Packy McCormick makes a startling claim: the invisible force that powers our entire digital and physical infrastructure is also the one thing human intuition has failed to master. While the world obsesses over large language models, he argues that the true bottleneck for the next century of progress isn't software, but our inability to "see" electromagnetic fields. This isn't just physics; it's an economic and strategic crisis where the "nervous system" of modern hardware is failing because we are trying to design it with blinders on.
The Invisible Bottleneck
McCormick opens by reframing electromagnetism not as a scientific curiosity, but as the silent operator of the global economy. He notes that "electrical and electromagnetic components are the 'nervous system' of modern hardware and contribute to 40-50% of failures." This statistic is jarring because it suggests that despite our reliance on chips and wireless networks, our fundamental grasp of the physics governing them is crumbling. The author argues that the nation's capacity to test and build these systems has declined, creating a dangerous gap between our ambition and our capability.
The core of his argument rests on the idea that electromagnetism is "black magic" to almost everyone. He writes, "There are maybe ten people in the world who can deeply intuit electromagnetism, who can see which shapes will create which EM fields in their mind's eye." This is a bold assertion, but it highlights a critical vulnerability: we are building increasingly complex systems with a vanishingly small pool of human experts. McCormick illustrates this by comparing human perception to gravity; we evolved to feel gravity, but we never evolved to "see" radio waves or microwaves. "Those of our ancestors' friends who wasted precious resources on seeing the full spectrum of EM waves wouldn't have lived to pass on these traits, either," he observes, explaining why our intuition is fundamentally mismatched with the technology we now rely on.
Humans can't intuit EM, and it's a bottleneck to the electric progress we both want to see.
The stakes are rising as the economy electrifies. McCormick points out that in 1970, electronics made up just 5% of a car's cost; by 2030, that figure is projected to hit 50%. In defense, the F-35 Lightning II already spends 35% of its cost on electronics, more than its engine. As the executive branch and defense contractors push toward next-generation platforms like the projected F-47, the reliance on electromagnetic design will only deepen. If we cannot scale the ability to design these systems, the entire electrification agenda stalls.
From Black Magic to Superintelligence
The solution McCormick proposes is as radical as the problem is dire: we must offload this cognitive burden to machines. He argues that while humans are blind to the full spectrum, artificial intelligence has no such evolutionary baggage. "Fortunately, AI doesn't share our blind spots. It is particularly good at seeing patterns, at making connections and understanding dependencies that are not necessarily intuitive to humans," he writes. This is the pivot from a hardware problem to a software one.
He introduces the concept of a "Large Field Model" (LFM), drawing a parallel to the large language models (LLMs) that have dominated recent discourse. Just as LLMs generalize across language, an LFM would generalize across electromagnetic fields. "We should be able to use this LFM to understand EM waves and shape them to do what we'd like them to do," McCormick explains. This approach aims to democratize the intuition that currently resides only in the minds of a few elite engineers. The historical context here is vital; just as the development of the Chain Home radar network during World War II required a massive leap in understanding how to manipulate radio waves to detect aircraft, today's challenges require a similar leap in computational intuition to manage the complexity of modern phased arrays and semiconductor design.
Critics might note that simulating quantum electrodynamics (QED) with the precision required for high-frequency engineering is a monumental computational challenge that current AI models may struggle to solve without massive energy costs. However, McCormick's argument gains strength when he notes that the alternative—relying on a shrinking pool of human experts—is not a viable path forward. The industry is already moving in this direction, with firms like Arena Physica deploying AI tools to debug hardware for major players like AMD and Anduril.
The New Frontier of Hardware
The article concludes by suggesting that the future of innovation lies in machines that can "see" the fields we cannot. McCormick writes, "This is an essay about how to teach machines to see the fields that we can't, and what the world might look like if we can." This vision extends beyond simple efficiency; it implies a new era of hardware design where the physical constraints of electromagnetism are navigated by algorithms rather than human trial and error. He references the skin effect, a phenomenon where high-frequency current flows only on the surface of a conductor, as a specific example of the complex, counterintuitive behaviors that machines could learn to optimize far better than humans.
The transition from mechanical switches to semiconductors was a leap in speed and reliability. McCormick suggests the next leap is from human intuition to machine intuition. "Electromagnetism secretly runs our world," he reminds us, but the secret is about to be solved by a new kind of intelligence. The implication is clear: the companies and nations that master this "Electromagnetic Superintelligence" first will define the next era of technological dominance.
Bottom Line
McCormick's most compelling insight is that the bottleneck for the future of hardware is not materials or manufacturing, but human cognitive limitation. By framing electromagnetism as an intuitive gap that AI can fill, he offers a fresh perspective on the AI race that moves beyond chatbots and into the physical world. The argument's greatest strength is its urgency, yet it leaves open the question of whether current AI architectures can truly replicate the deep physical intuition of a master engineer without a fundamental breakthrough in how models understand physics. Watch for the first commercial deployments of these Large Field Models; if they deliver on the promise of reducing hardware failure rates, the shift will be irreversible.