Summary

Computing¹ is vital to U.S. innovation, the economy, and national security. China, the United States’ largest competitor and rival, is poised to greatly outpace the United States in the production of science, technology, engineering, and mathematics (STEM) PhDs overall—especially in computing and related fields. To remain globally competitive, the United States needs to sustain a robust computing workforce—and in particular a sufficient supply of computing doctorates to continue powering industrial and academic advances and to train the next generation of computing innovators. The need is especially acute in artificial intelligence (AI), where there has been a surge in innovation and applications, in cybersecurity where the capability and capacity of the workforce falls short of society’s needs, and in national security roles where U.S. citizenship is required. Today’s computing workforce relies on more than half of the newly minted computing PhDs being international students.

Although there have been a series of boom-and-bust cycles in the computing sector, demand for computing expertise has continued to rise over the long term. Moreover, computing’s pervasive integration into the economy suggests a sustained and likely growing demand for talent in the future.

Computing doctorates are highly valued in the computing sector, a demand reflected in significant salary growth over the past decade. They take on engineering, research, and development roles that leverage their deep technical expertise, problem-solving skills, and research experience. These skills are also in high demand in other sectors that increasingly rely on computing such as agriculture, health care, and financial

___________________

¹ This report uses “computing” to describe the fields of science and engineering related to computing, including computer science, information science, computational science, and computer engineering.

services. Indeed, salaries and other incentives attract a significant fraction of doctoral degree recipients to industry, especially in high demand subfields such as AI.

Faculty vacancies, particularly at teaching-focused institutions and primarily undergraduate institutions, hamper the education of future computing professionals, which will make it harder for employers to fill critical computing and computing-related roles. Undergraduates majoring in computing fields more than doubled from 2012 to 2022, well above the 12 percent rate of increase for bachelor’s degrees overall. Demand for undergraduate computing degrees has historically followed a cyclical pattern—rising and falling with trends in the computing industry—although it has grown overall in the long term. Early signs suggest another downturn may be beginning.

Moreover, as computing has been integrated throughout all sectors and disciplines, demand by non-majors for computing courses, including advanced undergraduate and graduate courses, has grown dramatically. Demand for faculty has also come from growth in doctoral and professional master’s programs in computing.

The National Academies of Sciences, Engineering, and Medicine undertook this study to assess trends in supply and demand, pathways and flows toward advanced degrees in computing and computing careers, the balance between doctoral degrees awarded to U.S. and international students, and areas of potential shortfall and their implications for the health of computer and information science and engineering disciplines and academic programs. The study, sponsored by the National Science Foundation’s (NSF’s) Directorate for Computer and Information Science and Engineering (CISE), considers the ways that colleges and universities, scientific and professional societies, and industry partners could help address anticipated shortfalls in doctoral degree production, as well as the additional data that might be required to better assess trends and impacts for the future.

SUPPLY AND DEMAND FOR PHDS IN COMPUTING

Quantitative evidence and expert testimony provide differing perspectives on the supply of and demand for computing doctorates. NSF’s Survey of Earned Doctorates and the Computing Research Association’s Taulbee Survey provide large sample evidence on employment outcomes in doctoral-granting institutions and first career post-degree for doctoral students in computing by sector. Steady aggregate growth over the past two decades in the number of computing doctorates has largely kept pace with demand in industry and research-intensive universities. Put another way, if sustained investments had not been made to grow this supply, the result would have been mounting unmet demand.

The committee also heard broadly consistent testimony pointing to challenges in hiring in the following key areas:

• Faculty positions at non-PhD granting institutions;
• National security roles, national laboratories, federally funded research and development centers, university affiliated research centers, defense contractors, and in the federal government—a gap attributed to insufficient numbers of doctoral degrees being earned by domestic students; and
• Across sectors in areas of high national interest, such as AI, cybersecurity, and high-performance computing.

