In times when the gaps between nodes from TSMC or Samsung are growing larger and the performance or efficiency gains are becoming increasingly marginal, many are not asking whether quantum computers will replace the dying Moore’s Law but only when. But is this technically possible to entangle enough qubits to satisfy our constant hunger for computing power?
In 2023, the leading Taiwanese semiconductor producer TSMC rolled out the 3nm node, which is used, among other things, in the A16 Bionic Chip of the iPhone 15. This new structure density allows for a 1.6x increase in logic density and a 30-35% power reduction. While this may sound remarkable, it represents a gap of three years between this structure density and the 5nm structure density. The common consensus of Moore’s Law states that computing power or efficiency should double every 24 months. It should be clear by now that in the future, it will no longer be cost-effective to continuously shrink transistors. In the realm of structure reduction, only TSMC and Samsung continue to play a significant role.
Our deterministic computing is inevitably reaching a fundamental limit because some components on our chips are now only a few silicon atoms in size.
But what does the future look like?
Quantum computing is still in an early and experimental stage. To perform a quantum search, like the Grover algorithm, which generates quadratic acceleration, one can double the performance with each entangled qubit. However, with each qubit, one also encounters a more unstable system, as the trapped ions are very sensitive to environmental influences.
So, will we be able to entangle thousands, if not millions, of qubits in the near future, stable and error-free? Probably yes, but just as the performance doubles with each qubit, the complexity of executing this system stably and error-free also increases.
When looking back in time, we can draw parallels to the challenges Intel faced in developing the first microprocessor. In 1971, Italian engineer Federico Faggin, American engineers Marcian Hoff and Stanley Mazor, and Japanese engineer Masatoshi Shima pushed the limits of what was technically possible to bring the first commercial microprocessor to market. At that time, it seemed inconceivable, even physically impossible, to one day pack billions of microscopic transistors onto a chip and produce them in large quantities at an affordable price. Only time will tell by the end of this decade whether and when a similar development will occur in quantum computing.