What challenges do domestic chips face?

What Challenges Do Domestic Chips Face, and Why Can't China Produce High-End Chips?

The current state of the domestic chip industry shows that the self-sufficiency rate for high-end general-purpose chips is nearly zero. Overall, the composition structure of domestic and imported chips is roughly as follows: the proportion of chip domestication is very low, below 20%, and most chips must be purchased from abroad. There are about 40 global companies leading in general and key components.

There is a large trade deficit in integrated circuit imports and exports. In 2016, China imported $227.07 billion in integrated circuits, exceeding $200 billion for four consecutive years, making it the most valuable import commodity. At the same time, exports of integrated circuits were $61.38 billion, down 11.1%, resulting in a trade deficit of up to $165.7 billion. It is expected that integrated circuit imports will remain at a high level for the next few years.

Domestic chip manufacturing focuses on a few areas, such as certain communication chips, display processing chips, power management chips, discrete devices, microcontrollers (MCUs), and navigation chips. Most of these are in the low-to-mid-end, with low unit prices and profit margins, while mid-to-high-end performance chips and key components are monopolized by foreign manufacturers. Looking at the entire chip industry, domestic companies are not weaker than foreign manufacturers in the packaging and testing fields. Some digital chip design areas are not behind foreign manufacturers, but analog chip design remains far behind, and chip manufacturing equipment and processes lag foreign IDM or foundry companies by 1-2 generations.

Four Major Challenges Faced by Domestic Chip Manufacturers

1. Accumulation of Long-Term R&D Investment and High Tolerance: This manifests in the lack of microarchitecture design and low-level operating system design capabilities, with general CPUs lacking their own microarchitecture (most domestic PC/server operating systems are still based on Linux). Foreign ARM-like manufacturers experienced over 20 years of R&D accumulation before seeing success, or quickly introduced and grabbed top chip design talent.

2. Achieving a Positive Cycle of Heavy Capital Investment and High Output.

3. Surpassing Foreign Competitors in Performance and Stability in the Short Term.

4. Cultivating a Hardware Developer Ecosystem: Intel and Microsoft have developed curriculum systems, certification systems, and ecosystem cultivation systems in domestic universities for many years. Domestic companies rarely have such cross-level strategic operations.

Why Can't China Design Powerful Chips?

Currently, China's integrated circuit industry involves design, manufacturing, and packaging, with manufacturing available through foundries, and packaging and testing not far behind the United States. The biggest gap lies in design, with domestic chip design still heavily relying on foreign sources. An expert from a Beijing research institute said, "It's not that China can't design chips; it's that there's no environment for chip iteration."

"Companies like Intel and ARM, their first-generation chips were difficult to use, but they could iterate, leading to better chips eventually."

However, domestic conditions are different. Domestic chips have a hard time surviving because the industry doesn't give them a chance to iterate. On one hand, superior chips already exist in the market, often with lower costs. On the other hand, people lack patience for domestic chips to iterate, which directly limits China's ability to improve chip design.

"Many people could design great chips, but because the market doesn't give domestic chips a chance to iterate, they are often ignored, and domestic companies can't or won't give engineers a chance to 'sacrifice,' which is our problem."

Another reason why high-end general-purpose chips aren't produced domestically is a lack of sufficient R&D investment. Except for major national science and technology projects, other national science and technology plans don't include integrated circuit-related projects or funding. Two major national science and technology projects launched in 2008, "Core Electronic Devices, High-End General-Purpose Chips, and Basic Software Products" and "Very Large-Scale Integrated Circuit Equipment and Complete Process Technology," invest only 4-5 billion yuan per year in integrated circuits, representing just 5.2% to 7.7% of Intel's annual R&D expenses.

This is similar to why China can't produce DSLR cameras. The journey toward chip domestication may not be as distant as it seems, but it will require patience, significant investment, and an opportunity to iterate.