ITIF Search

Becoming America’s Seed Fund: Why NSF’s SBIR Program Should Be a Model for the Rest of Government

Becoming America’s Seed Fund: Why NSF’s SBIR Program Should Be a Model for the Rest of Government
September 26, 2019

The National Science Foundation’s Small Business Innovation Research (SBIR) program focuses on high-growth startups and commercializing federally funded research and development (R&D). Other federal agencies should consider emulating this model.


SBIR is a federal program coordinated by the Small Business Administration that funds small business R&D. The program has helped seed companies such as Qualcomm, Symantec, and 23andMe, and has been copied by 17 countries around the world.
Over the past two decades, the National Science Foundation (NSF) reinvented its SBIR program to specifically target growth-focused startups and to emphasize commercializing innovations derived from federal R&D.
Other agencies should follow NSF’s model and do more to target growth-focused companies, centralize program management, hire dedicated program directors, and coordinate SBIR awards with other agency programs that provide support for commercialization.
Congress should reform SBIR funding to grant agencies more autonomy, require agencies prioritize commercialization potential in funding decisions, allow awardees to use funds for commercialization activities, and increase federal funding for R&D.


The Issue Facing Innovative Small Businesses: A Capital Gap

The SBIR Program


Impacts of NSF’s SBIR Program

Implications of the NSF Model for Other Agencies


Appendix: Companies and Agencies Interviewed




If the United States is to solve its most pressing challenges, including sustaining prosperity, improving public health, ensuring national security, and combating climate change, it must expand, diversify, and accelerate innovation. Small businesses with big ambitions to develop and commercialize new technologies have the potential to play these critical roles in America’s innovation ecosystem. But many are not able to realize this potential due to a lack of early-stage capital.

The federal government’s Small Business Innovation Research (SBIR) and Small Business Technology Transfer (STTR) programs are designed to help fill this gap, but they do so imperfectly. The authorizing legislation for SBIR and STTR—which are typically considered one program and will be referred to here as SBIR, for simplicity—lays out several goals: stimulating technological innovation, addressing federal research and development (R&D) needs, supporting social and economic diversity among small businesses, and, finally, commercializing new technologies. The agencies implementing the program have substantial discretion to prioritize these goals.

NSF reinvented its SBIR program over the past two decades to focus more sharply on one of the program’s objectives: commercializing innovations derived from federal R&D.

The National Science Foundation (NSF) has an effective program that is worthy of examination—and perhaps emulation—across much of the federal enterprise. NSF reinvented its SBIR program over the past two decades to focus more sharply on one of the program’s objectives: commercializing innovations derived from federal R&D. And over the past five years alone, NSF’s SBIR awardees have received $6.5 billion in private investment and had 87 exits.[1] We draw on original interviews with program stakeholders, quantitative data, and prior research to further illustrate NSF’s progress toward promoting commercial success among small businesses.

The report concludes by setting forth a model, drawn from the NSF experience, SBIR programs in other agencies should consider. The approach would more effectively stimulate private-sector commercialization of SBIR-funded R&D, thereby enhancing the impact of this federal investment to benefit society. It involves:

  • Targeting growth-focused companies;
  • Centralizing program management;
  • Hiring dedicated program directors; and
  • Coordinating SBIR awards with other agency programs that provide support for commercialization.

We also describe steps Congress should take to support agencies seeking to implement the model, notably:

  • Reforming SBIR funding to allow agencies more autonomy;
  • Requiring agencies to increase the weight of projects’ commercialization potential in funding decisions;
  • Allowing small businesses to use a portion of SBIR awards for commercialization activities; and
  • Increasing overall federal funding for R&D.

