Is Moore’s Law Still Applicable?

For decades, Moore’s Law has driven the semiconductor industry forward. It predicted that transistor density on an integrated circuit (IC) would double approximately every two years while costs remained constant. This rapid advancement has powered everything from personal computing to AI-driven supercomputers.

However, as we approach sub-2nm technology, physics, economics and manufacturing complexity threaten the traditional interpretation of Moore’s Law. Is this the end of transistor scaling, or is the industry adapting in new ways? In this blog, we examine whether Moore’s Law is still relevant, the roadblocks ahead, and the innovations that will sustain semiconductor progress.

1. Why Moore’s Law is Slowing Down

A. Physical Barriers: The Limits of Shrinking Transistors

The persistent shrinking of transistors has brought us to a point where quantum mechanical effects begin to interfere with chip performance.

One major issue is quantum tunneling. As transistor gates become smaller than 2nm, electrons can unpredictably pass through barriers, leading to power leakage and inefficiency. Additionally, short-channel effects make it difficult for transistors to fully switch between “on” and “off” states, further limiting performance improvements.

To counter these problems, the semiconductor industry has moved beyond traditional FinFET (Fin Field Effect Transistor) technology, which has been in use since Intel's 22nm node (2011), and is now transitioning to more advanced Gate-All-Around (GAAFET) transistors, which offer better electrostatic control.

B. The Cost of Advanced Manufacturing

Even though transistor scaling has continued, the cost per transistor is no longer decreasing as it did in the past. The introduction of Extreme Ultraviolet (EUV) lithography—a critical technology for fabricating sub-5nm chips—has significantly increased the cost of semiconductor manufacturing.

• ASML’s EUV machines cost over $300 million each, making it challenging for smaller semiconductor companies to compete.

• Fabrication plants for sub-3nm nodes now require more than $20 billion in investment.

• Companies like TSMC, Intel, and Samsung are the only ones currently capable of pushing into 2nm and beyond, while others rely on older nodes for cost-effectiveness.

As a result, only the biggest players can afford cutting-edge chips while others opt for more cost-efficient solutions like chiplets.

2. The Future: Extending Moore’s Law with New Technologies

Even though classical transistor scaling is slowing down, the industry is finding new methods to continue performance improvements. These include new transistor designs, 3D stacking, chiplets and alternative materials. 

A.    Transistor Innovations: FinFET → GAAFET → 3D Stacking

To combat the limitations of FinFET transistors, GAAFET (Gate-All-Around Field Effect Transistors) are being introduced. Unlike FinFETs, which have a single gate wrapping around three sides of the channel, GAAFETs completely surround the channel, improving efficiency and reducing leakage.

• Samsung’s 3nm GAAFET (2022) was the first to enter mass production, offering 45% reduced power usage, 23% improved performance and 16% smaller surface area compared to 5nm  FinFETs.

• Intel’s RibbonFET (2nm, expected in 2025) takes GAAFET a step further by stacking multiple layers of gates, improving transistor density.

Beyond GAAFET, the next big leap is 3D stacking, where transistors are placed on top of each other rather than just side-by-side. This method, demonstrated by IBM’s 2nm prototype in 2021, improves performance by 45% while reducing power consumption by 75% compared to 7nm FinFETs.

B. Chiplets & Advanced Packaging: The True Moore’s Law Successor?

Instead of making monolithic chips larger and more complex, companies are now breaking them into smaller chiplets and using advanced packaging to integrate them efficiently.

• AMD’s EPYC Genoa & Intel’s Meteor Lake leverage chiplets to improve performance without relying on node scaling alone.

• TSMC’s CoWoS (Chip-on-Wafer-on-Substrate) and Intel’s Foveros 3D stacking enable better interconnects between different chiplets.

One of the biggest adopters of chiplets is NVIDIA, whose H100 AI GPU uses multiple dies and high-bandwidth memory to achieve superior performance. Chiplets are expected to dominate future semiconductor design, reducing reliance on extreme transistor scaling.

C. Beyond Silicon: 2D Materials & Carbon Nanotubes

With silicon-based transistors approaching their limits, researchers are investigating new materials that can replace or complement silicon in future chips.

• Graphene & Molybdenum Disulfide (MoS₂) – These 2D materials offer higher conductivity and better performance at sub-1nm scales.

• Carbon Nanotube Transistors (CNTs) – IBM & Stanford are experimenting with CNTs, which have the potential to increase performance 10x over silicon transistors.

  
  

While these technologies are still in early research stages, they could revolutionize computing within the next 10-15 years, enabling transistor scaling beyond silicon’s physical limitations.

3. Is Moore’s Law Dead or Just Changing?

While traditional transistor shrinking is slowing, computing power is still growing through:

• GAAFETs & 3D stacking to improve transistor density.

• Chiplets & advanced packaging instead of monolithic scaling.

• AI & specialized accelerators (like Google’s TPU, NVIDIA’s AI GPUs) are reducing reliance on Moore’s Law.

• Quantum computing & neuromorphic chips as long-term disruptors of classical computing.

Final Verdict: Moore’s Law is evolving into “More Than Moore”

The classical interpretation of Moore’s Law—shrinking transistors for higher performance at lower cost—is no longer valid in the same way. However, the spirit of Moore’s Law continues through alternative innovations, including chiplets, 3D stacking and AI-driven architectures.

The semiconductor industry is moving toward a future where performance scaling comes from smarter chip architectures rather than just making transistors smaller. The next decade will be one of revolutionary computing models, where the biggest advancements may no longer come from Moore’s Law, but from entirely new ways of building processors.

At McKinsey Electronics, we stay ahead of these industry shifts by providing cutting-edge semiconductor solutions and expert circuit design advisory to the GCC, Africa and Türkiye. Whether you're looking for the latest chip technologies or strategic guidance on next-generation architectures, we help businesses navigate the evolving semiconductor landscape with precision and expertise.