Keeping Intel lean and mean
Intel's analyst briefing day featured a parade of senior executives, many of whom aren't seen at other times of the year. One of them was Andy Bryant, the company's former CFO and now serving as its Chief Administrative Officer.
It's an odd title, he took a strange path to it. Bryant served as CFO for 14 years and had enormous cachet with the Wall Street types seated in the Intel auditorium, who spent as much time checking BlackBerries as listening to the spiels on stage.
Three years ago, Intel decided to bump him into this job because his understudy, Assistant CFO Stacy Smith, was clearly a talented and capable executive in his early 40s who was almost certain to be purloined by another Silicon Valley company for the CFO spot, and Bryant, in his mid-50s, was a long way from retirement.
Intel wasn't going to give the COO title to Bryant because he's not the heir (and if he were about five to seven years younger, he might very well be). So Smith got the big bean counter chair and Bryant got moved up into this odd new title, making him a COO in practice if not necessarily title. His talk was centered around Intel's operations, particularly its manufacturing processes.
He noted that in 2008, more than a dozen companies talked about shipping 32 nanometer design products. Only one company made it, and that's Intel. Intel's high-k metal gate transistor technology were the reason why, he said, because it allowed them to get down to that small process design.
This happened despite a massive and severe recession. During both the 2002 and 2008 recessions, Intel went through two rounds of layoffs, but the firm always preserved engineering headcount.
Instead, it attacked non-essential spending and protected its most important investments. The company accelerated the move to 45nm and 32nm, the introduction of the new Sandy Bridge architecture, and the Atom processor. "All the things you are hearing about today are things we focused on two years ago," he said. "It's not enough to survive, you have to thrive and do that through innovation and investing."
The efficiency of high-k metal gate technologies means that, if Intel were to go back in time technologically to 2006, when it was making chips at 65nm, it would need to spend $40 billion to build four more factories and need four times the staff to produce the volume of desktop, server, Atom and other chips as it makes today.
The newer manufacturing process means a 50 percent reduction time to ramp up manufacturing over the old 90nm process. Intel has reduced delivery time by 50 percent and is three time faster in responding to orders. It used to be if an OEM wanted to place an order for chips, Intel had to get back to them a few days later. Now it can tell them almost instantly.
Intel also has the ability to recover faster. The recession of 2008-09 was marked by a massive drop in sales in the first quarter of 2009, followed by a near total recovery by the end of the year.
If Intel had to respond to 2009's recovery with the manufacturing technology it had in 2003, it would have taken so much longer to ramp up production the company would have lost $1 billion in sales.
It's getting harder and harder to for competitors to maintain that level of competition, too. He notes that the bar for companies keeps going up. A company needs to maintain a 50 percent margin in order to have enough money to reinvest in its manufacturing process, and to do that, it has to be at least a $10 billion company, said Bryant.
That means massive investment, and Intel has fewer competitors left. Samsung is the only company that comes close. It will spend $6 billion on manufacturing process technology this year, topping even Intel and TSMC, both of which will spend $4.8 billion on new manufacturing technology this year.