Lu Chaoqun, founder and chairman of Etron Technology, said on the 6th that there is no problem with Taiwan’s OEM, but software understanding is the key. In particular, although Taiwan's semiconductor industry is firmly at the core of the AI wa...
Lu Chaoqun, founder and chairman of Etron Technology, said on the 6th that there is no problem with Taiwan’s OEM, but software understanding is the key. In particular, although Taiwan's semiconductor industry is firmly at the core of the AI wave, there are gaps in the understanding of software and algorithms that are difficult to ignore.
Lu Chaoqun said at the "2025 Vision Summit" that everyone knows the hardware architecture of AI. For example, the logic operation of the chip must be fast, the memory must be fast, and the transmission must be stable. However, without corresponding software understanding, these hardware cannot exert its true value. However, companies such as OpenAI, Anthropic, and xAI can dominate the industry not because they can make chips, but because they understand algorithms and models and can drive the entire system.
Lu Chaoqun pointed out that Taiwan has long established advantages in "manufacturing-oriented" thinking, from TSMC's advanced processes, Quanta's customer service, to a complete supply chain for heat dissipation and packaging. Regarding the links that Taiwan is still missing, he admitted that Taiwan is ready for the AI era, except for memory and software. And this is cultural. Americans like to be heroes and want to be OpenAI CEO Sam Altman, but no one wants to go to the factory to do it. Taiwan, on the other hand, is capable and willing to do it, but does not necessarily dare to dream.
Talking about the reorganization of power in the AI industry, Lu Chaoqun pointed out as an example that NVIDIA (NVIDIA) invested US$100 billion in OpenAI, and OpenAI then purchased computing power from Oracle (Oracle), which in turn purchased NVIDIA chips, forming a highly strategic cluster operation. Lu Chaoqun emphasized that this is not going around in circles, but a capital strategy in the AI era. Taiwan must also establish its own settlements, otherwise it will always be just a part of the OEM industry.
Therefore, Lu Chaoqun proposed a specific idea to establish a "Sovereign R&D Industry Execution Center." Funding does not rely on government subsidies, but taxes are allocated from profits from the semiconductor and AI industries, and funds are earmarked for AI language models and independent technology development. Lu Chaoqun emphasized that the need to create Taiwan’s sovereign AI is not just a Chinese model, but a complete ecosystem that combines local industry, culture, and applications.
