Universities, Industry, and Government:
A Complex Partnership Yielding
Innovation and Leadership
One measure of the impact of investment in information technology research and development is its contribution to the creation of numerous U.S. firms with annual revenues exceeding $1 billion and of entire new sectors that contribute billions of dollars to the U.S. economy.8 Many of these firms are household names, and their products and services underpin the digital economy—and indeed the economy more broadly. The combined estimated annual revenue of only the companies listed on Figure 1 is nearly $500 billion (Table 1).
Figure 1, an update of the 1995 “tire tracks” figure9 and the intermediate 2003 version,10 illustrates, through examples, how fundamental research in IT, conducted in industry and universities, has led to the introduction of entirely new product categories that ultimately became billion-dollar industries. It reflects a complex research environment in which concurrent advances in multiple subfields—in particular within computer science and engineering but extending into other fields, too, from electrical engineering to psychology—have been mutually reinforcing, stimulating and enabling one another and leading to vibrant, innovative industries exemplified by top-performing U.S. firms.11Figure 1 is of necessity incomplete and symbolic in nature; it would be impossible to chart all of the important cumulative contributions of research and their links to today’s products, firms, and industries. For example, Google could be thought of as having benefited from at least three research areas—networking, parallel and distributed systems, and databases.
Listed in the bottom row of Figure 1 are areas where major investments in basic research in subfields of computing and communications have had the impacts shown in the upper portions of the figure. Not depicted but equally important is research on the theoretical and algorithmic foundations of computing more broadly (Box 1). The vertical red tracks represent university-based (and largely federally funded) research, and the blue tracks represent industry R&D (some of which is also government funded). The dashed black lines indicate periods following the introduction of significant commercial products resulting from this research, the green lines represent billion-dollar-plus industries (by annual revenue) stemming from this research, and the thick green lines represent achievement of multibillion-dollar markets by some of the industries. The top rows list the present-day IT market segments and representative U.S. firms and products whose creation was stimulated by the decades-long research represented by the red and blue vertical tracks.
FIGURE 1 Examples of the contributions of federally supported fundamental research to the creation of IT sectors, firms, and products with large economic impact. Tracks added since the 2003 update of the figure are described in Appendix B. See also Box 1 and Appendix C.
TABLE 1 Annual Revenue Associated with the IT Industry Sector for Key U.S. IT Firms Listed in Figure 1
Industry | Company and Estimated Revenue ($ Billion)a | |
Broadband and Mobile |
Qualcomm* |
11 |
Microprocessors |
nVidia |
3.5 |
Personal Computing |
Dell |
34 |
Internet and Web |
Juniper |
4.4 |
Cloud Computing |
Google (non-advertising)* |
1.1 |
Enterprise Systems |
Oracle* |
31 |
Entertainment and Design |
Electronic Arts* |
3.6 |
Robotics and Assistive Technologies |
iRobot* |
0.4 |
NOTE: Revenues are for FY 2011 except as indicated by an asterisk for firms whose listed revenues are 2010-based.
aSources for estimated revenue listed are given in the section “Notes” following the main text of this report.
BOX 1
Research in the Theoretical and Algorithmic Foundations of Computing
Are there problems that simply cannot be solved by a computer algorithm? If so, what are they, and why is this so? For the problems that can be solved, how efficiently (in terms of time, memory, or communications requirements) can this be done? And for those that can’t be solved, can we make practical use of this fact, for example, to help ensure better privacy on our computer systems?
These are some of the most basic questions in computing. Research to address such questions is often motivated by the desire to understand the basic nature of computation rather than to find practical applications. However, time and again discoveries are made that provide new ways to solve difficult algorithmic problems. For example, research in coding theory, which investigates the fundamental limits in the encoding and decoding of messages, has led to methods for transmitting messages in ways that are highly tolerant of faulty communications channels, and ultimately to methods that achieve very close to the maximum possible efficiency and provide a foundation for nearly all of today’s wireless technologies, ranging from mobile phones, to WiFi, to deep-space communications.