There is also broad concern about a sufficient number of competitive candidates to meet these demands. It is, however, difficult to quantify these reported challenges across different types of academic institutions, in government, and in industry. For example, industry hiring data are limited and not broken out by subfield, and no one currently collects data on faculty hiring by non-PhD-granting institutions. Available data on private-sector job postings also do not specifically track positions that require a doctorate in computing. Moreover, characterization of workforce demand is inherently difficult because workforce demand cannot be directly measured. For example, organizations may post more positions than they intend to fill or, if the positions cannot be filled by a candidate with the desired skill set, organizations may find alternate candidates and close the gap through on-the-job training. Projecting future demand is even harder because it depends on various assumptions and macroeconomic factors. Thus, although one can measure supply (how many doctoral degrees are awarded in computing each year), it is difficult to estimate demand (current need for PhDs in computing).

Considering these factors and viewpoints the committee concluded that

• In areas of especially high demand, such as AI and cybersecurity, nearly all institutions face difficulty in recruiting doctoral recipients;
• Teaching-intensive universities have difficulty filling both tenure- and non-tenure-track computing faculty positions; and
• In other areas, the growth rate in the number of computing doctorates has largely kept pace with demand for doctoral recipients at leading research-intensive universities and in industrial research and development.

Another key dimension of supply and demand is the balance of international and domestic doctoral students. International students now account for a clear majority of doctoral degrees in computing, having increased from 48.5 percent in 2012 to 60.9

percent in 2022, mostly from China or India. The number of domestic doctoral degrees awarded increased 15.8 percent from 2012 to 2022 (785 to 909), but the overall share of domestic students in doctoral programs decreased. Indeed, despite increasing domestic undergraduate enrollment in computing, the number of domestic students pursuing PhDs has not seen a corresponding rise.

There are two principal reasons that the United States cannot rely primarily on attracting more international students to meet its future needs for doctoral-level computing talent. First, it creates the risk of a future workforce shortfall because both the supply and stay rates of international students are sensitive to changes in U.S. immigration policy,² opportunities in student’s home countries, and the global competition for talent.

Second, it precludes expansion of the talent pool needed to fill national security positions that require U.S. citizenship. Computing doctorates are a small but critical component of the national security workforce, encompassing roles in government, national laboratories, and the defense industrial base. These roles generally require U.S. citizenship and often security clearances, requirements that considerably reduce the pool of potential candidates. Moreover, it has proven hard to fill these positions despite various efforts to shrink the gap between salaries for these positions and those in industry, owing to both compensation and the nature of the work.

This picture points to the need for targeted interventions in cases where there is evidence of unmet demand along with actions to ensure a continued robust supply of doctorates in computing.

Recommendation 1: Federal agencies supporting research and education, universities that perform research and education, and the larger computing industry should at a minimum sustain and, if possible, expand doctoral fellowships and assistantships in computing independently and in partnership. At the same time, they should increase the fraction of fellowships and assistantships awarded to domestic doctoral students studying computing.

Recommendation 1 aims to achieve two coupled goals:

1. Ensuring a sufficient supply of talent to sustain the world-class research and teaching workforce in the United States and ensuring a sufficient supply of talent in areas that are important for U.S. economic competitiveness and national security.

___________________

² The committee notes that various changes to U.S. immigration policy could change the number of international applicants to U.S. doctoral programs in computing or increase the likelihood that international students stay in the United States following receipt of their doctoral degree, but that U.S. immigration policy is outside the scope of this study.

2. Reversing the decline in the fraction of domestic students pursuing a doctoral degree in computing, ensuring supply that is more robust to changes in global labor markets and helping meet demand for national security roles that require U.S. citizenship.³

There ought to be no problem finding additional highly qualified international and domestic candidates to meet these goals, given the enormous growth in the number of U.S. undergraduates earning computing degrees and the very low acceptance rates in the most selective doctoral programs. Additional steps will be needed to attract more domestic students to graduate study, including enhanced opportunities for undergraduate students to become aware of, be exposed to, and experience computing research, as well as paying attention to the opportunity costs of attending graduate school. It may also require changes in recruiting and admission practices and additional attention to doctoral student retention.

BETTER MEASUREMENT OF SUPPLY AND DEMAND

More detailed information would help NSF and others supporting doctoral education in computing to detect specific supply shortfalls and tailor appropriate interventions to mitigate them.

