It’s no news the Foundry business is consolidating. It’s been foreseen for years and has largely played out as expected, but, with one very big difference; Global Foundries (GF). Personally, I never sat through a panel discussion or read an article that said a fund from Abu Dhabi was going to make a run at TSMC. Yet, that’s exactly where we are.
And a run is what it will be. That is, to catch the undisputed foundry leader and take sustainable share for a relevant timeframe (defined as a period equal to what a Wall Street cynic would say constitutes a viable company). Even a year ago there were many of us that thought it would be too daunting a task to annoy, let alone challenge, TSMC. But some things are shaping up differently than maybe even the GF team anticipated.
Let’s take a look. First, GF may have started with the right first move; AMD. Unburdened by massive commercial overhead and commitment, yet buoyed by strong technology and predictable demand, the first leg of the stool has proven to be solid and manageable. AMD has run quality fabs and needed only to support themselves. Leg two, Chartered, has brought a merchant interface expertise but with technology that lags AMD. The market facing skills of Chartered may or may not be usable by the more advanced AMD fabs but it is a start. At least there is a small ecosystem that has developed tools, equipment, IP and design expertise for Chartered. (BTW, leg three will come fast and there are only a few choices left.)
This may relieve the Achilles Heel of GF, the ecosystem. But it is a far cry from the cosmic alignment of the moons and stars that supports TSMC. Simply, TSMC has it all. Further, TSMC has set the standard for customer service, quality and delivery in the twenty year old foundry segment. They will be virtually impossible to overcome on this front.
But does GF care? It’s doubtful GF is running around courting startups to make their one, high risk, dubious market IC. In fact, I am certain they are not. They need volume and they need it now. And those players with real volume control their own ecosystems. They don’t need 3rd party IP or libraries and even their processes are self-directed. More importantly, the big guys want GF to gain significant share and they’ll make sure they get it; and it’s not because they’re great guys. It’s because they can’t afford an ever growing TSMC to get too big and with any more market control.
No doubt, TSMC is the best. Everyone knows it. There is no debate. But no consumer buying 10K wafers per month or more will support a situation where they have no choice but to buy from Big T. They will create and prolong competition and the only thing GF has to do to benefit is to be good enough; a standard much lower than that set by TSMC.
The semiconductor industry has matured and there will be a new set of winners and losers. It appears Global Foundries may have drifted (or driven) into a place where their handcrafted stool may be just strong enough to support the long term plans of a customer base looking for an alternative to the TSMC chaise longue. In any case, it will be geek drama of the highest order and an endless source of speculation. Game on.
