A Quantum Cat

As if on cue, Amazon (AMZN) just threw its hat into the ring for quantum computing…

This morning, the e-commerce and cloud computing giant announced it has designed and manufactured its own quantum computing semiconductor, claiming its new innovative design is the world’s “first realization of a scalable architecture for bosonic error correction.”

I couldn’t help but feel similar to how I felt when I read Microsoft’s announcement about its own topological quantum computing semiconductor… also a “world’s first.”  We explored that development in yesterday’s Bleeding Edge – Quantum Breakthrough or Quantum Spin.

And just like with Microsoft’s Majorana 1, I have my doubts about Amazon’s new chip…

Amazon’s Novel Approach

Amazon’s new quantum chip is Ocelot – a superconducting quantum semiconductor similar to the approach taken by Alphabet (GOOGL) and Rigetti (RGTI).

Ocelot | Source:  Amazon

The quantum team at Amazon took a novel approach in designing Ocelot by building in a new approach to quantum error correction – the most significant challenge for any quantum computing system.

Amazon claims that this new approach, designed into Amazon’s new quantum semiconductor, will reduce error correction by up to 90% compared to approaches taken by other quantum computer companies.

Ocelot Semiconductor Before Packaging | Source:  Amazon

At the core of Amazon’s approach is the cat qubit, named after the thought experiment Schrodinger’s Cat, which supposes that a cat can be in two different states at exactly the same time (e.g. alive and dead at the same time).

Hence, the name of a cat, Ocelot.

An Actual Ocelot

Cat qubits are promising because they are more resilient against a certain type of error (bit flip errors).  With the new hardware design and by increasing the energy through the system, Amazon will, in theory, be able to dramatically reduce the errors through its quantum computing system.

Increasing the energy in the system will produce more photons, which helps to reduce the amount of quantum errors.  Photons are bosonic particles, hence Amazon’s use of the term “bosonic error correction.”

Promising… In Theory

There is a tradeoff though that Amazon didn’t mention, working with cat qubits can be tricker and much more difficult to scale than more conventional superconducting qubits like those used by Alphabet and Rigetti.

Kind of reminds me of the difference between working with cats versus dogs…

Again, in theory, Amazon’s design suggests that its path toward a universal fault-tolerant quantum computer using cat qubits would require a far smaller number of qubits compared to other quantum systems.

Amazon has a long way to go to prove that, however.  In its own words:

While today’s announcement is a promising start, Ocelot is still a prototype and AWS is committed to continuing to invest in quantum research and refining its approach.

“We’re just getting started and we believe we have several more stages of scaling to go through,” said [AWS Director of Quantum Hardware, Oskar] Painter. “It’s a very tough problem to tackle, and we will need to continue investing in basic research, while staying connected to, and learning from, important work being done in academia. Right now, our task is to keep innovating across the quantum computing stack, to keep examining whether we’re using the right architecture, and to incorporate these learnings into our engineering efforts.”

And just like Microsoft is not offering its Majorana 1 quantum computer to its Microsoft Azure cloud services customers, Amazon isn’t offering Ocelot to any of its Amazon Web Services (AWS) customers:

Source:  Amazon

Also telling is that the two most widely used quantum computers on Amazon Web Services – IonQ (IONQ) and Rigetti – are also the same quantum computing systems used by Microsoft on Microsoft’s Azure cloud services platform.

I expect it will be years before Amazon will be ready to offer its quantum computing platform alongside its existing partners.  After all, this is just Amazon’s first prototype.

A Matter of Scaling

We should, however, take Amazon’s efforts more seriously than Microsoft’s.

Amazon is building a superconducting quantum computer using manufacturing processes that exist today in semiconductor manufacturing.  And both Alphabet and, more importantly, Rigetti have already proven that this approach is effective and can scale.

The other reason this development is worth tracking is that Amazon has already had great success designing its own custom semiconductor for its cloud services business.

For the last couple of years, about 50% of all new computing capacity offered by Amazon Web Services (AWS) has been running on Amazon’s own Graviton semiconductors, which are all manufactured by Taiwan Semiconductor (TSM).

It has been a major success story for Amazon as its ARM Holdings (ARM)-based Graviton chips result in cloud service prices that are 20% lower for its customers and use as much as 60% less energy per unit of compute than chips designed on x86 architectures (like Intel).

The superconducting design makes perfect sense for a cloud services company like Amazon as well.  Superconducting quantum computers have to operate in pristine environments void of vibration and at near-absolute-zero temperatures, colder than space, in a dilution refrigerator.

AWS’s Quantum Computer | Source:  Amazon

Very few companies would have the expertise to maintain their own superconducting quantum computers on-premise, so accessing these kinds of services through the cloud will be the most common way.

Regardless of the latest announcements by Microsoft and Amazon, the last three months in the quantum computing industry have been the most exciting ever.

There is clear progress with quantum error correction and an increasing level of investment flowing towards quantum computing companies, as well as great progress with new encryption standards designed to be resilient against quantum computing hacks.

Leading companies in the industry are already commercializing their quantum computers.  It’s no longer the realm of just research and development.

Now it’s just a matter of scaling quantum systems and improving error correction technology.

Jeff