Recommendation 2: The National Science Foundation should, in collaboration with other federal statistical agencies and the Computing Research Association, collect additional data to better measure supply and demand for doctoral degrees in computing and provide the information needed to tailor future interventions.

Recommendation 2-1: The National Science Foundation should collaborate with other federal statistical agencies and the Computing Research Association to identify and collect additional data on the supply of doctoral degrees in computing.

Data are not currently being collected in aggregate to reflect whether there is a retention problem within computing doctoral programs, either compared to other

___________________

³ Another way to grow the number of PhDs with U.S. citizenship to fill national security roles would be to expedite paths toward citizenship. Changes to national immigration policy are, however, out of scope for this report.

fields or to past retention trends, and the committee heard no testimony to this effect. Additionally, more information is needed to assess the in-flow and out-flow of doctoral applicants and recipients, especially for domestic students. Data to collect would include (1) the number of applicants (preferably distinct—which would require a trusted third party to cross-link by name or unique identifier) to doctoral programs in computing each year; (2) the number of students admitted to doctoral programs in computing each year; (3) the number of years to complete doctoral studies; and (4) the completion and attrition rates for doctoral programs. The CRA Taulbee Survey (Zweben and Bizot 2024) has recently added questions that address the first two data points suggested, but longitudinal data are needed to assess the trends in applications and admissions and allow for appropriately tailored interventions in the future.

Recommendation 2-2: The National Science Foundation should collaborate with other federal statistical agencies and the Computing Research Association to identify and collect additional data that would allow them to better measure and project academic hiring demand for doctoral degrees in computing at non-doctoral-granting institutions.

Although some data are available on the success of faculty searches and hiring at doctoral-granting institutions, these alone do not show a complete picture of the challenges in meeting demand for faculty, especially those teaching undergraduate courses. Presentations to the committee suggested that it has not been possible to hire the faculty required to meet the enormous demand for teaching computing at various institution types. Data to collect in an expansion of the CRA Taulbee Survey would include faculty hiring data for non-doctoral-granting institutions; NSF could encourage or require institutions that receive research funding from NSF to participate.

Recommendation 2-3: The National Science Foundation Survey of Earned Doctorates should add a set of questions to track the supply of and demand for PhDs in computing and related fields, disaggregated by subfield.

Differentiating doctoral recipients by research area, interest, capability, and citizenship will help ensure that demand is met while avoiding oversupply. Doctoral students are not necessarily interchangeable across subdisciplines or topic areas, students interested in research careers may not be seeking faculty positions at teaching-focused institutions, and not all students may be interested or eligible to pursue national security careers.

STRENGTHENING RESEARCH ENGAGEMENT AND GRADUATE ADMISSIONS

Multiple factors shape undergraduate students’ decisions to pursue a computing doctorate and their success in being admitted to a doctoral program. Beyond financial pressures (see Recommendation 5), key factors include a lack of awareness of and exposure to research, unclear expectations for successful applications, and an opaque admissions process.

Raising Awareness of Research and Expanding Undergraduate Research Opportunities

Many U.S. undergraduate students—particularly those at primarily undergraduate institutions, regional universities, and community colleges—lack awareness of research and associated career pathways enabled by doctoral study. This gap has multiple causes—research is generally not well integrated into undergraduate curricula, and research is often not emphasized in student advising. Even in research-intensive institutions, many students may be unaware of the research being performed by faculty and graduate students. Yet another knowledge gap is that some students are not even aware that most doctoral students in computing are financially supported by research assistantships or fellowships. Various initiatives have attempted to close these gaps, but much more can be done to raise awareness among domestic undergraduate students.

Participants in undergraduate research are twice as likely to apply to computing PhD programs and more likely to be accepted than students with no formal research experience. It appears that only a small fraction of undergraduates has access to such opportunities, a gap exacerbated by growth in the number of computing majors, but solid data do not exist (see Recommendation 3-5 below).

