Summary

Nanomechanical computing (NMC) is computing going back to it’s roots, using Charle’s Babbage’s Difference Engine as inspiration, instead of using the flow of electrons around a circuit to perform binary operations, these computers use acoustic vibrations instead. Electronic computers based on CMOS technology have a minimum supply voltage of around 30 attojoules for error-free computation (known as Boltzmann’s tyranny). This is a factor of 104 higher than the fundamental Landauer limit, the minimum amount of energy needed for a computer to perform a computation. Nanomechanical computers do not have the same constraints and so can be order of magnitude more efficient than any electronic computer. The low-energy dissipation features of NMC makes them interesting for Reversible Computing and Quantum Hardware applications.

Viability (3)

A nanomechanical gate was invented in 2010 and attempts have been ongoing since to scale the gate up and build full integrated circuits. Different materials have been proposed to build the circuits to find the right balance between operation speed and energy required including nano-electromechanical systems (MEMS/NEMS), carbon nanotubes, Nano-RAM (NRAM) and molecules. The biggest restraint to scalable chips to date has been the reliance on electronics to interconnect gates. In 2022 an all nanomechanical architecture was demonstrated using a mechanical resonator to define logical states coupled into and out of the gate via nanomechanical waveguides. The approach is CMOS compatible facilitating both miniaturisation and scaling to complex architectures,

Drivers (2)

The main drivers for NMC is not to replace CMOS computers but to explore new approaches for extreme low-power, high temperature and high radiation environments. The development of nanoelectronics and nanotechnology tools for semiconductor fabrication has made it economically viable for researchers to experiment with the nanoscopic scale. The fragility of modern electronics in satellites to electromagnetic shocks such as solar flares or high altitude electromagnetic pulses (EMP) has led to increased attention on non-electronic computing.

Novelty (5)

Relative to CMOS-based electronic computers, nanomechanical computers are superior for applications that demand robustness, low power consumption and high-temperature. Operating speeds are order of magnitude lower so they are unlikely to ever be appropriate for high performance or even medium performance computing but are likely to find a niche where electronics are ill suited like high temperature and high radiation environments like space or deep underground.

Diffusion (2)

Like all other unconventional computing approaches, adoption will be slow. NMC will be challenging to design and fabricate simply because the $550 billion semiconductor industry is designed entirely around electrical components with a little photonics. Even if chips use silicon, new tools, new expertise and new supply chains are required to build mechanical components. The sub 1nm roadmap runs to 2036 with investment taking place a decade before commercialisation. Even with strong commercial demand from the space economy, it nanomechanical devices will take more than a decade to reach customers.

Impact (4) Low certainty

The high impact scenario is that nanomechanical gates and ICs enable Reversible Computing and error-correcting Quantum Hardware as a function of the operating close to the Landauer limit. High performance will always be the primary driver for the computing industry and extreme low-power computers are likely to always remain closer to a niche, at least on earth where energy is relatively cheap. The space economy is estimated to reach $1 trillion in 2040 and the biggest impact of NMC will be in space, eventually as the dominant computer for satellite's and other space-based devices as radiation protection becomes a high priority.

Timing (2030+) High certainty

High certainty that we will not see any nanomechanical chips in market before 2030 and maybe even 2035. Over the next 5 years we can expect further scaling up of nanomechanical ICs and the experimentation with different materials most suitable for commercially viable chips. We are likely to see a nanomechanical device in space before 2030 to better probe performance.