Note

170 People Showed Up. Seven Small Companies Walked Away With an Actual Contact.

That was the result of "Connect Day," an April 2026 event organised by the Malaysia Semiconductor Industry Association to put local SMEs in a room with global chipmakers — Micron, Intel, Infineon, Texas Instruments, and others. Out of 170 attendees, seven small Malaysian companies came away with a direct line to one of those firms. No financing was attached. No training programme. Just the chance to find out, in the organisers' own words, "what the MNCs require" — leaving each company to go figure out how to meet that requirement on its own. Penang plans to run a similar event in September. Compare that to what Petronas offered oil and gas vendors in 1993 — 439 specific, reserved contracts, with Petronas as the guaranteed buyer — and the gap between the two approaches is the story.

The scale of what a matchmaking day is up against

Malaysia's semiconductor industry needs roughly 50,000 skilled engineers to meet current demand. Local universities produce about 5,000 graduates a year — a tenfold shortfall that isn't closing on its own, especially with the industry losing an estimated 15% of its existing talent every year to Singapore, Taiwan, the US, and Europe, according to the Malaysia Semiconductor Industry Association's own president. Malaysia's STEM enrolment rate has sat below 50% since the year 2000, against a government target of 60%.

The government's response, the National Semiconductor Strategy, comes with a genuinely large RM25 billion behind it. But look at where the money and the headlines have actually gone: RM69.4 billion in newly approved semiconductor investment, a US$7 billion (roughly RM30 billion) Intel manufacturing plant. That's a strategy built overwhelmingly to attract more multinational factories to Malaysia — not to build the capability of the thousands of smaller local companies that already sit beneath the 350-plus MNC factories clustered around Penang and Kulim. Bringing in more anchors is one half of the equation. Building what sits underneath them is the other half, and it's the half that isn't getting anywhere near RM25 billion.

Where this bites hardest — the exact place AI needs it most

The clearest example of the gap is also, unhelpfully, the one closest to AI itself. Despite fifty years in semiconductor manufacturing and a 13% share of global chip assembly and testing, Malaysia currently has zero advanced packaging capability — the specialised process that stacks and connects chips together, and that's now essential for the high-bandwidth memory (HBM) chips powering AI hardware. Five Malaysian companies — Inari Amertron, Pentamaster, NSW Automation, SkyeChip, and a startup called FusionAP founded by former Intel engineers — have banded together as the Malaysia Advanced Packaging Consortium to try to build that capability themselves, targeting a modest 7% of the global advanced packaging market, worth an estimated US$5 billion a year, by 2035. Their first project targets HBM4, the memory format built specifically for next-generation AI chips.

It's a genuinely inspiring attempt at exactly the kind of coalition this argument keeps returning to — companies organising themselves to build a capability none of them could build alone. But it's revealing that even this effort is stuck on the same leg that tripped up Petronas's vendors before 1993 solved it: the consortium hasn't secured the roughly RM400 million needed for its pilot production line, and is having to acquire equipment incrementally instead. The industry can see what needs to be built. What's missing is the financing structured specifically for building it — the exact piece BNM-backed financing and bank support supplied for oil and gas vendors thirty years ago.

What's actually missing, side by side

Run this against the same four requirements: a credible anchor with contract-linked demand, capability-building matched to the real technical need, financing built for the transformation itself, and a coordinating layer tying it together. Penang and the National Semiconductor Strategy have plenty of anchors — Intel, Infineon, Micron, and the rest are all there, all real, all buying. What's missing is the deliberate mechanism connecting that anchor demand to the SMEs underneath it: no reserved scopes the way Petronas set aside 439 of them, no financing purpose-built for a smaller company absorbing advanced-packaging or AI-chip capability, and — Connect Day aside — no sustained coordinating body doing for semiconductor SMEs what SIRIM, MDEC, and TalentCorp did for oil and gas vendors over three decades, rather than one afternoon a year.

The question this leaves

Malaysia has the anchors. It has ambitious, self-organising SMEs willing to form their own consortium and go after the hardest, most AI-relevant part of the value chain, entirely on their own initiative. What it doesn't yet have is anyone playing Petronas's part — turning MNC demand into reserved, guaranteed opportunity, and backing the companies trying to meet it with financing built for that specific climb. A tenfold talent gap and a 15%-a-year brain drain don't wait for the next Connect Day. The honest question for Penang's semiconductor ecosystem is whether it's willing to build the other three pieces of the coalition it already half has — before the SMEs currently trying to do it on their own run out of runway.

Sources: MSIA "Connect Day" SME-MNC matchmaking event, The Star; Malaysia semiconductor talent gap figures, TechWire Asia; National Semiconductor Strategy funding and goals, CREST; Malaysia Advanced Packaging Consortium and HBM4 project, TechWire Asia.

← All notes