The Issue Facing Innovative Small Businesses: A Capital Gap

Solving the nation’s most pressing challenges requires bold technological innovations to be brought from conception to commercialization. Improving health outcomes depends in part on the development of new therapies and medical devices. Eliminating carbon emissions requires innovations in energy supply, management, and use. Creating whole new industries, which is central to the U.S. economic model, requires a robust innovation ecosystem.[2]

New growth-oriented, technology-driven businesses play a key part in this innovation ecosystem, whether they ultimately grow large themselves or their assets are eventually acquired by a large firm. Such innovative start-ups often pursue ideas more established firms fail to spot or won’t invest in. They may also pursue ideas very similar to those of larger competitors, but do so more nimbly or creatively. Each wave of information technology (IT) over the past few decades, to take the most prominent example, has been characterized by the emergence of start-ups that grew to be major employers and household names—from Intel in the 1960s and 1970s to Apple, Microsoft, and Qualcomm in the 1980s and 1990s to Google, Facebook, and Amazon in the 2000s and 2010s.[3]

However, a substantial body of research suggests that ambitious new businesses often face daunting barriers when they seek investment. To be sure, these ventures are risky and ought to pay a premium relative to more established and diversified companies to cover that risk. But even when analysts adjust for risk, “severe financial frictions,” as New York University economist Sabrina Howell put it, that disadvantage such firms remain.[4] 

Borrowing from banks is rarely an option for innovative new businesses because they typically lack collateral and revenue. Venture capital fills the capital gap in some sectors, particularly IT, where upfront investments are modest, and returns can be extremely large and relatively quick. However, in many other economic sectors, as well as in many regions of the country outside of technology hubs, venture capital is not available. For instance, as a recent working paper from the MIT Energy Initiative puts it, venture capital is “the wrong model” for clean energy hardware start-ups, because the upfront investment is large, and the returns may take a decade or more to materialize.[5]

Some indicators suggest that the capital gap may be growing, particularly for the earliest-stage start-ups. Venture capital has been flowing toward deals that are larger and later stage, often after the viability of a product has already been demonstrated. The number of seed-stage deals declined approximately 46 percent over the past three years, while overall venture funding has increased.[6] Moreover, only 20 percent of such funds have gone to ventures outside the IT sector.[7]

Many states have established seed-stage investment funds to aid local entrepreneurs who want to start and grow innovative companies. These funds tend to be modest in scale and constrained in scope; they have barely made a dent in the problem. The federal government’s SBIR program, which the U.S. Small Business Administration (SBA) has labeled “America’s Seed Fund,” deploys more than $3 billion annually, and has the potential to fill a much larger portion of this early-stage capital gap.

The SBIR Program

NSF program officer Roland Tibbets developed SBIR for NSF in 1977, after which Senator Edward Kennedy (D-MA), with SBA’s backing, spearheaded a push to expand it across the federal government. Partly in response to concerns about growing foreign economic competition, President Reagan signed the resulting Small Business Innovation Development Act into law in 1982.[8] A decade later, Congress enacted the STTR program, which differs from SBIR only in that it requires small business awardees to partner with research institutions.[9] The current authorizing language for both programs lies in the Small Business Act.[10]

SBIR has four statutory objectives:

  1. Stimulate technological innovation;
  2. Use small business to meet federal R&D needs;
  3. Foster and encourage participation by socially and economically disadvantaged small business concerns (SBCs) and women-owned SBCs in technological innovation; and
  4. Increase private-sector commercialization of innovations derived from federal R&D.[11]

Program Structure and Funding

Federal agencies that support over $100 million per year in extramural research (i.e., research conducted outside of the agency) are required to commit at least 3.2 percent of these funds to SBIR, while agencies with extramural research budgets over $1 billion are required to devote an additional 0.45 percent of these funds to STTR.[12] Five federal agencies with large extramural research budgets provide both SBIR and STTR awards and collectively fund 97.5 percent of the program: the Department of Defense (DOD), Department of Energy (DOE), Department of Health and Human Services (HHS), National Aeronautics and Space Administration (NASA), and NSF. Six other agencies fund the remaining 2.5 percent of the program, providing SBIR awards but no STTR awards: the Department of Agriculture, Department of Commerce, Department of Education, Department of Homeland Security, Department of Transportation, and Environmental Protection Agency.

