Determining when next-gen computing makes financial sense won’t be easy. A new framework finds that enormously complex business problems will benefit, but many workloads will still run best –and more economically — on classical computers.
By Peter Krass
Like many technologies, quantum computing is both a brilliant breakthrough and also a misunderstood advancement. Decades in development, quantum computing still only hints at its full capabilities — especially for business applications.
That’s why researchers Neil Thompson, lead of the MIT Initiative on the Digital Economy (IDE) Technologies that Create Prosperity research group, Sukwoong Choi, a postdoctoral scholar at MIT IDE, and William S. Moses, a postdoctoral scholar at MIT Computer Science and Artificial Intelligence Lab have created a framework to help tech-savvy executives evaluate the potential of quantum computing even before it’s widely available. The methodology was co-developed with Accenture, a founding member of the IDE.
Speaking at IDE’s recent annual conference, Thompson explained that while quantum computing has great potential for speeding up some computational tasks, it also has important limits.
Most likely, it will be used by businesses for specific niche problems while today’s conventional computers — what researchers call “classical” computers — will continue to be used for many years to come.
Given that, the challenge for leading-edge IT managers will be deciding which tasks are appropriate for which type of computer. Near-term, quantum computers, even once they’re commercially available in the cloud, are likely to be expensive. Therefore, putting the right job on the wrong technology could be a costly mistake. Thompson said one goal of the framework is to identify quantum’s sweet spot: the economic as well as its processing advantage for a given job.
Technical Progress
For example, where classical computers are built on silicon semiconductors, some quantum computers use what are known as electronic superconductors that must be kept extremely cold — within a degree of absolute zero, or -273.15 C.
Despite this and other technical issues, progress is being made quickly. IBM, an aggressive developer of quantum computers, expects to move from its current “Osprey” quantum computer with 433 quantum bits, or qubits, to “Kookaburra,” a computer with more than 4,500 qubits, by 2025. (The more qubits, the bigger the problems a quantum computer can tackle.)
Business adoption of quantum computing could follow quickly.
According to Quantum Computing Market Size, Share | Forecast Report, 2030 “the quantum computing market size is projected to grow from $928.8 million in 2023, to $6,528.8 million by 2030, at a CAGR of 32.1%.”
Healthcare, automotive, financial services, chemical, manufacturing, energy and utilities, transportation, logistics, and others are the end users of the market, the report says.
Tortoise or Hare?
For business leaders — especially those who aren’t technical experts — learning about the technology and figuring out which applications are good candidates for quantum computing can be a challenge. Once many of the technical issues are resolved, businesses will need guidance on evaluating their needs, allocating resources, and reaping the benefits.
In his presentation, “The Quantum Tortoise and the Classical Hare: Which problems will quantum computing accelerate?” Thompson offered a starting point for understanding the potential of quantum computing for real problems, even if the quantum algorithms haven’t been invented yet. “That may seem counterintuitive,” Thompson told IDE annual conference attendees, “but in fact, it’s a powerful feature of the model, and that’s going to be very helpful.” (A research article on the topic will be published shortly.)
In the tortoise and hare example, classical computers are generally faster, but require more steps to accomplish a task — they follow an inefficient path from Point A to Point B. In contrast, quantum computers have the potential to take a more direct route, which may result in a shorter path from Point A to Point B, but they also have slower processing speeds, so each step takes longer.
The point of the framework is to determine whether the shorter route or the faster computer is more important for getting from A to B quickly.
Thompson also explained that the cost of quantum computers should soon start coming down. “There’s going to be a time when these two technologies [quantum and classical computers] are cost-competitive,” he said. “And that’s going to create what we call the quantum economic advantage.” This happens when a particular problem can be solved more quickly on a quantum computer than any comparably priced classical computer.
As Thompson explained, this will give managers two main decision points to consider:
- Feasibility: Is the quantum computer big and fast enough to solve a particular problem?
- Algorithm advantage: If yes, would the quantum computer be faster than on a comparably priced classical computer?
Feasibility is a big issue because affordable, business-ready quantum computers don’t currently exist. And when they might be available is still unknown. In another session at the IDE annual conference, research scientist Jonathan Ruane estimated the timeframe as, “three to 10 years away.”
But Ruane also left open the possibility for technical breakthroughs that could speed things up. “Who could have predicted the success of ChatGPT?” he asked by way of example.
Ruane also said that quantum computers are accessible in the cloud now but “they won’t replace classical computers, they will work alongside of them.” Think of them as “another string to the bow,” not like DVDs replacing VHS media, he said. [Read more about Ruane’s work here.]It’s also possible that the technology’s complexity could lead to technical issues that bring progress to a crawl.
Despite the view of quantum computers as the next-generation, they actually run much slower than classical computers.
“There’s an enormous speed difference,” Thompson explained. This speed difference is why quantum computers are only beneficial if they offer a shorter route (i.e. the better algorithm) that overcomes the speed gap,
Fortunately for quantum computers, there can also be enormous differences in algorithms. Where classical computers often follow an inefficient path to solve a problem, quantum computers may have a more direct algorithm. The benefits of a better algorithm grow as problems get bigger, for example having the right algorithm for alphabetizing a set of files is more important if you have 1 million files than if you have 3. As a consequence, quantum computers have more advantages for large problems.
The takeaway? Quantum computing won’t be better for every business problem, but it should be great for the right ones. Namely, problems that are difficult, small enough to run on a quantum computer, yet also big enough to enjoy quantum’s algorithmic advantage. That’s the new sweet spot. By contrast, problems that already have adequate algorithms, as well as most day-to-day business problems with small data, will continue to run best on classical computers.
Watch the video of the session here.
Peter Krass is a contributing writer and editor to MIT IDE.