Van Beurden then asked about problems with innovating out of trouble when chip development is getting more expensive, quoting costs of $25 million at 45-nm through $50 million at 32-nm and going as high as $110 million as the cost of developing a chip at 22-nm. The cost is going up demanding higher sales volumes to justify development which means that the number of markets that can support chip development is diminishing, he asserted.
"This may be true in VLSI but it's not true in automotive, industrial, sensors and medical," said Bozotti. He continued by pointing out that his business stands on two pillars. The one comprises sensors and power and the other is multimedia convergence. "The first side is healthy. The second side is challenged."
Bozotti also said that the major part of the cost is in the essential software that goes with a so-called platform offering. "The major part of the spending is in software development. Producing an LTE solution is very expensive." The answer to these problems is to achieve the right mix of market focus and partnering, Bozotti added. "The other ingredient is speed of execution." He praised his company's progress in MEMS – where is has gone from zero to $900 million in revenue in a few years – as an example of speed. "The difference between success and failure is speed."
Van Beurden returned to the topic of manufacturing and noted that ST runs a mixed business where it outsources some leading-edge digital manufacture but retains manufacturing for other areas. "There is a lot of added value in manufacturing otherwise I would outsource MEMS making. In the other areas differentiation is not there. And getting to 20-nm process requires collaboration," said Bozotti.
So van Beurden then asked whether, as Mark Bohr of Intel had elsewhere suggested, the fabless business model was collapsing (see Intel executive says fabless model "collapsing"). "I don't think it's true. Both will coexist; fabless companies will remain successful."
Joep van Beurden, chairman of the GSA, interviews Carlo Bozotti, CEO of STMicroelectronics on stage at the GSA European Executive Forum.
The discussion then turned to industry relevance and whether chip making gets its share of attention from politicians and young people. "The semiconductor is ubiquitous but it is a secret," said van Beurden.
Bozotti responded by saying he works for ERT, the European Roundtable of CEOs that comprises about 50 CEOs who are seeking to address a chronic shortage of science, technology, engineering and math (STEM) graduates that the continent faces. "We need hundreds of thousands of students over the next few years. That's why we've created the Ingenious program to try and persuade high school students to take up STEM and provide a better understanding of the challenges of globalization. ERT is trying to reach 40,000 students with specialist courses and contests," Bozotti said. "But it is also important that people, politicians understand that every dollar of semiconductors sold drives $22 of additional products and services."
I think companies like Quallcom, Samsung and Nvidia have broadly captured the processor market for mobile phones and tablets. It could e a touch ride for ST to come back in this area. But their analog business is definitely on a positive growth.
Yes you can argue that most of ST's problems stem from a past over-reliance on Nokia. As Nokia's fortunes have tumbled so have ST's.
But there are still question marks about the fab-lite strategy. And whether ST should still be trying to be a broad supplier.
It is noticable that NXP and Infineon have downsized considerably and seem to be doing the better for it.
Can ST-Ericsson compete with Qualcomm, Intel, Samsung without a significant boost from someone?
I believe ST has what it takes to make successful Mobile SoCs with their advanced power management techniques and modem IPs. What is missing is a long term partnership with a "successful" cellphone maker.
Bozotti might have a dream, but many dreams are not relating to reality that well...we have two horses: more than Moore (their analog and MEMs biz) and more Moore (their digital biz including Ericsson venture)...riding both is tough abd very few companies can pull that off, most just ride one...Kris
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.