Semiconductor R&D Crisis?

Rising costs and a consolidating industry are forcing companies to rethink where to place their dollars; Europe and Asia step up investments.


Research and development is a sometimes forgotten but critical element in the semiconductor industry. The delicate R&D ecosystem enables many of the key breakthroughs in the business.

But there could be a troubling trend, if not a crisis, brewing on two fronts in the R&D arena. On one front, R&D costs for semiconductor technologies are escalating at each node. Higher R&D costs are not only making it more difficult for vendors to move to the next node, but are also putting a strain on budgets and contributing to the current wave of consolidation in the IC industry.

On another front, Asian and European R&D organizations are expanding their efforts while the United States is taking a step back. Case in point: Sematech, a major R&D chip consortium in the U.S., is falling by the wayside, at least as a standalone organization.

Sematech’s issues began showing up in March, when Intel confirmed it had exited from the R&D consortium. Subsequently, Samsung and TSMC have pulled out of Sematech, leaving GlobalFoundries as one of the few remaining members in the organization, according to multiple sources in the industry.

Then, in May of 2015, Sematech lost its status as a standalone chip consortium—it was unceremoniously folded into the State University of New York Polytechnic Institute (SUNY Poly), a high-tech university system in New York. Sematech still exists within SUNY Poly, but it is following a new and untested charter, sources said. Sematech and SUNY Poly declined to comment.

Intel, Samsung and TSMC are still part of the Global 450 Consortium (G450C), which is a 450mm effort under the auspices of SUNY Poly but separate from Sematech. G450C is still active, although 450mm technology has been put on hold indefinitely.

In any case, Sematech’s former charter as a member-based chip consortium simply ran its course. Fewer chipmakers are at the leading edge, so vendors must rationalize their R&D dollars, and work with fewer research partners, to avoid duplication.

For example, Sematech’s current and former members, such as GlobalFoundries, Intel, Micron, Samsung, TSMC and others, currently belong to a rival R&D organization—Belgium-based IMEC. In fact, many hi-tech multinationals are not only flocking to Imec, but also to other non-U.S. research organizations, such as A*Star and CEA-Leti. These organizations are luring multinationals with fresh funds and state-of-the-art facilities.

“It’s interesting to note that Imec has been a relative success. Sematech is not doing quite as well and is sort of on its way out. Recently, Intel announced its departure. Samsung also announced its departure from Sematech,” said Robert Maire, president of Semiconductor Advisors, a consulting firm. “I find it interesting that we haven’t been able to support an organization, at least in an ongoing basis in the United States that is a clearing house for research much like what Imec does in Europe. I don’t know if it can be fixed, or will be fixed, but certainly it’s an ongoing issue.”

Others have a different perspective regarding the state of R&D in the U.S. “If you look at the companies in the U.S., you think of Intel and Micron. Similarly, if you look at the fabless side, you have Qualcomm and others. From that perspective, the U.S. is definitely not losing its edge and it won’t lose that anytime soon,” said Risto Puhakka, president of VLSI Research.

“The real issue is what’s happening at the university and national institute level in the U.S. regarding fundamental research,” Puhakka said. “Is there any interest from the U.S. government perspective to grow and fund that type of activity? This is the question here. The companies have been forced to pick up the slack.”

R&D crisis?
Puhakka is referring to one troubling R&D trend: U.S. federal funding for total R&D is projected to reach $133.7 billion in 2015, up a mere 2% from 2014, according to a new report from the National Science Foundation (NSF). Within those figures, U.S. funding for basic research, the lifeblood of R&D, is projected to fall by 1% in 2015, according to the NSF.

On the bright side, the top-10 chipmakers worldwide spent a total of $31.8 billion in R&D alone in 2014, up 11% over 2013, according to IC Insights. R&D as a percentage of sales remained healthy at 16.9%.

But over time, it’s become more difficult for chipmakers to stay on the leading edge. Soaring IC design and manufacturing costs are only part of the problem. “Higher R&D costs mean fewer companies can continue to innovate, which leads to increased incentives to consolidate,” according to a recent report from AlixPartners, a consulting firm.

Indeed, the IC industry is experiencing a new round of merger and acquisition (M&A) activity. And based on the trends, the M&A activity could accelerate.

For example, on average, IC design costs are projected to rise from $30 million for 28nm, to $80 million for 14nm, to $120 million for 10nm, according to Gartner.

Process technology costs alone are increasing by about 35% at each node, according to AlixPartners. The cost to develop a 16nm/14nm process is about $1.3 billion, compared to $900 million for 28nm, according to GlobalFoundries.

On the fab tool front, the development cost is also high for CVD, etch and related gear. “If you look at how much it costs to develop a new set of equipment, it’s typically $100 million,” said Jeff Marks, vice president of emerging businesses at Lam Research, in a recent presentation.

And amazingly, the industry has spent between $14 billion to $21 billion to develop extreme ultraviolet (EUV) lithography over the years, analysts said. Despite that investment, EUV is still not commercially viable.

So all told, chipmakers and fab tool vendors must carefully allocate their R&D dollars and make the right bets, in a challenging environment. And as the costs escalate, the industry must collaborate more than ever before, said E.S. Jung, executive vice president of the semiconductor R&D center at Samsung. “We are doing three nodes at the same time in our R&D center,” Jung said at a recent event. “How can we make it happen? We need tools, materials and open innovation. Also, we cannot do it all by ourselves.”

