I keep wondering, if it was right to stick around here in the US, or if I should have just packed up and moved to Shanghai.

The company I might have worked for announced earnings, and it's interesting sometimes when there is press on them. Yeah, I know they hired a lot of inexperienced engineers, and had a core group of engineers to train all the younger engineers. That's what it was like in China, they have people to debug their fab.

I agree with the article, they're probably ahead of their biggest competitor, but now that TSMC has shipped all their equipment from a fab in Taiwan, to replicate in Shanghai, there might be some real competition.

I don't know why I wanted to go out there anymore, but I do. Perhaps indeed it's a calling from God.

China's largest foundry poised to turn a profit
by Mike Clendenin, EETimes
Silicon Strategies
12/01/2003, 11:28 AM ET

SHANGHAI, China -- A year ago, not many companies were interested in working with Semiconductor Manufacturing International Corp., China's largest pure-play foundry.

Today, even with wafer prices rising, there's interest from Texas Instruments, Elpida, Infineon, Nvidia, Marvell, PMC-Sierra, Broadcom, Motorola and others. There is even word of potential joint ventures with STMicroelectronics and SanDisk.

And it's likely the foundry will turn an operating profit in December, which will help allay nagging doubts and criticism about its viability. SMIC is estimated to have brought in about $350 million in revenue this year-up from about $90 million in 2002; and it could triple that number next year as more capacity comes online.

"We have more and more customers supporting us and becoming long-term partners, so profitability will be relatively easy for us to maintain," said chief executive officer Richard Chang.

Getting to that transition point hasn't been easy. Chang has navigated his company through a minefield of obstacles inherent in the Chinese market, including inconsistent regulations, underdeveloped supply chains and inexperienced local workers.

SMIC hired a lot of young engineers with no experience and had to rely on a core group of foreign engineers, mostly Taiwanese, to train them. It took nine months. And much of that time was spent on teaching methodologies and changing attitudes, rather than on improving specific skills.

The senior engineers had to encourage the younger ones to follow the standard manufacturing recipes but also to question those procedures as they saw fit. "Sometimes the attitude in China is that people don't ask; they feel ashamed to ask questions. [So if young engineers had a problem] so they would just try to figure it out -- and then when they did [something differently from the established practice], the yield would go down," Chang said. The credo thus became: "If you have any questions, ask. If you have a good idea, propose it; but do not just change something without notifying people."

The company focused on DRAM and that product's sensitive manufacturing process to train its engineers and debug its fab. Meanwhile, it was also qualifying logic products; a process that took more than several months for some products. Slowly, things improved.

Now Chang is under pressure to expand SMIC's logic capacity. The requests come largely from foreign clients; few domestic design houses use the advanced processes. SMIC's acquisition of a Motorola fab in Tianjin was motivated in part by the need to ramp capacity for offshore demand. It is also in the market for used equipment, in part because lead times for new gear are stretching as the upturn takes root.

Chang said he would limit DRAM to less than 25 percent of the output at SMIC's current fabs. A 300-mm fab, now under construction in Beijing, would also start with DRAM and bring on logic products, Chang said.

Chang has negotiated orders from Infineon Technologies and Elpida Memory to get the fab into the black faster than if it had started on logic. It's a "small amount of profit initially, but it is still profitable. That's the key," he said.