Figure 1 shows total funding provided by SBIR since its founding. The program obligated over $3.1 billion in fiscal year (FY) 2018. 

Figure 1: Total funding for the SBIR program across all agencies has increased to $3.1 billion as of FY 2018. The awarded amount is shown for 2014 and beforehand, and the obligated amount is shown for 2015 and thereafter[13]

SBA serves as the coordinating agency for the SBIR program, although it does not fund R&D. It reviews the implementing agencies’ progress, and reports on the program to Congress. Most important, SBA is responsible for issuing a policy directive that outlines implementation guidance for agencies, which was updated most recently in May 2019. The policy directive provides specific instructions on a range of topics, such as the timelines on which solicitations must be issued and reviewed; how to minimize regulatory burdens; reporting requirements; and procedures to ensure awardees are more likely to receive additional agency funding beyond the scope of the SBIR program.[14]

To be eligible for SBIR, an applicant must qualify as an SBC.[15] An SBC is a for-profit entity with a place of business in the United States, over 50 percent U.S. ownership, and no more than 500 employees. In 2018, federal agencies awarded nearly 6,000 SBIR awards to over 3,000 SBCs.

As stated in the program objectives, the SBIR program endeavors to increase participation by women, socially or economically disadvantaged individuals, and businesses in underrepresented areas. The SBA’s Federal and State Technology (FAST) Partnership Program funds state and regional efforts to increase the number of applications from these groups to the SBIR program.[16]

A key feature of the SBIR program—and one that is common across awards from all participating agencies—is its three-phase model. Table 1 shows the objectives, eligible applicants, maximum award amounts, and award durations for each SBIR phase. In brief, Phase I awards help agencies determine the feasibility of a project, and Phase II awards enable companies to further carry out their projects, complete the R&D requested of them by the agency, and develop a technology that may be commercialized. These awards each require separate applications to an agency’s SBIR program, and typically only Phase I awardees are eligible for Phase II awards. Phase III refers to an agency’s continued support for former SBIR awardees’ commercialization efforts without using SBIR funds, typically through government procurement contracts. For phases I and II, agencies also have some flexibility in instituting award amounts and durations above and below the values shown in Table 1.[17]

Table 1: The SBIR program funds businesses using a three-phase model[18]

Table 2 shows the total number of awards and budgets associated with each of the major SBIR agencies for FY 2018.

Table 2: Five agencies fund the vast majority of all SBIR awards as of FY 2018[19]

DOD and HHS provide the majority (76.4 percent) of funding across the federal government; DOE, NSF, and NASA provide most of the remainder (21.2 percent); and the other six agencies funding the program collectively provide a relatively small amount (2.5 percent).

It is worth noting that for many SBIR agencies, SBIR set-asides are not treated as a single pot of money. Agencies such as DOD and the National Institutes of Health (NIH) at HHS essentially operate a number of autonomous programs across different branches, institutes, and centers. And for certain agencies, such as DOE, congressional language dictates specifically how much SBIR funding should be allotted to specific programs and offices.[20] This restriction can limit an individual program to only funding one to a few awards per year, which could impede its ability to support new and existing projects effectively. Subdivisions of agencies can also be protective of their SBIR budgets. When the set-aside is perceived as being drawn directly from their full research budgets, the program is often seen as a tax.

At NSF, appropriations are allocated as a lump sum, there are no congressional restrictions on how they must be spent, and the program is not seen as a tax on its directorates because they play no role in the program; rather, the whole SBIR program’s management is centralized in one office.

Within the framework provided by Congress and SBA, agencies have significant flexibility in how they administer the program. Their choices reflect their missions, resources, and priorities among the program’s four goals. They may structure the program to meet these goals through the designation of R&D topics, issuing of solicitations, review and selection of grant proposals, and assessment of companies’ performance.