National programs offering paid undergraduate research experience enable students to engage in computing research, generally at an institution other than their home institution. Examples include NSF’s Research Experiences for Undergraduates (REU) Sites and Supplements, Department of Energy (DOE) Science Undergraduate Laboratory Internships, the Computing Research Association’s Committee on Widening Participation (CRA-WP) Distributed Research Experiences for Undergraduates (DREU), as well as other summer research programs established at academic institutions, national laboratories, federal agencies, and industry. Many of these programs target students from 4-year colleges—primarily undergraduate institutions and minority serving institutions—providing the opportunity to work in a research environment that would otherwise be unavailable. However, the number of research opportunities has not kept pace with growing undergraduate computing enrollment. This points to an opportunity—increasing the number of opportunities, which would help boost domestic enrollment in doctoral programs.

Many research experiences provide little to no compensation for relocation or housing expenses during the experience, which can preclude some students from participating. Also, stipends for summer undergraduate research programs are lower than even entry-level service sector pay and are significantly lower than for industry internships (most of which are not research focused).

Recommendation 3: Research funders, computing departments, professional societies and organizations, and industry should take actions to increase the number of undergraduate students aware of and exposed to research, participating in research experiences, and pursuing research careers in computing.

Recommendation 3-1: Computing departments at colleges and universities and computing professional societies and organizations should enhance and coordinate efforts to provide more and better information to undergraduate students about the nature of computing research, undergraduate research opportunities, skills needed for success in research, and the availability of financial support for graduate study in computing.

Recommendation 3-2: Research institutions (doctoral-granting computing departments and computing companies) and teaching-focused institutions (primarily undergraduate institutions, regional universities, and community colleges) should establish partnerships to enhance undergraduates’ awareness of computing research and doctoral study opportunities.

Such partnerships can provide undergraduate students with more information about computing research opportunities, exposure to computing research conducted at partner institutions, coaching on the admission process, and financial support to facilitate pathways into computing doctoral programs. Successful partnerships could help both computing departments and companies identify and recruit domestic talent.

Recommendation 3-3: Computing undergraduate research programs and computing departments should increase the number of research opportunities and provide sufficient funding and other support to expand student and faculty involvement in undergraduate research experiences.

To achieve these outcomes in undergraduate research experiences, funding will need to be significantly expanded across institution types and not limited to research at doctoral-granting institutions, with creative solutions to reach a wider range of students or more students at lower cost, including but not limited to online, self-paced, or group-based research.

Recommendation 3-4: Industry partners should not only expand research-intensive internships and offer other opportunities for industry research but also consider expanding support for undergraduate research programs at academic institutions.

Recommendation 3-5: Programs funding undergraduate research experiences in computing should require participating institutions to report on the number of students applying and acceptance rates to gauge demand.

Adopting a Holistic Approach to PhD Admissions

Several aspects of the admissions process create challenges for prospective students. Expectations for program acceptance are often unclear; information about the overall competitiveness of programs can be hard to come by; and the process is opaque, especially for students who do not have a faculty advisor or mentor to guide them. Gauging the chance of success is even harder for departments where individual faculty members select students for admission.

Admissions processes vary widely among doctoral programs in computing. Programs that look for prior research experience disadvantage students without such experience, a circumstance most common for students from primarily undergraduate institutions and especially from schools that do not send many graduates to PhD programs.

Several additional factors make it less likely that domestic students will matriculate in doctoral programs. Top programs receive a large number of high-quality domestic applications, but not all of even the top domestic applicants will be accepted, given the limited number of slots available and competition from highly qualified international applicants. Domestic students not successful in gaining admission to a top program are more likely to take an industry position than attend a doctoral program in computing at a lower-tier university and are less likely to reapply in following years. Consequently, the majority of applicants at most other institutions are international.

Recommendation 4: Universities and computing departments should adopt a more holistic approach to engaging, encouraging, and evaluating applicants to doctoral computing programs, especially for domestic students with limited research experiences, holders of degrees in non-computing-related disciplines, and those who have not identified a preferred research advisor.

