SMIC Founder Says Chip Success Isn’t Just About 2nm — And SEA Should Pay Attention
The chip industry loves one flex above all: smaller nodes. 3nm, 2nm, next-gen wafers, Apple chips, NVIDIA AI hardware — that is usually where all the attention goes.
But SMIC founder Richard Chang thinks that view is too narrow.
In a new interview, Chang argued that semiconductor success should not be measured only by whether a company can reach cutting-edge 3nm or 2nm production. According to him, that mindset misses a much bigger part of the market: mature manufacturing processes.
Mature chips still run most of the world
Chang said advanced manufacturing accounts for less than 20% of overall semiconductor demand, while mature processes make up more than 80% of the market.
That is a big reminder, especially for readers in Malaysia and SEA. Not every important chip is powering an AI server or flagship gaming GPU. A lot of real-world demand comes from less glamorous hardware: car electronics, routers, appliances, industrial systems, budget phones, payment terminals, IoT devices, display controllers, and plenty of the stuff that quietly keeps daily tech running.
For Malaysia, this matters because we already sit in an important part of the global semiconductor chain, especially assembly, testing, packaging, and electronics manufacturing. The hype may be around 2nm AI accelerators, but the business opportunity for SEA is often in the practical chips that companies actually need in volume.
SMIC may be looking away from the 2nm race
TSMC’s rise has been built on elite execution at advanced nodes, with major customers like Apple and NVIDIA relying on its ability to stabilise difficult processes quickly. That side of the market is still extremely important, especially with the current AI boom.
But Chang’s point is that not every chip company needs to fight on the same battlefield.
He suggested that domestic semiconductor companies should look at niche areas where overseas firms currently dominate. Instead of trying to do everything at once, companies should identify specific bottlenecks and solve them properly.
That is a more pragmatic route: pick one missing link, get excellent at it, and build from there.
Why this connects to AI hardware too
Chang also pointed out that much of the AI chip conversation is centred on cloud computing. Big data centres, giant GPU clusters, trillion-dollar companies — memang all the spotlight is there.
But distributed AI is a different story. That means AI running closer to users, devices, factories, cars, cameras, stores, and local systems instead of everything being processed in massive cloud setups.
For SEA, that could be huge. Think AI features in affordable devices, smart city systems, retail cameras, factory automation, logistics, agriculture tech, or even gaming and creator hardware that does not depend fully on cloud servers. These scenario-based applications may not require the absolute latest 2nm node, but they do need reliable, cost-effective chips.
Chang also warned AI startups against simply burning huge amounts of money trying to challenge the biggest players head-on. His view is that startups should find specific use cases and build around them instead.
The China chip problem is still real
Of course, SMIC is not suddenly on equal footing with TSMC or Samsung. China’s chip industry still faces serious limits because SMIC does not have access to advanced EUV lithography tools, which are needed for 5nm and below.
Because of that, SMIC remains tied to older DUV equipment and is limited around the 7nm class. That makes competing directly with TSMC’s most advanced production extremely difficult.
But Chang’s argument is basically this: if the cutting edge is blocked, win somewhere else.
And honestly, that is not a bad read. The world still needs a massive amount of mature-node chips. For Malaysia and SEA, where cost, supply stability, and practical hardware matter a lot, the mature-chip market may end up being just as important as the headline-grabbing AI race.
Source: Wccftech Gaming
