Two Singapore brothers build encryption company on math problem no algorithm can solve

Two brothers in Singapore have built a data-encryption company on pure mathematics, betting that a problem no algorithm can solve will keep files safe from the quantum computers expected to break much of today's encryption within years.

Sometime in the next several years, the encryption that protects bank transfers, medical records and state secrets is expected to start failing. Large quantum computers, still being built, will eventually unravel the math that keeps that data private far faster than any machine today.

In Singapore, two brothers have built a company on a bet about how to stop them.

Lim Meng Liang, 38, and his older brother Ken Lin, 45, founded Aires Applied Quantum Technology in 2023 to sell encryption they argue will hold up even against quantum attacks. Their wager rests on an unusual foundation: a class of pure-mathematics problems that no algorithm can ever fully solve, The Straits Times reported.

Lim, a National University of Singapore applied mathematics graduate who also runs his own investment firm, specializes in Diophantine equations, polynomial equations for which mathematicians search for whole-number solutions.

Contrary to how they are often described, these are not simply equations "without solutions." Their power comes from a deeper property: no general method can decide whether an arbitrary Diophantine equation has integer solutions at all, a problem essentially equivalent to the unsolvable halting problem.

That undecidability, the negative answer to Hilbert's tenth problem settled by mathematicians in 1970, is what Lim turned into a way to scramble data.

In 2022 he was granted his first U.S. patent, No. 11,522,674. According to the filing, his encryption method pairs this undecidability with artificial intelligence, building cipher text out of equations whose solvability cannot be feasibly determined and generating decoy or "faux" equations to mislead attackers.

The company has described the approach as among the first to combine AI with what it calls undecidable encryption, based on problems that no algorithm can solve.

A researcher configures a dilution refrigerator used to cool a quantum computer at SEEQC's facility in New York. Quantum machines are expected to break much of today's encryption within years. Photo courtesy of SEEQC/Handout via Reuters

"I could have just framed the patent and kept it on the wall," Lim told The Straits Times. "But after some hard thinking and discussions with my brother, we decided to do something with it."

Lin, who had spent 15 years in finance, quit to join him. The two carry different surnames because of an error their parents made on Lin's birth certificate.

The threat the brothers are selling against is real, if not yet present. No quantum computer can break today's encryption now, and IBM's most powerful processor holds just over 1,100 qubits.

But the timeline is tightening. Google researchers reported in May 2025 that a machine with fewer than a million noisy qubits, running for about a week, could crack RSA-2048, a common encryption standard, a 20-fold drop from the company's 2019 estimate, the trade publication The Quantum Insider reported.

IBM, meanwhile, has said it aims to deliver its first large-scale, fault-tolerant quantum computer, named Starling, by 2029, according to IEEE Spectrum.

The U.S. National Institute of Standards and Technology, which finalized its first post-quantum encryption standards in 2024, has urged organizations to retire vulnerable encryption after 2030 and drop it entirely by 2035. Even so, expert surveys compiled by the Global Risk Institute place a genuinely code-breaking machine in the 2030s or later.

Lin puts the contrast bluntly. "A standard computer may take 1,000 years to crack current encryption methods, but it will take just two days for quantum computers," he told The Straits Times, adding that Diophantine-based encryption would stay practically impossible to solve even for such machines.

That claim is the company's, not a settled fact. Encryption built on Diophantine equations has not been proven unconditionally secure, and academic researchers have shown that some versions can be broken using AI, A* search or genetic algorithms.

It is also not among the lattice- and hash-based schemes that NIST adopted as its first post-quantum standards in 2024, which most governments and large firms are now migrating toward.

Aires is selling anyway. Its flagship app, LionGuard, encrypts files on devices, in the cloud and across networks, working through a key the sender generates and shares through the app for the recipient to unlock.

Still in beta at S$388 (US$300) per user a month, LionGuard has more than 100 subscribers across oil and gas, commodities trading, education technology, cloud services and banking, the company said.

Aires says it holds four international patents, including one for encrypting QR codes and barcodes now being piloted in Europe. It has raised more than US$2 million from private investors and government agencies including Enterprise Singapore, and is eyeing a listing in the U.S., Singapore or Japan to raise "an eight-digit figure."

The brothers are riding a wave of state money. Singapore unveiled its S$37 billion (US$28.7 billion) Research, Innovation and Enterprise 2030 plan on Dec. 5, 2025, a 32% increase on the previous five-year budget, The Straits Times reported, and has separately committed S$300 million (US$230 million) to quantum specifically under a National Quantum Strategy unveiled in 2024.

The same month, CNBC reported, Horizon Quantum became the first private firm to deploy a commercial quantum computer in the city-state, ahead of a planned U.S. listing.

When Aires first pitched in 2023, Lin said, audiences confused quantum cryptography with cryptocurrency.

"Today, I don't get such questions any more," he told The Straits Times. "Organizations are now asking practical questions about integration, deployment timelines and long-term readiness."