Malaysia Makes 13% of Global Chips. The AI Boom Depends on It
Quick summary
Malaysia produces 13% of global chips. Intel, Infineon, TI, and Chipbond run major packaging and testing in Penang. NSS secured RM63B. What developers should know.
Read next
- SK Hynix and Samsung Are the Only HBM3E Suppliers. That Is the AI Bottleneck.SK Hynix holds ~60% of HBM and has sold out through 2026. Only two firms make HBM3E at scale. HBM market to hit $54-58B in 2026. What developers should do.
- AI Data Centers Are Consuming 70% of the World RAM. Memory Prices Have Quadrupled. Your Next Laptop Will Cost More.Data centers will consume 70% of all memory chips produced in 2026. DRAM prices are up 300-400% from mid-2025. Budget PCs under $500 may disappear. Here is what is happening and what developers and buyers should do.
Malaysia makes roughly 13% of global chip output by some measures and is the largest semiconductor packaging and testing hub outside the United States. Intel, Infineon, Texas Instruments, and STMicroelectronics run their biggest back-end facilities there. Under the National Semiconductor Strategy (NSS) launched in May 2024, Malaysia had secured over RM63 billion in semiconductor investments by March 2025, less than a year in. The AI GPU supply chain runs through Penang, and almost no English-language tech coverage explains why that matters.
Why Malaysia Dominates Chip Packaging and Testing
Packaging and testing happen after a wafer is cut into dies. The dies get mounted, wired, and sealed; then they are tested. Most of that work is done in Asia, and Malaysia does more of it than anyone except Taiwan. Penang alone hosts dozens of fabs and OSAT (outsourced semiconductor assembly and test) plants. The country has been in the game since the 1970s, when Intel set up in Penang. Today the ecosystem includes substrate suppliers, equipment vendors, and chemical suppliers, plus a workforce that has been doing this for decades.
Intel Penang and Kulim are among Intel's largest manufacturing sites globally. Infineon is building what it calls the world's largest 200 mm silicon carbide power fab in Kulim. Texas Instruments, STMicroelectronics, ON Semiconductor, and NXP all have major Malaysian operations. When an Nvidia H100 or AMD MI300 is assembled, a large share of that assembly, testing, or supply chain touches Malaysia.
The Numbers: What Is Actually There
- 13%: Malaysia's share of global semiconductor output by volume (packaging and testing heavy)
- RM63 billion: investments committed under the National Semiconductor Strategy as of March 2025 (RM58 billion foreign, RM5 billion domestic)
- RM575 billion: Malaysia semiconductor export value in 2024 (roughly USD 130 billion)
- 13: number of Malaysian companies the NSS aims to develop as national champions (e.g. Carsem, plus Infineon and NXP expansions)
- RM1.2 billion: government allocation over five years to train 60,000 engineers for the industry
- RM800 million: Chipbond Technology advanced packaging and testing facility opened in Penang (Batu Kawan) in early 2026, with initial capacity of 10,000 wafers per month and 100 million wafer-level chip-scale packaging units
Chipbond is Taiwanese. Its new Penang plant is a signal that advanced packaging is spreading beyond Taiwan as Taiwan capacity stays maxed out. Malaysia is the obvious next stop: same time zone, existing OSAT base, and government incentives.
The AI Boom Runs Through Penang
AI GPUs need leading-edge logic fabs (mostly Taiwan and South Korea) and advanced packaging: CoWoS, InFO, 2.5D and 3D stacking that connect the GPU die to HBM. A lot of that packaging capacity still sits in Taiwan, but the testing, substrate supply, and a growing share of assembly are in Malaysia. If you are building AI infrastructure, your H100s and H200s have almost certainly passed through Malaysian facilities or Malaysian-dependent supply chains at some point.
That is why the country matters for the AI boom even though it does not make the logic chips themselves. Any serious supply chain or risk assessment for AI hardware has to include Malaysia and Penang.
What the National Semiconductor Strategy Is Doing
The NSS is not just marketing. It has attracted real capital and set concrete targets. The 13 national champion companies include homegrown OSATs and the big multinationals expanding in Malaysia. The 60,000-engineer target addresses the main bottleneck: skilled labour. Infineon Kulim SiC fab is the kind of project that locks in Malaysia as a long-term node for power semiconductors, which matter for EVs and data centre power as much as for logic.
