ST Explains: Why is S’pore investing $700m in quantum computing? (2024)

SINGAPORE – More firepower, led by a $300 million investment top-up by the Government over the next five years,will be added to fuel quantumtechnology research and talent grooming in Singapore.

The aim is to build quantum processors on the island, a move that would extend the influence of the little red dot in developments such as the discovery of new drugs and new materials for building longer-lasting batteries.

Quantum computing could also turbocharge generative artificial intelligence (gen AI) capabilities and efficiency.

The $300 million injection, announced on May 30 by Deputy Prime Minister Heng Swee Keat, adds to the initial $400 million research and development budget that the National Research Foundation has poured into quantum technology since 2002.

Due to high costs, among other reasons, only a handful of other nations – including the United States, China, France, Finland, Germany, South Korea and Japan – are on this quest to build a national quantumcomputerto gain first-hand experience with the technology.

The Straits Times explains why Singapore is investing $700 million in quantumcomputing research, and how the technology could unlock the power of gen AI.

1. What isquantumcomputing?

It is similar to traditional computing, but operates at the far cooler temperature of nearly absolute zero, the temperature at which a thermodynamic system has the lowest energy – corresponding to minus 273.15 deg C.

Under layers of casing and cryogenic components to attain this extreme cool state – colder than in outer space – quantumobjects such as an electron or a particle of light are manipulated to execute complex mathematical calculations beyond the reach of traditional computers.

Traditional computers store information as either 0s or 1s. Quantum computers, on the other hand, usequantumbits (or qubits) to represent and store information in a complex mix of 0s and 1s simultaneously. As the number of qubits grows, aquantumcomputerbecomes exponentially more powerful.

Quantumcomputing’s long development history dates back to the 1970s, when the late American physicist Paul Anthony Benioff demonstrated the theoretical possibility ofquantumcomputers.

By harnessing quantum physics, quantum computing has the potential to comb vast numbers of possibilities in hours and pinpoint a probable solution. It would take a traditional computer hundreds of thousands of years to perform a similar task.

Japan’s first prototypequantumcomputer, unveiled in 2017, could make complex calculations 100 times faster than a conventional supercomputer.

Google’squantumcomputercreated in 2019 could perform in 200 seconds a computation that would take the world’s fastest supercomputers about 10,000 years.

A year later, in 2020, a team at the University of Science and Technology of China assembleda quantum computer that could perform in 200 seconds a calculation that an ordinary supercomputer would have taken 2.5 billion years to complete.

But none of these machines was given practical tasks to perform.

2. What are the real-world benefits ofquantumcomputing?

Its first real-world application could be in chemistry.

Today, a traditionalcomputercan simulate the structure of a simple water molecule, but only powerful quantum computers will be able to accurately simulate more complex molecules such as enzymes, proteins and DNA. Complex molecular simulations are key to the discovery of new drugs.

Riding on the success of their messenger RNA (mRNA) Covid-19 vaccines in the Covid-19 pandemic, biomedical firms are developing mRNA vaccines for malaria, tuberculosis and HIV, all of which are still causing many deaths in lower-income countries. This is a potential application for quantum computing.

More powerfulquantumcomputers can also speed up the discovery of new materials to address present-day battery woes. Specifically, more efficient, safer and greener batteries are needed for electric and autonomous vehicles.

Lithium-ion batteries that power automotives use liquid electrolytes to move energy around. But these batteries can be slow to charge and do not last long enough for longer journeys. They also freeze in sub-zero temperatures and contain flammable material. What’s more, their production requires the extraction of rare earth materials using large amounts of water.

Hyundai is working with quantum computer maker IonQ to explore ways to supercharge lithium-ion batteries. They are also exploring new metal catalyst chemical reactions, which could enable Hyundai engineers to develop higher-performance electric vehicles at reduced costs.

IBM and Daimler, the parent company of Mercedes-Benz, are working on next-generation lithium-sulphur batteries for a more powerful charge that lasts longer than lithium-ion ones.Meanwhile, Nissan, Honda, Volkswagen, Ford, Mercedes-Benz, BMW and Toyota are working on safer solid-state batteries. These are applications waiting for a quantum computing breakthrough.

Advances in this space will spill over to mega-battery grids, currently also based on lithium-ion technologies, to ensure a smooth supply of renewable energy to homes and offices even when there is no sunshine or wind. Mega-battery storage facilities are located across the US as well as in England, Lithuania and Chile as the charge towards net-zero emission goals intensifies.

Also of interest to the energy sector is a more efficient way to plan global logistics, such as those for transporting oil and gas. This would also benefit from quantum computing.

3. What can quantum computers do for gen AI?

Quantumcomputers, which work in similar ways to the human brain’s neural network, are best suited for gen AI tasks that involve understanding the context and nuances of human language.

Specifically, techniques inspired by quantum physics could have the potential to reduce the computational costs for large language models (LLMs), and complex gen AI tasks such as simulating large virtual environments or creating high-resolution content.

Today, training LLMs on traditional computers to understand and respond to human natural language inputs requires a lot of computing power. The process emits a lot of heat that requires cooling by air or water.

OpenAI’s GPT-3 model was estimated to emit over 500 tonnes of carbon dioxide during training. Its latest GPT-4 model is a lot bigger and has about 1,700 billion parameters (values or instructions a model can independently learn from its training data), making it far more pollutive.

The biggest challenge yet is in automating the detection and removal of disinformation and toxic social media posts, videos and images – much of which can be easily created using gen AI tools. Quantum computers hold the promise of enabling such complex computation.

4.Quantumcomputing has been talked about for decades. When will it finally deliver on these benefits?

It will requirequantumcomputers to become cheaper to solve real-world problems such as simulating drug molecules and processing images, natural language and software codes to filter hate speech, misinformation and software bugs.

Each quantum computer costs upwards of US$100 million (S$135 million) to set up now.

Today, the world’s fastestquantumcomputeris based on a 1,121-qubitquantumprocessor by IBM rolled out in December 2023.

IBM said it has developed a new way of connecting quantum computing chips inside machines and linking these machines together that could produce error-proof quantum machines for real-world tasks by 2033. The company is planning to incorporate the new chip in its AI platform,watsonx, for business users.

5. Will aquantumcomputing chip be in my smartphone in the future?

No. Traditional computing methods are still needed to run most smartphone and computing applications such as data crunching, storage and management, spreadsheet calculations and video gaming.

Quantumcomputers are good only with specific tasks governed byquantumphenomena, such as modelling and designing complex materials. So,quantumcomputers will not take over traditional computers. They need to work together.