The average number of SBIR awards won per firm is an indicator that offers insight into how agencies manage their programs. For instance, NSF provided a single award to each company it funded in 2018, whereas DOD provided on average over two awards to each company. These numbers reflect NSF’s model of funding start-ups, and DOD’s frequent use of the program for contract research to deliver specific technologies for its use.

Commercializing Innovation Through SBIR

SBIR funding has helped enable awardees to generate 70,000 patents, found nearly 700 publicly traded companies, and garner approximately $41 billion in venture capital investments in its 35-year history.[21] SBIR alumni include success stories such as 23andMe, Symantec, and Qualcomm. An ITIF review of award-winning innovations from 1970 to 2006 highlighted by R&D Magazine found that about one-quarter were developed by companies that had been supported by SBIR.[22] The program has been copied by 17 countries around the world.[23]

Individual agency programs have also been shown to have significant economic impacts. A 2018 study of the National Cancer Institute’s SBIR program related $26 billion in economic output, $9.1 billion in sales, and over 100,000 new jobs to companies that were awarded Phase II grants between 1998 and 2010.[24]

Still, there is room for improvement, especially in improving the commercialization of innovations funded by the program. Some agencies administer the program with a greater emphasis on the technologies funded than the businesses supported—often to address specific R&D needs that have been identified by agency subject-matter experts or to build technologies that the agency itself may need to use.

This strategy is certainly a legitimate use of the SBIR program. But a consequence of prioritizing technologies over businesses is agencies may concentrate a share of SBIR funding among a subset of the applicant pool: companies that have previously received many awards, use contract research as their business model, and have relatively little incentive to grow and commercialize their work.

Prior research suggests that favoring repeat awardees detracts from the program’s objective of promoting the commercialization of federally funded innovation and economic growth. A study measuring outcomes across SBIR programs found that companies that continually sought SBIR contracts were less likely to commercialize the outputs of their research.[25] Another study, which examined SBIR awards from DOE’s Office of Energy Efficiency and Renewable Energy, found that firms tend to produce 20 percent fewer patents with each additional previous SBIR award, and firms with previous SBIR awards have half the probability of acquiring subsequent venture capital investment compared with firms with no prior awards.[26]

NSF’s SBIR program concentrates on seeding new companies that can develop transformative technologies, and distributing its funding widely among companies with high potential for growth. However, the program has not always been managed this way.


NSF is one of the five largest agency funders of the SBIR program. As shown in table 2, in FY 2018, with a budget of $197 million, NSF funded 400 SBIR awards that amounted to 6 percent of SBIR funding across the federal government. The program is housed within NSF’s Division of Industrial Innovation and Partnerships within the Directorate of Engineering.

NSF has reinvented its program over the past two decades in ways that distinguish it from those of many other agencies, and bring to life the government-wide slogan, “America’s Seed Fund.” Perhaps most important, the agency has chosen to strengthen the program’s role in promoting the commercialization of SBIR-funded innovations and company growth in line with the statutory objectives for SBIR. A 2015 study of the NSF SBIR program by the National Academies of the Sciences, Engineering, and Medicine (NASEM) finds that the program does meet all of the statutory objectives for SBIR, with the exception of sufficiently encouraging participation by women and minorities.[27] (This failure is not limited to NSF, which performs about as well on this objective as the other major SBIR agencies.)[28]

To this end, the program has aimed most of its funding at start-ups, as opposed to well-established small businesses. NSF has also shifted away from a decentralized management structure in which program directors across the agency dedicate a small part of their time to the SBIR program; dedicated SBIR program directors are based in a single office and support the program full-time, engaging with companies before and after they are awarded grants. NSF has also instituted additional programs within and outside SBIR to further support the ability of awardees to commercialize the outputs of their R&D.[29]

A Focus on Commercializing Innovation

According to Benaiah Schrag, who serves as NSF’s senior program director for SBIR and has been with the program since 2009, these changes to the program happened relatively organically. They were driven by a changing vision among program administrators of what NSF’s SBIR program should focus on—driving commercialization and economic growth—and a recognition of the types of awardees and program structures that would enable NSF to meet its objectives.[30] Three elements of NSF’s program demonstrate its goal to maximize commercial output: its focus on start-ups, the topical flexibility of its solicitation, and the emphasis on commercialization in its Phase I and Phase II solicitations.