The impact of theoretical and algorithmic research is wide-ranging. Algorithms for network congestion provide the key building block for today’s content-distribution networks. Modern logistics systems, such as those used by the airline industry or package delivery systems, depend on a deep understanding of the limits of computation and algorithms for optimal allocation of resources and for scheduling. All modern search engines make use of fundamental knowledge of how mathematical concepts such as eigenvalues can be used to rank Web pages. All electronic commerce today is built on foundational concepts of so-called one-way functions, developed in some of the most theoretical computing research endeavors. And today’s speech and natural language understanding systems apply large-scale statistical analysis algorithms in sophisticated ways.
Additional examples of the impacts from algorithms research are provided in Appendix C.
Although the tracks in Figure 1 were chosen to illustrate through prominent examples how each selected research area is connected with a closely linked industry area, in reality each research area is linked in many ways to one or more industry areas. Research outcomes in one area have continued to affect and enable research in other areas. Furthermore, synergies among research areas often lead to surprising results and have impacts on industry that were not originally intended or envisioned (Table 2). This characteristic of technological innovation is most evident in the broad-based impact of research on basic questions in computing. Such research often starts as a search for fundamental knowledge but time and again produces practical technologies that enable significant economic impact, in areas as diverse as optimal resource allocation and scheduling, compact encodings of signals, efficient search algorithms, fair auction and voting mechanisms, and ultralarge-scale statistical analyses. (Box 1 provides further discussion.)
The arrows between the vertical tracks represent some salient examples of the rich interplay between academic research, industry research, and products and indicate the cross-fertilization resulting from multi-directional flows of ideas, technologies, and people (examples are given in Appendix B). Also illustrated in Figure 1 is how products arising from industry can shape academic research. (For example, Microsoft’s Kinect sensor is now being used in many research applications, and Google’s practical application of MapReduce introduced new ideas about web-scale distributed computing to the research community.) Arrows spanning research areas provide a few indications of the interdependence of research advances in various areas.
TABLE 2 Original Goals, Unanticipated Results, Current Results, and Possible Future Directions for Research Topics in Figure 1
Research Topic | Original Goal | Unanticipated Results |
Digital Communications | Untethered communication | Wireless local area networking for computers, cell phones |
Computer Architecture | Tools to manage increasing complexity of microprocessor designs; new architectures to dramatically increase processing power | Powerful computation in things such as cars, televisions, kitchen appliances, and mobile devices |
Software Technologies | More effective use of computing power for specific tasks, and the creation of common systems on which to run them | Open-source movement that inspired many to gain powerful technical skills and become entrepreneurs; the ability to create software systems of extraordinary scale and complexity |
Networking | Sharing computational resources and data among computers | Network e-mail; widespread sharing of software and data; the interconnection of billions of computers and other devices |
Parallel and Distributed Systems | Using multiple computers and/or processors to solve a complex problem | Emergence of businesses such as Google and Amazon that use multiple very large data centers to deliver services at large scale |
Databases | Tools for managing, discovering, and locating information | Search engines, digital libraries, and data mining and analytics on massive data sets; advances in databases that have led to the development of enormous data repositories—improving knowledge and supporting new forms of scientific discovery |
Computer Graphics | Display of real-time graphics and text on an external screen | Graphical user interfaces; techniques for realistic modeling and simulation applied for near-realistic video games and movies; support by these technologies for applications in training and scientific exploration |
Artificial Intelligence and Robotics | Simulation of human-level intelligence, including language understanding, vision, learning, and planning | Robotic-enabled prosthetics and artificial organs; fly-by-wire avionics and antilock brakes; cars capable of parallel parking themselves; intelligent ranking of Web search results |
aSources for details listed are given in the section “Notes” following the main text of this report.
