SMIC Founder Says Chasing 3nm and 2nm Alone Is a “Misconception” as He Pushes Niche Semiconductor Markets as the Real Breakthrough Path
SMIC founder Richard Chang is pushing back against one of the most common assumptions in the chip industry, arguing that semiconductor success should not be judged only by whether a company reaches 3nm or 2nm. In a newly reported interview covered by MyDrivers, Chang said it is a misconception to treat advanced process competition as the only meaningful measure of success, and instead argued that real industry contribution can also come from mastering a specific niche and solving a major bottleneck problem.
According to the report, Chang believes the current industry conversation has become too concentrated around advanced nodes and large scale compute, while many niche semiconductor segments that are still strategically important have been neglected for years. He said advanced process nodes account for less than 20% of the market by product volume, while more than 80% of demand comes from mature processes and specialty technologies. That is the foundation of his argument. Rather than trying to compete head on in every popular segment, he says domestic chipmakers should focus on the missing links and underserved sub markets where overseas suppliers still dominate and where local companies may have a more realistic chance to break through.
Chang’s framing is notably pragmatic. He said China’s semiconductor industry does not need to do everything at once or aim to be number 1 in every category. Instead, he argues that prioritization matters more, and that becoming exceptional in one focused segment can already be a major contribution to the broader industry. In his view, crowding into the same high profile tracks creates homogeneous competition, while deeper work in overlooked areas may deliver more practical and durable value.
That same logic also extends to AI. The interview says Chang sees the current AI race as too focused on cloud side large model infrastructure, while edge and distributed AI applications remain underappreciated. He divides AI development into 2 directions, one being cloud based large model and supercomputing platforms with extremely high capital barriers, and the other being scenario based distributed AI deployments where practical applications still have huge room to grow. He argues that the second path offers a stronger opportunity for startups and domestic innovators, especially in areas such as industrial control, automotive electronics, and wearable devices, where there is still significant unmet demand for hardware adapted to real world use cases.
This is what makes Chang’s comments more interesting than a simple defense of older process nodes. He is not saying leading edge manufacturing does not matter. He is saying that too many people mistake leading edge manufacturing for the only path that matters. His real point is about strategic positioning. If the largest global attention is fixed on 3nm and 2nm, then the more open opportunities may lie in mature nodes, specialty processes, and application driven chip design where barriers are different and competitive gaps remain easier to identify.
For SMIC, that perspective also lines up with present reality. The company remains under pressure from export restrictions and equipment limits, and the report notes that China’s largest foundry is still constrained by lack of access to advanced EUV machinery, leaving it dependent on older DUV equipment and limiting it to less advanced production. In that context, Chang’s comments read less like a rejection of advanced technology and more like an argument for pursuing the most achievable and strategically useful wins first.
From an industry standpoint, this is a meaningful reminder that semiconductor competition is broader than the leading edge node race. The glamour and investor excitement may sit around 3nm, 2nm, and AI training silicon, but the market itself is still heavily supported by mature node demand, specialty chips, and scenario specific hardware. Chang’s argument is that solving those gaps can matter just as much, and in some cases may be the more realistic way to create national capability and long term value.
Do you think the semiconductor industry puts too much attention on 3nm and 2nm prestige, or is leading edge manufacturing still the only benchmark that truly matters?