NSF focuses on funding start-up companies, as opposed to more well-established small businesses. Schrag said that NSF’s emphasis on start-ups was not the “result of a top-down mandate,” but rather of “working with a lot of different types of companies and seeing over time that the start-up companies tend to outperform other companies in terms of the commercial outcomes.”[31] This approach is in contrast to some other SBIR agencies, such as NASA and DOE, that tend to give greater consideration to the capacity of firms to fulfill a specific R&D need, with less regard to size or the number of prior SBIR awards.[32]

To target start-ups, NSF values applications from companies that have not previously received SBIR awards, and limits the number of awards given to the same company. One program director interviewed as part of the 2015 NASEM study on NSF’s SBIR program stated he intentionally avoids offering multiple Phase II awards to the same company.[33] NSF has institutionalized this approach by limiting the number of applications a company can submit: currently one application per each of the two annual calls for proposals.[34] This strategy has increased the number of new applicants to the SBIR program, reduced the number of applications it receives, and dissuaded companies from using NSF as a “second shot on goal” for proposals that had been rejected from other agencies.[35] The award data in table 2 shows that each of the 400 companies awarded by NSF in FY 2018 received precisely one award.

While NSF is not the only federal agency committed to supporting basic research, its mission is unique in that it seeks to advance all fields of science and engineering (with exceptions in the medical sciences) and does not aim to use the outputs of the research it funds. Consequently, NSF’s SBIR program uses its wide latitude to promote the most promising businesses and ideas across all fields, wherever there are opportunities for commercial and societal impact.[36] This flexibility is reflected in NSF’s SBIR solicitation. The agency requests projects that fall within a wide range of scientific topics, and does not automatically exclude proposals that fall outside of these areas (for which there is an all-encompassing “Other Topics” category).[37] The full set of topic areas, shown in table 3, is reflective of the areas of research the entire agency supports. In the words of one program director, it is “unlikely that a good project would not find a topic.”

Agencies such as NASA, DOE, and DOD only support technologies that align with their narrower missions. In many cases, they are also looking to serve as downstream buyers of the technologies they fund. They therefore typically request that SBIR applicants address narrower, predefined R&D needs. As an example, an FY 2016 DOE solicitation issued a call for a “single bounce monolithic axis symmetric x-ray mirror optics with parabolic surface profile.”[38] At NSF, a similar project might be submitted to the broader call for advanced optical components and systems under the Photonics topic area.

Finally, NSF has integrated its focus on commercialization into its solicitation. One driver of this effort was Errol Arkilic, who served as a program director from 2003 to 2011 and was among the team that was charged with improving the SBIR program to increase commercial output. Arkilic described working with a team of investors and industry partners to reform the commercialization work plan, an element of the Phase II application. The new commercialization work plan better reflects what private investors look for, such as market potential, subsequent finance plan, and the company’s track record of commercialization. A similar line of questioning evaluating companies’ commercial potential was added to the Phase I application, where previously there had been no discussion of commercialization.[39]

These changes, especially the limitation on applications per company and inclusion of a commercialization history, initially caused “an enormous amount of friction” among some SBIR applicants, according to Arkilic. He described a community of small companies, referred to by some as “SBIR mills,” that had been using the SBIR program for 20 years as a means of conducting contract research, with relatively few attempts to commercialize their work. These companies would often use NSF as a backup application for SBIR proposals submitted and rejected elsewhere, given NSF’s all-encompassing solicitation. Eventually, he said, the new measures instituted by the agency caused these companies to stop applying for NSF’s SBIR support.[40]

Table 3: The Technology Topic Areas offered by NSF’s SBIR program cover a wide range of disciplines and applications. NSF will also consider applications outside of these areas[41]