Todaya | Advances Expected with Continued Commitment to IT Researcha |
Wireless and broadband industry. Nearly 6 billion cell phone subscribers worldwide, including more than 320 million subscribers in the United States; 54% of the U.S. population as active mobile-broadband subscribers, and more than 87 million fixed broadband subscriptions | Pervasive/ubiquitous communications and access to data and computing resources; mobile sensors for monitoring environment and health in real time |
Microprocessor industry. 8.3 billion microprocessors produced annually and used pervasively; $40 billion in annual revenue | Increased interplay between hardware and software to achieve performance while managing power and providing easy programmability |
Personal computing industry. 1.4 billion PCs in use worldwide as of 2010; U.S. smartphone sales expected to be nearly 100 million in 2011 | Parallel software to better use parallel hardware; improved tools for processing very large data sets |
Internet and Web industries. One-third of the world’s population is online, and 45% of those are under the age of 25; more than 18 billion searches were conducted in October 2011 in the United States (across five major search sites); U.S. retail e-commerce sales for the third quarter of 2011 were $48.2 billion and accounted for 4.6% of total sales; worldwide, annual e-commerce sales were almost $8 trillion; banking, trading, and other financial transactions done by means of the Internet | “Internet of things” (virtually every device/object networked); sensors embedded everywhere, enabling dramatic improvements in automation, efficiency, and safety |
Cloud computing industry, an emerging and rapidly growing industry sector. Health IT alone expected to spend more than $1 billion on cloud services by 2013; enterprise spending on public cloud computing services expected to expand 139% from 2010 to 2011 | Renewal of efforts in parallelism to sustain growth in computing performance; improvements in scalability with reductions in operating costs for very large data centers |
Enterprise systems industry. Widespread use of enterprise resource planning software; world’s largest civilian database, Walmart’s data warehouse, stores more than 583 terabytes of sales and inventory data built on a massively parallel system | Natural language searches, data management to promote energy-efficient computing, cloud-based data analyses in heterogeneous environments, and other large-scale data management systems |
Entertainment industry. CGI movie “Toy Story 3” the highest-grossing film of 2010; 12 feature-length computer-generated-imagery animated films released in 2011; modeling and simulation commonplace in manufacturing and engineering; video games using advanced computer simulation techniques | Tying visualization of large data sets to the simulation code, increased use of augmented reality, search based on images; photography becoming computational |
Robotics and sensing industries. Automation commonplace in manufacturing and in specialties such as robot-assisted surgery; use of aerial drones for surveillance becoming commonplace; some household robots | Artificial intelligence agents capable of abstraction and generalization beyond their initial programming; “household” robots for more than vacuuming |
Consider how research in the leftmost area in Figure 1—digital communications—has propelled the communications revolution that continues to unfold today:
• Code division multiple access, which had origins in World War II anti-jamming technology and later was used in military communications satellites, was developed and commercialized as a new standard for cellular telephony in the 1990s by Qualcomm, a company founded by DARPA-funded university researchers. It uses unique mathematical codes to modulate transmissions, thus allowing multiple users to efficiently share a radio channel and providing relative immunity to interference.
• Research in the 1990s on multiple-input and multiple-output techniques, beginning with closely related university research and followed by research at Bell Laboratories, has been a fundamental enabler of today’s wireless communications technologies.
• Research and serious engineering efforts in universities through the 1990s led to the ability to use complementary metal oxide semiconductor technology for radio-frequency signals, a development that made it possible to include WiFi, GPS, and Bluetooth at low cost in small, mobile devices.
• Early academic research into packet switched networks provided an underpinning for the local area networks that connect computers within homes and businesses as well as for the Internet that links the globe.
• A university spin-off company developed and commercialized a practical approach to digital subscriber line (DSL) technology, which made it possible to provide high-speed data networking over public telephone network lines.
A similar list could be constructed for each of the research areas represented in Figure 1. As Figure 1 and Table 2 illustrate, investments started more than four decades ago have been critical enablers of the products and services in use today. They also illustrate how research can yield important results not originally contemplated when a first investment was made. Finally, they describe some of the open questions that researchers pursue today and suggest some of the potential applications that lie ahead provided there is a continued commitment to IT research.