By 
As AI chips get denser, the hardest limits are no longer inside the chip. They are at the boundaries, the tiny interfaces where one material meets another and heat, current, and signals pile up and get stuck. nanoSkunkWorkX, a Malaysian deep-tech startup, is built around that unglamorous insight, and it has just raised $2 million in a seed round led by Singapore’s Tin Men Capital to prove its fix works at scale.
The company’s lead product, nSD-I, is a thin graphene-copper film for the copper interconnects inside advanced AI chip packaging. Its pitch is a small change with an outsized payoff: help the copper carry more current and shed heat faster, which reduces thermal throttling and cooling demands and, in turn, lowers the cost of running AI. Crucially, it is designed as a drop-in improvement, no replacing copper and no changes to existing manufacturing. In peer-reviewed testing, the company says the film moved heat across its surface more than twice as effectively as plain copper, and held that edge after use in pre-qualification components.
One idea, three markets
What could look like a lack of focus is instead a single thesis applied three ways. The same interface engineering that helps a chip shed heat also governs how efficiently a hydrogen electrolyser transfers charge, and how sensitively a biosensor detects a signal. So alongside semiconductors, nSWX is working on green hydrogen, aiming to cut reliance on expensive platinum-group metals, and on diagnostics, including a paid joint research programme with the US Navy’s medical research unit in the Indo-Pacific. The unifying belief is simple: at scale, the interface becomes the ceiling, so improving that thin boundary can unlock big gains everywhere without disturbing the materials and machines already in place.
The team behind it is unusually credentialed for a seed-stage company. Co-founder and CTO Dr Amani Salim is a former NASA principal investigator, and CEO Iqbal Shamsul is an MIT-trained computer scientist and former Morgan Stanley commodities executive, both of whom returned to Malaysia to build frontier hardware. And they have been strikingly frugal: three product lines, peer-reviewed data, and first revenue, all on under $1.5 million of prior capital.
The mountain is qualification
Here is where realism matters. Materials for semiconductors is one of the slowest, most punishing businesses in technology. Getting a new material qualified into chip-packaging supply chains takes years of rigorous testing, and incumbents like plain copper are entrenched precisely because they are proven. The drop-in design lowers the barrier, but qualification is still the mountain to climb, which is exactly what this funding is meant to pay for. The performance claims, impressive as they are, remain pre-qualification; the real test is whether they survive contact with a manufacturer’s production line.
Malaysia gives the bet a genuine home-field edge, handling roughly 13% of the world’s semiconductor assembly, testing, and packaging, with state funds now channeling money toward exactly this kind of frontier work. That matters, because deep-tech hardware needs patient capital and nearby customers, and nSWX has both within reach.
So can a $2 million startup crack a bottleneck that constrains the entire AI industry? Not on its own, and not quickly. But it has picked a real problem, framed it elegantly, and proven it can do a lot with a little. In a field obsessed with ever-bigger chips, betting on the thin, overlooked space between materials is a quietly contrarian idea, and it might just be where the next gains hide.
Comments