Transistor scaling no longer defines the full trajectory of computing progress. As chipmakers face limits in size, cost, and complexity, they are exploring new paths to improve performance. Erik Hosler, a consultant known for his work in semiconductor patterning and next-generation device strategies, points to a shift already underway. Performance gains now depend on integrating technologies that operate outside the boundaries of logic transistors.
In this landscape, components like MEMS and photonics are no longer peripheral. They are becoming central to how chips deliver speed, efficiency, and responsiveness. As traditional node scaling slows, the industry’s focus is turning toward how physical sensing, optical communication, and domain-specific functions can be layered into systems. This approach defines the post-node era as a continuation of Moore’s Law through new forms of integration.
Table of Contents
Beyond Transistor Counts
For decades, the industry thrived on a simple formula: halve the feature size, double the transistor count, and reap performance gains. However, sub-7nm nodes have introduced challenges that make further scaling increasingly expensive, complex, and inefficient. Issues like heat dissipation, quantum tunneling, and patterning variation make it clear that traditional scaling is unsustainable.
As a result, chipmakers are rethinking what “progress” means. Instead of counting transistors, the focus is shifting to system-level gains: faster data transfer, greater energy efficiency, smaller form factors, and enhanced functionality. It opens the door for new classes of technology to contribute to performance without shrinking features.
Introducing the Post-Node Era
The post-node era does not abandon scaling but expands how we pursue it. It acknowledges that improvements will come from integration, not just miniaturization. Chiplets, 3D stacking, heterogeneous packaging, and domain-specific accelerators all represent new forms of advancement.
MEMS and photonics introduce capabilities that logic transistors alone cannot provide. MEMS enables mechanical sensing and actuation at miniature scales, while photonics supports high-speed, low-power data movement. Integrating these technologies expands the functional range of chip architectures and supports new approaches to performance, efficiency, and adaptability.
What MEMS Bring to the Table
Microelectromechanical systems are already well-known in certain applications, including accelerometers in smartphones, gyroscopes in drones, and pressure sensors in automotive systems. But their role is expanding in advanced computing. MEMS can serve as tunable filters, actuators, timing devices, and environmental sensors embedded directly within chips.
By integrating MEMS into semiconductor devices, designers gain new levers for system optimization. MEMS can dynamically adjust system parameters, enable sensing at the edge, or provide compact mechanical functions that would otherwise require external modules. As edge devices become smarter and more responsive, MEMS will become essential to balancing performance with miniaturization.
Photonics: Lighting the Way Forward
Photonics offers another change in basic assumptions. As electrical interconnects approach fundamental speed and power limits, transmitting data using light becomes more attractive. Silicon photonics allows optical components to be manufactured using conventional semiconductor processes, making integration with CMOS more feasible.
The advantage of photonics lies in bandwidth and efficiency. Optical interconnects can transmit terabits per second with minimal loss and heat generation, which is vital for high-performance computing and data center applications. On-chip photonic links reduce latency between processing cores and memory, improving overall throughput.
Converging Domains for Smarter Systems
MEMS and photonics become most effective when they are integrated directly into chip architecture rather than added as separate components. In modern systems, data must move fast, adapt intelligently, and interface seamlessly with the physical environment. MEMS and photonics bring those capabilities into the chip domain.
This convergence is redefining what a chip can be. It is no longer a monolithic slab of transistors but a platform that blends logic, sensing, and communication. Systems-on-Chip (SoCs) increasingly resemble systems-on-multiple-domains, with electrical, mechanical, and optical elements working together.
By leveraging each domain’s strengths, the limitations of each one are overcome.
A Different Kind of Scaling
The post-node era also introduces a different mindset about scaling. Instead of making components smaller, designers make systems smarter. It includes:
- Functional scaling, where new capabilities are added (e.g., in-sensor computing)
- Vertical scaling, through 3D integration
- Energy scaling, by optimizing power use through domain-specific hardware
These approaches deliver the kinds of user-visible improvements, faster boot times, more responsive interfaces, and lower latency that Moore’s Law once guaranteed via node shrinkage. The outcome remains the same: better computing delivered more efficiently. But the route to get there is now multifaceted.
Strategic Implications for the Industry
This shift has implications far beyond fabrication. For chip companies, it means rethinking product roadmaps around integration rather than pure speed. For system designers, it offers flexibility to solve bottlenecks with the right technology mix. For startups, it opens space to specialize in emerging technologies that feed mainstream platforms.
In the post-node era, the value was not created solely by leading process nodes. It was created by orchestrating complementary technologies in a cohesive, scalable way. That is why the most forward-thinking companies are investing heavily in MEMS and photonics, not just to stay relevant but to redefine relevance itself.
Moore’s Law, Reimagined
The essence of Moore’s Law was never about transistor size. It was about delivering more capability, more efficiently, over time. That spirit is alive and well, but expressed differently today. It lives in the ability to create systems that are adaptive, multifunctional, and attuned to human needs.
This development was captured during a panel at the SPIE Advanced Lithography symposium. Erik Hosler highlights, “Finally, the solution to keeping Moore’s Law going may entail incorporating photonics, MEMS, and other new technologies into the toolkit.” His comment underscores a growing consensus: the tools we now use to advance computing must be broader and more interdisciplinary than ever before.
Reframing the Future
As the industry leaves behind the comfort of a single-node roadmap, it enters a more dynamic and demanding phase. Progress now comes from the interplay of technologies and from system-level thinking rather than from the performance of any one component.
MEMS and photonics are not just stopgaps or supplements. They are fundamental to a broader understanding of what it means to build faster, smarter, and more capable devices. They are proof that Moore’s Law is not obsolete; it has simply developed.