For developers and infrastructure teams, the takeaway is geographic: the GPU you rent or buy depends on a chain that runs through Taiwan, South Korea, and Malaysia. Supply chain diversification talks often focus on fabs. Packaging and testing are just as concentrated, and Malaysia is at the centre of that concentration. Regional disruption (natural disaster, geopolitical shock, or labour shortage) would affect lead times and availability globally.
Key Takeaways
- 13%: Malaysia's share of global chip output; largest packaging and testing hub outside the US
- RM63 billion: committed semiconductor investments under the National Semiconductor Strategy as of March 2025
- Penang: core hub; Chipbond opened RM800 million advanced packaging facility there in early 2026
- Intel, Infineon, Texas Instruments: largest back-end or expansion facilities in Malaysia
- For developers: AI GPU supply chains depend on Malaysian packaging and testing; factor Malaysia into any supply chain or risk assessment for AI hardware
- What to watch: How much advanced packaging (CoWoS-style) moves to Malaysia in the next 3 to 5 years as Taiwan capacity stays maxed out; NSS national champion progress
FAQ
Frequently Asked Questions
How much of the world chips does Malaysia make?
Malaysia produces roughly 13% of global chip output by volume. It is the largest semiconductor packaging and testing hub outside the United States, with the bulk of that activity in Penang.
Which chip companies have major facilities in Malaysia?
Intel, Infineon, Texas Instruments, STMicroelectronics, ON Semiconductor, and NXP all run large packaging, testing, or manufacturing operations in Malaysia. Intel has been there since the 1970s and counts Penang and Kulim among its biggest global sites.
Why does the AI GPU supply chain depend on Malaysia?
AI GPUs need advanced packaging (CoWoS, 2.5D, 3D) and testing after the logic die is made. A large share of that work is done in Asia, with Malaysia as a central hub. H100s and H200s typically pass through Malaysian or Malaysia-linked facilities in the supply chain.
What is OSAT and why does it matter for Malaysia?
OSAT stands for outsourced semiconductor assembly and test. It is the industry that packages and tests chips after fabrication. Malaysia is one of the biggest OSAT bases in the world, which is why it has such a large share of global chip output even without leading-edge logic fabs.
Free Weekly Briefing
The AI & Dev Briefing
One honest email a week — what actually matters in AI and software engineering. No noise, no sponsored content. Read by developers across 30+ countries.
No spam. Unsubscribe anytime.
More on Hardware
All posts →SK Hynix and Samsung Are the Only HBM3E Suppliers. That Is the AI Bottleneck.
SK Hynix holds ~60% of HBM and has sold out through 2026. Only two firms make HBM3E at scale. HBM market to hit $54-58B in 2026. What developers should do.
AI Data Centers Are Consuming 70% of the World RAM. Memory Prices Have Quadrupled. Your Next Laptop Will Cost More.
Data centers will consume 70% of all memory chips produced in 2026. DRAM prices are up 300-400% from mid-2025. Budget PCs under $500 may disappear. Here is what is happening and what developers and buyers should do.
Japan Is Spending $67 Billion to Rebuild Its Chip Industry in 10 Years
Japan is investing $67B over 10 years. Rapidus targets 2nm in Hokkaido by 2027; TSMC Kumamoto is producing and building a 3nm fab. What it means for the supply chain.
Nvidia Rubin Has 288GB of HBM4 Memory — And It Will Make Your Laptop More Expensive
Nvidia's next-generation Rubin GPU architecture will pack 288GB of HBM4 memory — 3.6x more than H100's 80GB. But the DRAM supply squeeze caused by AI chip demand is already pushing up prices for consumer laptops, phones, and PCs.
Free Tool
Will AI replace your job?
4 questions. Get a personalised developer risk score based on your stack, role, and what you actually build day to day.
Check Your AI Risk Score →Written by
Software Engineer based in Delhi, India. Writes about AI models, semiconductor supply chains, and tech geopolitics — covering the intersection of infrastructure and global events. 941+ posts cited by ChatGPT, Perplexity, and Gemini. Read in 167 countries.
