commercial partnerships and international research collaborations.

2.2 Increased International Collaboration

Data on coauthored papers presented at several of the sampled conferences1 discussed above and in Appendix F were used to examine how international collaborations have changed over time. In the network connectivity graphs2 of Figure 2-1, the nodes (circles) represent individual countries, and the size of each node represents the number of papers produced by that country. The edges (lines connecting two circles) represent collaborations on coauthored papers, and the weight of each edge indicates the number of papers that share coauthorship between nations.

In each area except architecture, the network graphs show an increasing geographical diversity in research and a tremendous increase in international collaborations. The network graphs show that between 1996 and 2011, international participation and collaboration between the United States and other nations has dramatically increased. In the devices and circuits areas, many of the international collaborations come from work that spans multiple international sites within the same company. This trend toward greater collaboration across national boundaries will likely continue due to the increasing global investments in research by both nations and global industries.

International research collaborations in computer architecture have not increased dramatically, although more papers are being published as collaborations between U.S. and foreign researchers. The emergence of the ARM architecture in the mobile computing space provides impetus for foreign investment in architecture research, particularly in Europe, as European funding agencies prefer to invest in activities that are synergistic with European-based technologies.

Today, leading U.S. universities are linking to remote campuses in Asia and Europe and are describing themselves as “global universities.” This trend, as well as the growing number of global companies, may have an impact on future U.S. competitiveness.

2.3 Commercialization of Technologies

This section provides a snapshot of the global landscape in the commercialization of semiconductor and computing hardware and software technologies using data from iSuppli, Gartner, the Hardware Top 100, and the Software Top 100.3

2.3.1 Semiconductor Commercialization

The committee began by analyzing revenues from the largest semiconductor, as well as computing hardware and software, companies. Table 2-1 shows the top 20 semiconductor companies, ranked by 2010 revenues4 and includes companies that sell semiconductor components.5 The chart shows the nation where the company is headquartered, its primary technology area, whether it has its own in-house fabrication capability, 2010 revenues in U.S. dollars, and the fraction of the global semiconductor market.

These top 20 companies account for a total of $197 billion, which is about two-thirds of the global semiconductor market. Of these top 20 companies, the United States accounts for 47 percent of revenue. Japan and Korea account for about 20 percent each, while Europe accounts for 10 percent. Historically, being a major semiconductor company required owning and operating significant semiconductor fabrication factories. However, the rising cost of deploying such facilities, both in R&D and capital investments, combined with the availability of “fab-for-hire” foundry services from companies such as Taiwan Semiconductor Manufacturing Corporation (TSMC), have given rise to an increasing number of fabless6 semiconductor companies. Foundries such as TSMC have grown to be about 10 percent of the overall semiconductor component market (Gartner7 estimate is U.S. $28.3

_______________

1Conferences included in each of the four technology areas are as follows: (1) Architecture: ASPLOS, HPCA, ISCA, and MICRO; (2) Programming: ECOOP, OOPSLA, PLDI, POPL, and PPoPP; (3) Applications: SIGGRAPH, SC, VLDB, and WWW; and (4) Semiconductor Devices and Circuits: IEDM and ISSCC.

2Coauthor networks were generated with the Science of Science (Sci2) Tool: Sci2 Team (2009). Science of Science (Sci2) Tool. Indiana University and SciTech Strategies, http://sci2.cns.iu.edu.

3See www.isuppli.com; www.gartner.com; www.hardwaretop100.org; and www.softwaretop100.org.

4See http://www.isuppli.com/Semiconductor-Value-Chain/News/Pages/Intel-Reasserts-Semiconductor-Market-Leadership-in-2011.aspx. Last accessed on August 16, 2012.

5Companies that supply only fabrication services (such as TSMC with 2010 revenues of over $13 billion) are not included. Systems companies that design their own chips (such as Apple) are included in Table 2-1 below.

6Fabless semiconductor companies specialize in the design and sale of hardware devices and semiconductor chips, as opposed to device fabrication.

7“Semiconductor foundry revenue increased 40.5%, reaching $28.3 billion in 2010. Foundry fab utilization reached its peak in 3Q10 after several quarters of good growth. Leading-edge technologies (65 nm to 45 nm) have been in high demand from foundries, increasing in revenue contribution.” Available at http://www.gartner.com/id=1634315. Last accessed on February 7, 2012.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement