Consolidation in the semiconductor industry continues apace, with more than $100 billion in mergers and acquisitions announced in 2015, and more to come in 2016. “With our industry growth rates being so low, it’s a lot cheaper to acquire market share than it is to invest and beat your competitor over the head,” said analyst Bill McClean, speaking at SEMI’s Industry Strategy Symposium (ISS) in January.
One potentially negative impact of consolidation is reduced innovation, said Ivo J. Raaijmakers, Chief Technology Officer and Director of R&D at ASM International, speaking at ISS on consolidation in the equipment supplier market. “The tail has been cut off. A lot of innovation happens in this tail,” he said. “The question we have to ask ourselves is how can be ensure efficient innovation in such a consolidating landscape of equipment suppliers?”
The is compounded by exponentially increasing complexity and R&D spending. The diversity in the number of materials used is also increasing rapidly. This is important, since it most future innovations will require the use of these new materials.
“More and more innovations are needed per node, per year,” Raaijmakers said. “Life was easy in the ‘60s, ‘70s and ‘80s. In the ‘90s, there were silicides added. In the 2000s, there were tantalum oxides, spin-on glasses and copper and low k. In 2010, we had high k and metal gate. We had porous low k materials, new barriers, SOI coming in. in 2015, we saw a lot of materials used for patterning. In 2020, we will attack the channel material, going from silicon to germanium to III-V materials and all associated materials.”
Raaijmakers provided an equation that captures the mathematics of innovation:
dI/dt α I x η/τ
where I is the number of innovations being worked on: loosely relates to R&D budget
η is the average success rate: what fraction of projects are successful, and
τ is the time constant: how long does it take from innovation to production
Since R&D budgets are fixed at about 15% of the overall revenue, the only two knobs to turn are the success rate and affect the time constant. “We are in deep trouble,” he said, “unless we manage η and τ.”
Raaijmakers noted that industry consolidation lowers the number of innovation projects. At the same time, success rate decreases with complexity as does development time.
The time factor does not offer much potential. On average, most successful innovations take 7-10 years from concept to high volume manufacturing. “Can we decrease this by collaboration along the value chain?” Raaijmakers asked. “I think it will be difficult and if you can do it, it will not be a huge gain.”
On the other hand, there’s much to be gained by the other factor, which is increasing the efficiency. “Collaboration along innovation chain can significantly lower the risk of adoption. Are we all working on the right things to push through into manufacturing. How quick can you narrow down choices?” said Raaijmakers. He added the such collaboration may increase the speed, but that’s not the major effect.
Collaboration across the chain increases R&D leverage and improves efficiency. “That means that two parts in the supply chain are not doing competing things or different things. They are working on one solution which is the solution for the industry.”
He said successful companies will bring R&D much closer to the fab and high volume manufacturing, but while not forgetting R&D. “You have to be good at taking things into volume production and supporting it there. And you have to be good in R&D. Maintaining those two traits is not so easy, he said.
Given the high cost to bring a device from development to market of up to 50 million or more in today’s paradigm one might argue that there can only be innovation with the scale to support it so consolidation is needed. Although as the article suggests the temptation might be to have the innovation be closer to fab process for efficiency, yield etc… rather than real breakthrough application innovation.
That being said history shows time and again that technology disruption will come about that will destroy existing paradigms and perhaps the massive post consolidation companies will struggle with a big infrastructure based in old models. Kind of like what is happening with cloud computing for tradition software and even hardware companies.
Consolidations happen in mature industries because in many cases innovation has slowed down and growth to keep investors happy happens only through M&A. This in turn reduces innovation further since competition is reduced. Its a vicious cycle.