Moore’s Law, once the cornerstone of semiconductor progress, has entered a period of serious scrutiny. The doubling of transistors every two years, once a given, is now a benchmark that demands increasingly complicated and costly technological feats. In this era of blurred boundaries between hardware, software, and materials science, Erik Hosler, a strategic consultant who highlights the importance of technological convergence, reflects a growing recognition that no single solution will extend Moore’s Law’s legacy.
The conversation has shifted from how long Moore’s Law will hold to how creatively it can develop. With physical scaling reaching atomic limits, researchers and engineers are turning to innovation as the last and best hope. However, innovation itself is multifaceted, relying on alignment between various scientific domains. The question is no longer just whether innovation can save Moore’s Law but whether the kind of innovation needed is even possible in isolation.
Scaling Is No Longer Straightforward
Lithographic scaling once delivered chip improvements like clockwork. Transistor sizes dropped, power efficiency rose, and computing performance surged. But now, at the edge of 3nm and beyond, the laws of physics push back. Material limits, quantum tunneling, and heat dissipation all function as barriers that no amount of fine-tuning can fully overcome.
Today’s advances require more shrinkage. They demand novel approaches to architecture, materials, manufacturing, and even chip design. It has made innovation not only necessary but also inherently more complex.
The Myth of the Lone Breakthrough
There’s a common belief in tech circles that a single “eureka” innovation might reinvigorate Moore’s Law. But this overlooks the ecosystem-wide nature of semiconductor development. No solitary breakthrough in etching, lithography, or materials can carry the load anymore. Progress depends on the convergence of advances across many fronts.
This multi-sector challenge is exactly why Erik Hosler emphasizes, “It’s going to involve innovation across multiple different sectors.” This quote, delivered during the SPIE Advanced Lithography symposium, frames the issue succinctly: siloed innovation won’t suffice. What’s needed is systemic innovation collaboration that spans industry, academia, and design disciplines.
Collaboration Over Competition
In the past, chipmakers raced to outdo one another in feature size and transistor density. Today, leaders like ASML and Lam Research are joining forces to develop solutions for 3D stacking and Extreme Ultraviolet (EUV) lithography. These partnerships represent a new model of progress, one where shared knowledge outpaces proprietary advantage.
ASML’s roadmap includes higher numerical aperture scanners, while Lam focuses on enabling etch precision in nanoscale layers. Together, they navigate problems neither could solve alone. This cooperative approach doesn’t just foster innovation. It enables it.
New Materials, New Momentum
Materials science has become a linchpin in scaling efforts. Traditional photoresists and silicon substrates face increasing limitations as feature sizes approach the atomic scale. In response, companies like FUJIFILM and IMEC are developing advanced resists and alternative materials to enhance pattern fidelity and reduce stochastic defects.
The success of these innovations, however, depends on how well they integrate with existing infrastructure. It highlights a recurring truth: innovation rarely arrives as a plug-and-play solution. It must be translated, validated, and scaled often with help from multiple sectors.
AI, Automation, and Algorithmic Design
Artificial intelligence and machine learning play a rising role in process optimization. From defect detection to predictive modeling, AI helps manufacturers fine-tune complex systems. But even this promising field is not a silver bullet. It relies on massive datasets, integrated sensors, and real-time feedback mechanisms.
For AI to effectively support the future of chip design, it must be embedded into the semiconductor lifecycle, from layout planning to quality control. Again, this means attracting innovation from sectors as diverse as software engineering, systems modeling, and applied physics.
Redefining Success: Beyond Transistor Counts
Moore’s Law was originally a statement about economics and density. But today’s innovation landscape is pushing the industry to redefine what success looks like. Instead of smaller transistors, the goal may be reduced energy consumption, increased reliability, or better real-time processing.
Chiplets, for example, offer modular design benefits without demanding smaller transistors. Likewise, photonic interconnects promise massive bandwidth increases without scaling traditional copper traces. These solutions don’t extend Moore’s Law in the classical sense, but they do achieve its spirit: delivering more with less.
The User Experience as the Ultimate Benchmark
The value of a semiconductor innovation is judged by what users experience. Whether powering a smartphone, an autonomous vehicle, or a quantum simulator, chips must deliver responsiveness, power, and connectivity. It places less emphasis on transistor density and more on application-centric performance.
Technologies like MEMS, neuromorphic computing, and heterogeneous integration aren’t merely academic; they enable real-world capabilities that reflect Moore’s original goals. In this context, innovation becomes more about impact than scale.
Economic Pressure and the Cost of Innovation
Innovation isn’t just a technical challenge; it’s a financial one. The cost of producing innovative semiconductors has skyrocketed, with top-of-the-line fabs requiring tens of billions of dollars in capital investment. It limits participation to only a few players globally, which in turn narrows the diversity of innovation sources.
Even within those firms, resource allocation must be strategic. Betting on the wrong architecture, material, or toolchain can result in years of sunk costs. In this context, innovation becomes as much about risk management as scientific discovery. That’s why cross-industry partnerships aren’t just helpful. They’re vital to sharing both risk and reward.
Case in Point: How TSMC Embodies Multi-Sector Innovation
TSMC’s recent strategies exemplify the broad innovation needed to sustain momentum. The company has made significant advances not only in node shrinkage but also in chiplet architecture, 3D integration, and foundry-customer collaboration. By collaborating closely with firms like Apple and AMD, TSMC ensures that design meets manufacturability at every step.
Its exploration of new packaging methods and co-optimization of software and hardware stacks shows that innovation isn’t isolated from fabrication. It spreads throughout the supply chain. TSMC’s approach reflects the reality that Moore’s Law lives or dies based on ecosystem thinking, not engineering brilliance alone.
Integrating Innovation: The Path Beyond Scaling
A breakthrough in isolation won’t define the future of Moore’s Law. It will depend on how well we integrate different breakthroughs, linking insights from chemistry, AI, lithography, photonics, and beyond. The innovation that sustains Moore’s Law must itself change, becoming more agile, adaptive, and collaborative.
We may be nearing the end of classical scaling, but that doesn’t mean we’ve reached the end of progress. With the right mix of creativity, coordination, and curiosity, the next chapter in semiconductor development could be even more transformative than the last.
