A fab success
The smartphone boom has been a boon for a pioneer in semiconductors
WHEN he founded Taiwan Semiconductor Manufacturing Company in 1987, Morris Chang recalls,
Nobody thought we were going anywhere.
Back then the rule was that semiconductor companies both designed and made chips.
TSMC was the first pure foundry, making chips for designers with no factories, or fabs, of their own.
The doubts of others suited TSMC nicely.
Mr Chang, at 82 still chairman and in his second stint as chief executive, says that meant it suffered no competition in its first eight years.
These days the idea is more popular.
Last year foundries made about half of all logic chips.
United Microelectronics Corporation, a slightly older Taiwanese company, turned itself into a pure-play foundry in 1995.
GlobalFoundries, with factories in America, Germany and Singapore, was set up in 2009.
Yet the pioneer still dominates.
This year, predicts Samuel Wang of Gartner, a research firm, TSMC's revenues will exceed those of all other foundries combined.
He reckons it has 90% of the world market for advanced 28-nanometre chips, which are essential to smartphones and tablets.
TSMC's sales in the second quarter, reported on July 18th, were NT156 billion, 21% more than a year before. Its net income rose by 24%, to NT52 billion.
That said, growth should slow in the second half of the year, Mr Chang told analysts, because smartphone-makers have been building up their inventories of chips ahead of new-product launches and because sales of PCs and some smartphones have been weaker than expected.
Sales could even shrink in the fourth quarter.
TSMC has thrived on a mixture of serendipity and anticipation.
The market moved in the direction in which we were heading, says Mr Chang.
The boom in mobile computing has meant a bonanza for several long-standing customers, which range from Qualcomm, an American firm that dominates the market for smartphones' application processors, to MediaTek, a Taiwanese neighbour that has burst onto the scene more recently.
Sales to Chinese designers such as Allwinner, Rockchip and Spreadtrum, whose chips power inexpensive Android tablets and phones, have doubled in the past year;
these all license low-energy chip technology from ARM, a British firm that has been a partner of TSMC's since 2004.
The grand alliance of TSMC and the chip designers, as Mr Chang has called it, has done far better from the shift to mobile devices than Intel, the world's biggest chipmaker.
Intel both designs and makes chips.
Its processor chips are the brains of most personal computers, but demand for these has been falling.
Although Intel's newest processors are much less thirsty and the firm is determined to break into the mobile market, it has so far struggled.
Its second-quarter revenues fell by 5% and its net income by 29%.
A combination of fabless designers and a pure foundry works well, Mr Chang explains,
because the designers don't have to worry about the capital-intensive part of the business any more: the foundry provides the scale.
At the same time, fabless companies, of which he reckons there are about 50 with annual revenues of at least 100m, compete with one another and with IDMs.
This diverse group has in total produced more innovation than any single IDM…Just look at the mobile products, he says.
The battle lines between the alliance and its rivals continue to shift.
Intel has entered the foundry business—and recently snaffled a contract with Altera, an American designer that has been a customer of TSMC's for 20 years.
I really regret that very much, Mr Chang says. I had an investigation into why that happened.
On another front, TSMC is fighting Samsung.
The South Korean company makes processors not only for itself but also for Apple, even though the two are deadly rivals in smartphones and tablets.
In addition it is the world's biggest maker of memory chips.
Recently the Wall Street Journal reported that TSMC, after years of effort, had won a contract to make chips for iPhones and iPads—a victory that will have stung Samsung.
Mr Chang says that he takes neither Intel nor Samsung lightly, but insists: We have one big advantage: we are a pure foundry. We do not compete with our customers.
No matter how hard Intel and Samsung try, he says, they will not enjoy the same trust.
The big question hanging over TSMC is whether it can find a worthy successor to its octogenarian boss.
It has had one false start already:
he stood down as chief executive in 2005, but took the helm again in 2009.
No handover is planned yet,
but last year three men were appointed co-chief operating officers.
Probably at least two of these, or maybe all three, will end up as joint chief executives.
A menage a trois could prove farcical. But perhaps TSMC's board thinks its founder is too good an act for one man to follow.