Recommendation 4-1: Doctoral programs in computing should strive to make the admission process transparent. This includes providing (1) basic data on the number of applications and admission rate, (2) information on what strong applicants look like, and (3) clear explanation of program logistics, including but not limited to funding opportunities and deferral possibilities.

Recommendation 4-2: Admission decisions should be the result of a hybrid evaluation approach that includes individual faculty selecting students matching their research interests as well as programs reviewing applications from promising individuals who have not identified a research area or who have non-traditional career pathways.

Recommendation 4-3: Doctoral programs in computing should inform all applicants of the admissions decision. Where feasible, non-admitted but promising applicants, especially domestic students, should be given constructive feedback on their application and be encouraged to reapply.

Recommendation 4-4: Computing departments should consider developing transitional programs that allow additional time to complete course prerequisites for otherwise promising students who do not meet traditional admissions criteria, especially domestic students. These programs, which might take the form of a transitional master’s program, should offer faculty advising and financial resources, such as teaching assistantships, to help support students in making the transition.

An increasing number of universities have created pathways into computing master’s programs for students who did not study computing as undergraduates, which also creates an opportunity for students to take advanced courses or gain research experience to prepare them for admission to a doctoral program.

INCREASING GRADUATE STUDENT SUPPORT

In computing, unlike other STEM fields, a PhD is not required for career advancement, as lucrative job opportunities and career paths are available to undergraduates. Although most computing PhD programs offer financial support through assistantships and fellowships, stipends are not competitive with industry starting salaries. Moreover, stipends at many institutions are insufficient to provide a reasonable standard of living, even for single students without dependents. This gap is likely a challenge in many STEM fields; this report addresses only implications and solutions for doctoral programs in computing.

Recommendation 5: All institutions supporting fellowships and assistantships to doctoral students in computing should explore available options to increase the amount of financial support provided to each student in order to allow a reasonable standard of living, make graduate study in computing more attractive relative to employment, and make graduate study more attractive to domestic students.

Recommendation 5-1: Academic institutions and funding agencies should increase the minimum stipend for graduate students in computing and index it to the inflation rate.

Recommendation 5-2: Funding agencies should consider establishing fellowships to provide supplemental funding for benefits such as housing allowances, health insurance for dependents, parental leave, and childcare.

Recommendation 5-3: Academic institutions and funding agencies should pursue the development of partnerships with computing industry leaders to provide supplements to department stipends or augment fellowships.

Most doctoral students in computing rely on federally supported research assistantships. It is thus impossible to both sustain or grow the number of graduate students (Recommendation 1) and provide more resources per student (Recommendation 5) without either growth in federal research investment or new partnerships that expand or augment research assistantships.

Industry partnerships are likely an essential ingredient for implementing Recommendation 5. The combination of federal and institutional resources will probably be insufficient to increase stipends to the needed level and variation across an institution may be otherwise limited because university rules, practices, and labor agreements may dictate uniform stipends. Federal research budgets have not seen significant growth in recent years, and future growth is far from certain. A partial solution in such an environment would be for industry (both the computing sector as well as other sectors that rely on computing talent) to step up and provide more support to help meet their immediate (i.e., new PhDs) and long-term (i.e., faculty to teach future employees) needs.

IMPROVING GRADUATE STUDENT RETENTION

Another way to maximize the number of doctorates earned in computing is to increase completion rates. Earning a PhD in computing is a challenging endeavor that typically takes a median of 7 years from the start of graduate school. The process is essentially a multi-year apprenticeship with a faculty advisor. Thus, effective and supportive mentoring and advising plays a significant role in a student’s success. A collaborative and supportive research group and departmental environment can also contribute to student retention. Such efforts may be especially effective for retaining domestic students, who have a wider range of options available should they leave before program completion.

Recommendation 6: Computing departments, in concert with professional organizations, should improve mentoring and advising for graduate students, especially for domestic students, to improve student retention and success.

Recommendation 6-1: Professional organizations and the Computing Research Association should create and disseminate materials on mentoring best practices for adoption within computing departments. Computing departments should implement and adopt these mentoring practices department wide.

