SMIC 65nm arrived this year, and the company lost its founder in the same stretch, which tells you most of what the year was about. The node is real, another rung above the 90nm the company hit in 2006, roughly where the leaders stood three or four years ago. But the headline out of Shanghai in 2009 is not the process. It is the courtroom, where TSMC finally won, and the price SMIC paid to settle.
A California jury found in early November that SMIC had misappropriated TSMC’s trade secrets, and the shares were suspended in Hong Kong the day the verdict landed. Rather than fight a damages phase it was likely to lose, SMIC settled within days. The terms are a humiliation dressed as a truce: 200 million dollars in cash, 1.79 billion new SMIC shares handed to TSMC with a warrant for nearly 700 million more, enough to make the competitor it was built to challenge an eight-percent owner of it, ten if the warrant is exercised, and the 2005 patent cross-license that had given SMIC some legal cover was torn up in the bargain. Richard Chang, who started the company on farmland outside Shanghai and drove it from nothing to 65nm in nine years, resigned the same week, with Applied Materials veteran David Wang stepping in to run what was left. The man and the method left together, because the method, catching up fast by absorbing what Taiwan knew, was exactly what the verdict punished.
The lawsuit did something larger than cost money and a founder. It put on the record, in a courtroom, the thing everyone knew and nobody said plainly: SMIC’s speed came substantially from knowledge that originated at TSMC. Strip that to its lesson and it is uncomfortable for the whole national project. The fast climb was partly borrowed, and borrowing has limits, legal ones now as well as practical ones. If SMIC is going to keep closing the gap, more of the closing will have to come from inside, from its own engineers solving problems the leaders already solved, the slow way, without the shortcut the court just closed.
A normal company might not survive this, the founder gone, a rival owning part of it, the core method discredited, no profit since 2007. SMIC survives because it is not really a normal company. The state wants a Chinese foundry, the state is patient, and a setback that would kill a startup is, for a national project, just a bad year. New leadership comes in, the losses keep getting absorbed, the fabs keep running. The foreign tools keep arriving, because in 2009 the global equipment market still sells to whoever pays, and SMIC still pays. The climb slows but does not stop. Beijing has too much riding on it to let it stop.
For all the drama, the competitive picture is sobering. SMIC at 65nm is still years behind TSMC, which is wrestling its own 40nm into yield while still pushing toward 28nm, running at scale the foundry model SMIC has only partly copied. The field is also consolidating above it: AMD’s spun-out fabs became GlobalFoundries this year and swallowed Chartered to jump to number three, while UMC absorbs the mainland foundry HeJian, leaving SMIC sliding toward fifth in a tier tidying itself up. The gap did not close this year; if anything the distraction of the lawsuit and the leadership upheaval let it widen. SMIC’s problem was always that it had to hit a moving target, and the target moved again while SMIC was in court. Catching up was never going to be a straight line. 2009 is the year the line bent backward.
The company that comes out of 2009 is chastened and slower and still, underneath, backed by a state that will not quit. Perso, I think SMIC recovers and resumes the climb, because the money and the will are intact even if the founder and the shortcut are gone, and a country this determined does not abandon its only foundry over one bad verdict. The path just got longer and lonelier. More of the work has to be original now, which means slower, which means the gap to TSMC may stay wide for years before it starts closing again. None of that changes the destination Beijing has in mind. It only changes the timeline, and timelines are the one thing a patient state can afford to be wrong about. SMIC will keep climbing. It will just take longer than anyone in Shanghai hoped.