• The advanced packaging lab is integrating lithium niobate onto silicon, and says it is in talks with UMC.
• Director Aaron Thean says the approach could open the technology to smaller firms, and that Malaysian companies are welcome as research partners.

The National University of Singapore is working on advanced packaging problems that commercial foundries have not taken on, and says one of its methods could let smaller companies adopt the technology without dedicating a production line to it, according to Professor Aaron Thean, who directs the university’s SHINE research centre.

That work centres on lithium niobate, a crystal that converts electrical signals into light more efficiently than silicon can, and on getting it onto a silicon chip without contaminating what is already there.

“You saw what we are doing with lithium niobate and chips,” Thean said. “A lot of the advanced foundries cannot even do that right now. So we’re even ahead of the industry itself.”

Photonics is where the industry is heading as chips run out of room to shrink. Thean, who is also NUS deputy president for academic affairs and provost, and was vice president of logic technologies at Belgian research institute imec, said AI models are scaling at roughly 10 times every two years against a best case of two times for transistors. That gap has pushed the industry towards connecting chips, boards and entire data centres at high speed rather than shrinking transistors further, which is the work advanced packaging does.

Asked about UMC and lithium niobate photonics production in Singapore, Thean said NUS is “in talks with them,” but declined to give details. UMC, the Taiwanese contract chipmaker, opened a 22nm fab expansion on the island last year.

Why the advanced packaging method could suit smaller players.

Foundries have held back from lithium niobate because of contamination risk, according to Thean. Etching the material on a silicon wafer exposes everything underneath it, and “people are afraid of lithium because it can leach out,” he said. A fab can work around that, he said, but it means dedicating a process line to it, which carries cost.

NUS took a different route. “We pattern lithium niobate on a source wafer, fully form the device, and we bring it in at the back end,” Thean said. By that stage, he said, most of the chip is already done.

The approach, he said, “may allow for early adoption of the technology, or smaller players coming in to adopt the technology.”

SHINE, the Singapore Hybrid-Integrated Next-Generation μ-Electronics Centre, is a national medium-sized research centre hosted at NUS and funded by Singapore’s National Research Foundation to work on heterogeneous integration. 

It reports S$134 million in cumulative funding since 2021 and five technologies deployed. Two companies have spun out of the work. AiTexel, started by a colleague Thean named only as Benjamin, who is chief executive, has raised seed funding from US investors that Thean described as significant, and remains largely in stealth. NextStrong, working on 2D materials, has recently come out of stealth with partners Thean declined to name.

The centre also works with companies ahead of their own roadmaps. “We work with companies like Marvell to look at these things, and they don’t have this kind of technology,” Thean said. TSMC, he added, “do engage us in different projects.”

“We work on problems ahead of them so that ultimately we can help them,” Thean said. “You can’t help them if you’re behind them. You have to get ahead of them.”

Where Malaysia fits

Asked whether a Malaysian company could partner with SHINE rather than simply supply it with tools, Thean said: “Totally, yes.” Asked whether any such partnership exists today, he said: “Not yet.”

Ties with Malaysian universities do exist, but not in research. “From the university side we already have partnerships, but mostly on the education side,” he said. “But how do we bridge that? That’s the key.”

Thean was candid about why he wants the partners. “Singapore has its own weakness,” he said. “Our weakness is that land cost and energy cost are high. Malaysia has more resources.”

The binding constraint, he said, is people. “You may have technologies, but if you don’t have enough people in terms of numbers to train, it doesn’t work out, because ultimately the capability comes from the number of people. Malaysia has a lot of industry experience, because in Penang and the surrounding areas, there are a lot of people already in the industry.”

He put the argument in terms of the supply chain rather than the balance sheet. “There’s no reason to compete, because we are all filling different parts of the gap in the ecosystem,” Thean said. “We are just not filling it together.”

He has a specific proposal. “Why not have a joint education and training programme where some Malaysian folks can come over, and we set up co-research activities in Malaysia, and then allow the diffusion to happen,” he said.

“This is something the universities probably should look into doing together. The university is a very neutral ground to do this with, so it doesn’t have as many concerns when it comes to proprietary technology,” Thean said.

Why it has not happened yet

“The gap exists not because of cultural differences, because culturally Malaysia and Singapore are almost identical,” Thean said. “The main issue is that people are just too busy.”

Both countries are mid-build. “Malaysia is aggressively growing semiconductors in AI, so is Singapore,” Thean said, adding that each wants its own house settled before turning outward.

The funding reflects that. Singapore is spending S$800 million on an RIE Flagship in Semiconductors covering 2026 to 2030, and added five industry collaborations last Friday, all with foreign companies. Malaysia committed RM185 million in May to the Malaysia Advanced Packaging Consortium, formed by SkyeChip, Inari Technology, FusionAP, Pentamaster Instrumentation and NSW Automation, targeting 7% of the global advanced packaging market by 2035. Neither has been announced with a cross-border research component.

Thean said closing the gap would take “more incentives on both sides” to reduce the friction, or at least the inertia.

On talent, he argued that retention is not solved by holding people in place. “The key is to make sure that the opportunities exist and match up,” he said. “Can they get enough industry in quickly to absorb them? People don’t go overseas because they just want a vacation. They are looking for opportunities.”

He extends the same logic to Vietnam, which he described as having a large STEM-trained workforce, many of them trained in AI, and a shorter research history that he does not regard as permanent.

“The region needs a strong partnership pact on dealing with technology, because each of us has strengths and weaknesses,” Thean said.