Why HBM4 Matters for AI
AI chips need more than powerful processors.
They also need extremely fast memory.
HBM, or High Bandwidth Memory, sits very close to AI processors and allows huge amounts of data to move quickly between memory and computing units.
HBM4 is the next generation of that technology.
But placing processors, memory and multiple chiplets close together requires very advanced packaging.
That is the area Malaysia now wants to develop locally.
Five Malaysian Companies, Five Different Roles
The MAPC brings together companies with different semiconductor capabilities.
SkyeChip is responsible for chip and package design.
FusionAP focuses on advanced packaging process development.
Inari Technology provides assembly and testing capabilities.
Pentamaster contributes automated testing and inspection equipment.
NSW Automation contributes precision manufacturing and dispensing equipment.
Instead of one company trying to build everything alone, the five companies are building different parts of the same local supply chain.
RM185 Million Pilot Programme
The programme is receiving about RM92 million in government R&D funding, matched by approximately RM93 million from the participating companies.
The 24-month project began in early 2026.
The immediate target is to install pilot-line equipment at Inari by the end of 2026.
The actual SkyeChip test chip, silicon interposer and HBM components are then expected to move through the packaging process during 2027.
The full programme is scheduled to run until February 2028.
Why This Is Important for Malaysia
Malaysia already has a strong semiconductor industry.
But much of the country’s historical strength has been in assembly, packaging and testing for multinational companies.
The bigger opportunity is to own more of the technology, process knowledge and intellectual property.
Advanced packaging is especially important because modern AI chips increasingly combine:
- GPUs or AI accelerators
- HBM memory
- Chiplets
- Interposers
- High-speed connections
These components need to work together as one system.
If Malaysian companies can develop their own advanced packaging capability, they can participate in a much higher-value part of the AI semiconductor industry.
Malaysia Is Starting From a Small Base
The opportunity is large, but the challenge is also significant.
Industry representatives have said that local Malaysian companies currently have very limited independent advanced-packaging capability, even though multinational semiconductor companies already operate advanced processes in Malaysia.
The MAPC project is therefore not simply about producing one test chip.
It is about learning how to build the processes, equipment, skills and supply chain required for advanced packaging.
That knowledge could eventually be applied to other AI and high-performance computing products.
Closing Thoughts
Malaysia does not need to build the world’s next NVIDIA GPU to participate in the AI chip boom.
There are valuable opportunities throughout the semiconductor supply chain.
Advanced packaging is one of them.
If the MAPC pilot succeeds, Malaysia could move from mainly supporting global semiconductor production towards owning more of the technology behind next-generation AI hardware.
That would be a meaningful step up the semiconductor value chain.
