SMIC has just reported its results
and, once again, they failed to impress. China's largest chipmaker did
make a profit but that was largely due to one-off gain from the sale of
used chip-making equipment -- one of the consequences of being in a
rapidly evolving industry.
BusinessWeek's Bruce Einhorn writes here about the problems that China's SMIC -- and, by extension, other emerging market chipmakers -- have competing in a capital-intensive industry where even the established giants struggle to consistently make money.
It was AMD's founder Jerry Sanders who once said "only real men have fabs" -- thanks to Mike Magee's The Register for the attribution -- and the phrase neatly sums up the enduring fascination that semiconductor companies still have for owning "fabs" -- the capital-intensive fabrication plants where silicon wafers get turned into chips -- despite the ample evidence that owning a fab is akin to pouring money into a bottomless pit.
The huge cost of new fabs -- more than a billion dollars -- means their owners are involved a race against time to get yields up and recoup the investment before the technology becomes obsolete. It is a high-risk game best left to a handful of experienced giants -- Taiwan's TSMC and Intel of the US, for example.
It is particularly hard to see the economic rationale for a developing nation like China to be pouring money into building fabs. SMIC has earmarked $720m this year on capital expenditure to expand production capacity.
Nevertheless, SMIC seems intent on being a foundry player and the industrial logic is that it can help China's local chip companies at an early stage of their development -- the subtext being that they would get treated worse by foreign foundries. And as Bruce Einhorn notes, "in China, at least, there's no shortage of governments willing to throw around money to get into the chip business."
But I can't help thinking the money would be better spent if it was used to boost China's nascent fabless semiconductor design sector. After all, the most consistently successful chip companies in the west are fabless, meaning they outsource manufacturing to a "foundries" like TSMC, thus leaves them free to concentrate on design.
These companies have realised that the real value of a chip business is not sunk into plant and equipment but rather lies in its intellectual property.
