy, which suggested that global talent “undercuts American workers.” The change reflects a growing recognition that the U.S. cannot maintain its technological edge without access to the world’s best minds .

The strategy also calls for expanded STEM education, apprenticeships, scholarships, and stronger pathways for researchers to commercialize intellectual property .

Physical AI: The Investment Boom

While government funding grabs headlines, private capital is moving in the same direction. In the first half of 2026, global venture funding in Physical AI—machines that can perceive, interpret, and interact with the physical world using AI—totaled $47.4 billion across 521 deals .

Several megadeals drove this:

• Waymo raised a $16 billion Series D, valuing the self-driving car unit at $126 billion
• Anduril Industries raised $5 billion at a $61 billion valuation
• Shield AI, a drone developer, secured a $2 billion Series G at a $12.7 billion valuation 

Why Hardware Is Becoming a Moat

In the software-as-a-service era, startups could scale quickly with minimal capital. Today’s Physical AI companies are different. They are hybrid businesses that integrate advanced AI with bespoke hardware—warehouse robots, autonomous vehicles, intelligent manufacturing lines, and energy systems .

Hardware, once considered a venture risk, is increasingly viewed as a competitive advantage. As one investor put it, the new investment thesis is less about choosing between hardware and software than about owning the intersection of both .

The Strategic Stakes

The push for Physical AI extends beyond commercial returns. As Emma Norchet, lead private technology investor at T. Rowe Price, observed: “Nations that automate production domestically can control costs and supply chains, while those that cannot risk structural dependence on foreign manufacturing” .

The economic pressure is acute. Industries facing labor shortages—manufacturing, logistics, construction, energy, and mining—are early adopters .

SandboxAQ’s $500 Million Award

A separate but related development: In June 2026, the Commerce Department signed a definitive agreement with SandboxAQ for a $500 million award under the CHIPS Act .

SandboxAQ’s platform uses AI to accelerate materials discovery for critical semiconductor bottlenecks, including alternatives to PFAS “forever chemicals,” advanced catalysts, rare earth-free magnets, and novel battery chemistries .

The Commerce Department will also receive a minority, non-controlling equity stake in SandboxAQ as part of the agreement .

Conclusion

What’s emerging across these developments is a pattern of deliberate, coordinated action. The White House strategy provides the framework. The CHIPS Act’s R&D investments target specific technical bottlenecks. Private capital is flowing in the same direction.

What makes this moment different from previous technology waves is the scale of public-private coordination—and the willingness to take equity stakes rather than just hand out subsidies. The U.S. is not simply funding research and hoping for the best. It is building the infrastructure—physical, intellectual, and institutional—to support technology leadership for decades to come.

The question is whether this strategy will succeed. The early signs suggest a deliberate, if challenging, effort to maintain American technological leadership in an era of intensifying global competition.