Imagine being dropped into a vast mountain range, such as the Adirondacks and being tasked with finding a specific lake.
You are told that the lake is located at the lowest part of the mountain range and the only tools you are given are a compass and a method for recording the paths you have taken. As you can imagine, this task would be next to impossible. You may be lucky and find it right away, or you may spend years searching.
Now imagine being in a helicopter above the mountains and being asked to do the same thing. Looking below, at the entire mountain range at once, you can view all possible paths and find the lake almost immediately.
I often use this analogy to illustrate the distinction between quantum and traditional computing. Classical computers move step-by-step, evaluating one path at a time. Quantum computers leverage the physics of the quantum world to explore many paths simultaneously. They operate in an exponentially expanding solution space – giving them the potential to solve certain classes of problems in seconds that could take classical machines centuries.
Why quantum matters (and why it feels familiar)
Quantum computing may feel like a futuristic headline, but it echoes a pattern we’ve lived through before.
I remember watching Steve Jobs and Steve Wozniak on the evening news when I was a kid – two guys in a garage building what society said were really oversized, more powerful calculators. It is hard to imagine that it would become Apple, one of the world’s most impactful companies.
However, from an applied, everyday perspective, the same thing occurred with the internet. Back in the 1980s, modems were so slow that they tied up the home phone line, and you were often yelled at by your parents. A final example is the mobile phone, which started off as a sizable briefcase with a corded phone attached. In all three cases, no one would have ever predicted that each would have changed the world the way it did. Finally, we are currently living through this again with AI and we are about to experience it again with quantum computing.
Quantum AI: What it is and why it matters
From theory to traction
Professor Richard Feynman introduced the idea of quantum computing in the 1980s. I first read about it while in graduate school in a biophysics lab in the late 1990s. Fast forward to today and I would compare current quantum computers to traditional computers of the 1970s/1980s. They work; they perform calculations using quantum physics, but are limited to very specific use cases, largely due to current hardware limitations.
However, I’m seeing the writing on the wall and believe its limitations will be overcome. When they are, well, who knows? Let your imagination run wild. It could redefine industries, scientific discovery and national security, among other impacts.
Key advancements and a moving timeline
I first started programming quantum computers about six years ago. At that time, the estimate of fault-tolerant quantum computers with enough logical qubits – the basic unit of information in a quantum computer – to be useful was estimated to be around 30 years away. Now, quantum roadmaps are aiming for 2030, as published by IBM Quantum, IonQ, QuEra and Google.
The industry has made tremendous progress in areas such as qubit quality, error mitigation and correction, logical qubit counts and architectural connectivity between the qubits. All of these taken together bring us closer to near-term quantum advantages. I have no doubt that 2030 will be continually adjusted to be sooner.
What's next in the quantum race?
What is also accelerating quantum computing is like what happened in the 1970s/1980s. Back then, there were about 45 unique computer manufacturers (e.g., Apple, IBM, Commodore, TRS, etc.). Most of them have disappeared through mergers and acquisitions, resulting in technological acquisitions.
As in the past, today there are well over 100 quantum computing companies across various quantum modalities (e.g., superconducting, photons, neutral atoms, trapped ions, etc.). We are also beginning to see mergers and acquisitions in the quantum computing sector.
According to McKinsey, the potential economic value from quantum computing 2035 will be about $1.5 trillion, resulting in tremendous investment to be leaders in this space.
I would consider quantum computing as a sort of arms race that was initially sparked when Professor Peter Shor from MIT published his quantum factoring algorithm, which showed that with a powerful enough quantum computer, RSA could be decrypted. All our encryption is based on some kind of variant of this and governments took notice.
Public investments have increased year after year. As of 2023, there was a record of $42 billion in government investments. Now, some countries have a version of a quantum initiative aimed at building the most powerful quantum computer while developing methods to prevent quantum RSA attacks. This has also evolved into large national defense projects across most three-letter agencies. On the private funding side, investments have decreased between 2022 and 2023, from $2.33 billion to $1.7 billion, which is due to a few reasons, including the stabilization of the quantum landscape and expectations of ROI. When we talk about ROI, it’s important to distinguish between short-term and long-term.
Most companies are focusing on long-term ROI, adopting a 'build it now, use it later' mentality. Companies are publishing articles, algorithms, and patents with the knowledge that their usefulness for real-world applications depends on the maturity of quantum hardware. These companies are looking to own IP for the long-term ROI.
This is not to say short-term ROI is not important, because it is. Using imaginative methods and hybrid processes, short-term ROI has been achieved in areas such as optimization and machine learning. This is not universal and tends to be on a case-by-case basis, typically centered around speed, ability to represent complex combinatorial data structures, more expressive models created with less data, and decreased power consumption. For short-term ROI, imagination and thinking outside the box are important.
Hybrid reality and convergence
Currently, quantum computers are still in their early stages of development. Professor John Preskill coined what the industry refers to as the Noisy Intermediate-Scale Quantum (NISQ) era – meaning systems are powerful, but still noisy and limited. They can solve only small, specialized problems and they need help from traditional computers to get meaningful results.
Quantum computing will complement an HPC center, serving as another tool in the computational toolbox that can be utilized as an accelerator, a higher-dimensional feature engineer, or a simulator, among other applications.
There has been extensive research in the QML areas, including quantum neural networks, quantum reservoirs, quantum natural language processing, and quantum LLMs, with mixed results. This is to be expected with such new and young technology, which is very different from anything the world has seen before. This is uncharted territory with no established rules or best practices; these are being defined in real-time. Being creative and thinking outside the box is imperative – maybe even thinking more like a physicist than a statistician will facilitate breaking out of the box.
Where we stand
Quantum computing right now is like standing on the event horizon of a black hole.
We have no idea about how some aspects of quantum computing work, but we know how to use it. We can peer into it, measure its effects, but don’t have the complete picture.
Every week, there are publications of incredible new results or discoveries. We are pushing technology to the quantum level, taking advantage of the strange world of quantum physics, where things can exist in two states simultaneously, be entangled across light-years, tunnel through barriers, and be teleported. This makes for an extremely challenging field, but that’s why it’s so much fun. In fact, I think it is the most exciting technological evolution and it will have the greatest impact on humankind.
