To The Moon

$IONQ Inc.(IONQ)$  When you read press releases or news about quantum computing, you'd certainly come across the term "qubit".

- "Google unveils the 105 Qubit Willow Chip..."

- "Ionq achieves 36 Algorithmic Qubits..."

- "D-Wave: The Advantage system... with over 5,000 qubits..."

- "Rigetti launches 84-qubit Ankaa-3 quantum computer..."

In simple terms, a qubit is the basic unit of information that a quantum computer uses to perform calculations. It might make sense that the more qubits the more powerful quantum computers will be. However, that is simply not true since qubits are not the same nor perfect. In this post, I'll explain the differences in qubit technologies and what really matters beyond the qubit count. Hopefully, this will help you make better sense whether press releases or news coming from various companies are substantial or just fluff

But first about qubits...

3 THINGS TO KNOW ABOUT QUBITS

1. Qubits store information as a combination of 0 and 1 (Superposition)

Information in classical computers are stored in bits which can either represent 0 or 1. Qubits, on the other hand, can store a combination of 0s and 1s. This is due to one of the fundamental principles of quantum mechanics called Superposition. Think of it as a compass: a bit can represent either N(orth) or E(ast) while a qubit can represent NE or NNE or anything in between. This characteristic makes quantum computers more useful for certain types of calculations than classical computers. 

2. Qubits can connect with each other (Entanglement)

A pair of qubits can be linked together such that the state of one qubit can affect the other. This is called Entanglement. Once qubits are entangled, they stay connected no matter how far they are apart in space. This is incredibly useful in certain complex calculations where changing the state of one qubit instantaneously changes the state of the other.

3. Qubits are fragile (Decoherence)

Qubits are very suspectible to noise or disturbances in the environment. After a certain time, a qubit will lose its superposition and entanglement. This is called Decoherence. The result is errors introduced into calculations which isn't particularly useful.

All the above affects the performance of quantum computers. Therefore, almost every hardware companies' R&D efforts goes into:

1. Increasing qubit count (superposition)

2. Increasing qubit connectivity (entanglement)

3. Increasing qubit fidelity (quality) / lowering qubit errors (decoherence)

However, depending on the type of qubit technologies, the challenges can be quite different.

3 MAJOR TYPES OF QUBIT TECHNOLOGIES

1. Superconducting

Solid-state circuits maintained at extremely low temperatures are used to form qubits.

PRO

- Easy to manufacture (able to achieve high qubit count)

- Higher gate speeds

CON

- Lower fidelity (quality) due to imperfect qubits (manufacturing defects)

- Refrigeration requirements (energy intensive)

- Shorter coherence time

- Limited connectivity (qubits that are laid out close to each other)

2. Trapped Ions

Ions are suspended in a vacuum chamber to form qubits.

PRO

- Higher fidelity since ions are nearly identical (no defects)

- Able to operate at room temperature (low energy)

- Longer coherence time

- All-to-all connectivity (all qubits can interact with each other)

CON

- Low qubit count (not easy to fit many ions in a vacuum chamber)

- Slower gate speeds

3. Photonics

Light particles are generated to form qubits.

PRO

- Able to produce many qubits (theoretically)

- Able to operate at room temperature

- Resistant to noise

CON

- Photons are easily lost

- Hard to form entanglement between qubits

NOT JUST ABOUT THE QUBIT COUNT

So back to the question about qubit count - is more the better? Quantum annealing (D'wave) may have the highest qubit count but it's only suitable for optimization problems. Superconducting gate-based QC have the next highest qubit count, but suffer from decoherence and low qubit fidelity. Ion traps have the highest quality qubits and connectivity but are slow and limited number of qubits. Photonics have noise-resistant qubits but also hard to achieve entanglement. Experts have estimated that we'll need approximately 1 million qubits to make useful quantum computers and we are far from that. In fact, even if we were to reach 1 million qubits, without a high qubit fidelity (at least 99.9999%) is hardly meaningful because errors will compound across consecutive operations. 

WHAT TO LOOK OUT FOR INSTEAD

In my opinion, besides qubit count there are 2 other areas that I'll keep an eye on

1. Scalability

Regardless of which qubit technology (except maybe photonics), there will be physical limits to how many qubits can fit into 1 system. There will be a need for complementary technologies to scale the system. Currently, "photonics interconnect" seem to be the most promising candidate for that.

2. 2-Qubit Gate Fidelity

As mentioned earlier, complex calculations require operations of entangling 2 qubits. Therefore errors will compound even if you have seemingly high qubit fidelity. For example, 99.9% fidelity quickly drops to 36% across 1,000 operations. So rather than qubit count, I'd look out for news about "99.99%", "error mitigation", "error correction", "qubit fidelity" and "fault-tolerance" instead.

Thank you for reading. In the next article, I'll be sharing my thoughts about IonQ - the good and the bad - and why it is better positioned to succeed than its peers.

Disclaimer: Information provided here are based on my own research. There may be inaccurate or outdated data presented here.