In fact, most vendors work with various R&D partners. But much has changed from the glory days of the central research labs in the U.S., when Bell Labs and IBM dominated the R&D landscape.

In the 1980s, Sematech and SRC emerged in the U.S. And in Europe, Imec rose from near obscurity to become one of the most prominent sources of advanced semiconductor R&D. In addition, France’s CEA-Leti continues to expand. And in 2014, the European Union launched the Horizon 2020 program, which is Europe’s biggest hi-tech R&D effort ever with nearly €80 billion (US $88.9 billion) of funding available over the next seven years.

Asia continues to be a hot bed of R&D activity. Last year, for example, A*Star, a Singaporean R&D organization, and a bevy of multinationals launched four R&D labs in the following areas in Singapore — lithography, metrology, packaging and assembly. This year, A*Star and others launched nanoimprint lithography and fan-out packaging consortiums in Singapore.

“If you look outside the U.S., particularly in China and Korea, they are investing heavily in the infrastructure of the semiconductor industry. China is putting more than $100 billion in the semiconductor industry,” Lam’s Marks said. “What we are seeing is a competitive dynamic that is going to be harder for the U.S. to stay competitive going forward, unless we make more investments as a country in this space.”

What about Sematech?
In the U.S., it’s a mixed bag. U.S. chipmakers and fab tool vendors continue to invest in R&D. The SRC continues to hold its own. And earlier this year, several companies launched a silicon carbide (SiC) manufacturing consortium in the U.S.

On the other hand, U.S.-based government labs and universities face more budget cutbacks. And the other question is what happened to Sematech?

Alarmed by share losses to Japanese chipmakers in the 1980s, the U.S. semiconductor industry responded by forming Sematech in 1987. Sematech’s initial members included IBM, Intel, HP, Micron and Texas Instruments. As part of its mission, Sematech hoped to pool their R&D resources, and develop new technologies, to regain lost share in both the chip and fab equipment segments.

Having accomplished many of its goals by the 1990s, Sematech altered its charter. Over time, it made the transition from a U.S.-centric consortium to an international R&D powerhouse. It enrolled chipmakers and tool vendors from Asia and Europe.

Sematech, however, ran into various issues, when it recently made a transition from a self-supporting, member-based organization to a public-private funding model under the state of New York, according to one former Sematech member. “The authority of its long-time industry members was marginalized. Collaboration stalled, and the consortium fell apart and was ultimately dissolved,” the source said.

“Intel is very practical. So are TSMC and Samsung. The members were probably questioning that if we are putting a (certain) amount of money in here, then what are we getting in return,” VLSI Research’s Puhakka said. “The world is also changing around Sematech. Finding partners is a hard thing to do, given the consolidation in the industry.”

Now, as part of its new and untested charter, Sematech is being folded into SUNY Poly. SUNY Poly will continue to conduct basic R&D. “SUNY will get access to the manufacturing assets from Sematech,” Puhakka said. “Fundamentally, it’s a major re-tooling of Sematech’s business model. It used to be a membership-driven business model. Now, it’s becoming more of a project-driven business model. Sematech has various centers of excellence. Companies can sign up and gain access to the results and IP from those projects.”

Sematech, according to Puhakka, will continue with its advanced CMOS, lithography and fab equipment programs. It will also add solar to the list.

The continuation of these programs depends on several factors, however. “Once they move into the project-oriented format, there will be projects that won’t be supported. Then, there are new projects that will be supported. It’s a market-driven type approach. That will define how successful they are going to be,” Puhakka said.

The problem for Sematech, as well as other worldwide R&D organizations, is that there is a finite number of R&D dollars to go around. Organizations will have to fight for those dollars, as the IC industry matures and consolidates. In addition, they must compete for funds with faster growing, and perhaps more critical, industries such as life sciences, energy and security.

  • Scott

    There is much R & D going on with nano-IC technology. These nano size IC’s are being produced down to 10 nm by etching the transistors with a process known as reactive ion etching. Researchers have been able to us a plasma system to get the heat and the power consumption down to impressive levels. Let’s hope this technology becomes widespread.

  • Dev Gupta

    Mark as usual a great article ! Can you list Sematech’s significant technical accomplishments in the last 20 years and who have been the principal beneficiaries ?

  • SemiMike

    Motorola was early member of Sematech and benefitted slightly by some of their training materials, and equipment characterization efforts. But IBM and TI dominated, and Intel was not very open (and now has dropped out) on many key technology issues. One good thing I remember was the APC Workshops (I attended about 13 of those over many years) where real-time control systems data integration with near-realtime modeling tools became a strong toolset.

    SEMI and SEMATECH cooperated and the Technology Roadmaps evolved, but it was the learning-in-the-effort that mattered more than the final roadmaps. The industry was “growing up” and is now “mature” by many measurements, but remains innovative, just very expensive, and thus divided into legacy and “bleeding edge” operations. R/D in general has been best supported by near-monopolies in most industries (IBM in mainframes, ATT’s Bell Labs in phone and solid-state technologies). Consortiums or Alliances often degenerate into games of “liars poker” as in 450 MM efforts.

    Here is to Jean Hoerni whose planar process patent saved Fairchild for a while, and make the IC manufacturable by everyone who licensed that patent.