Recommendation 6-2: Computing departments should provide training for all faculty members on how to mentor students effectively, including providing structured guidance for all new junior faculty members during their first 3 years. This should include emphasizing

the roles of junior faculty members as research advisors, teachers, and mentors, providing concrete advice on how to be successful in those roles.

Recommendation 6-3: For students who enter a doctoral program without a matched research advisor, computing departments should initially assign a department advisor to provide support and guidance, and to assist students in finding an advisor.

Recommendation 6-4: Computing departments should periodically assess the state of graduate student support, mentoring, department climate, student experience, and the impact of interventions on graduate student retention.

TARGETED INTERVENTIONS TO ADDRESS HIRING DEMANDS

Students who successfully complete their doctorate in computing have a wide range of occupations open to them, particularly those who study in high-demand subfields. The practice of academics pursuing industry roles while remaining in academia for a percentage of their time is becoming more common and more widespread. Industry’s demand for computing doctorates is primarily driven by a desire to advance innovation and translation in areas where economic value can be derived, resulting in a demand for skills associated with specific computing research subfields and the skill of conducting computing research itself.

Recommendation 7: To encourage their employees to pursue doctoral degrees in computing, industry and government laboratories should provide fellowships for graduate study in exchange for a commitment to return for a period following completion.

Owing to the large portion of undergraduates with degrees in computing that pursue positions in industry following graduation and perceived lack of doctoral-level talent available to pursue positions in industry and government, especially in high-demand areas such as AI and national security, these institutions should invest in their own employees to help meet their internal demand for doctoral recipients. This support could take the form of fellowships to cover graduate student stipends, release from work

to pursue doctoral degrees, or resources to supplement and encourage research toward a doctoral degree.

Recommendation 8: The National Science Foundation and other federal agencies that support computing research as well as industry, potentially through government–industry partnerships, should create new fellowships and assistantships in computing that include requirements to serve in faculty positions for a specified period.

For industry partners, the benefits include sustaining and growing the faculty that will educate future employees, growing the supply of talent in high-demand areas. Such fellowships and assistantships could be targeted in various ways, including for faculty positions at non-doctoral-granting institutions for subfields that are in especially high demand. There are several examples of federal computing fellowships with requirements for service, including the NSF CyberCorps® Scholarships for Service and Artificial Intelligence Scholarships for Service Initiative.

Ensuring adequate production of PhD recipients requires a sufficient academic faculty. Faculty hiring and faculty composition, including in areas that are currently highly sought after, directly influence PhD production.

Recommendation 9: Computing departments and academic institutions should take a broader view of what merit is considered in faculty hiring and promotion, not just the number of publications or PhD pedigree, but other attributes, including, but not limited to, commitment to best practices in mentorship, preparing students for their future careers, and research that builds bridges with other fields.

BETTER UNDERSTANDING THE SHIFTING COMPUTING TALENT LANDSCAPE

New bachelor’s level degrees are creating additional pathways into doctoral computing programs. For example, blended interdisciplinary degrees between computer science and other disciplines (e.g., CS+X) produce students educated in core computer science degree topics in addition to their second discipline. There are also increasing online degree offerings at both the master’s and bachelor’s degree levels in computing and adjacent fields. There are limited data available about the cohorts who attain such degrees and their associated career trajectories.

Recommendation 10: The National Science Foundation and the Computing Research Association should collect data on undergraduate and graduate student educational and career trajectories—inclusive of computer science and engineering (CSE), information science, data science, CS+X interdisciplinary degrees in non-CSE departments, computational science, and computational physical and biological sciences, and similar computing-adjacent fields.

Recommendation 10-1: The National Science Foundation and the Computing Research Association should collect data on post-graduation employment or educational plans for undergraduates in computing and computing-adjacent fields.

Recommendation 10-2: The National Science Foundation and the Computing Research Association should collect more granular data on the educational background, including undergraduate and graduate degrees and field of graduate study for students entering doctoral programs in computing and computing-adjacent fields.

* * *

U.S. leadership in computing has depended critically on decades of investment in educating computing doctorates. Meeting the competitiveness and national security challenges of the coming decade will require not only continued investment but also targeted interventions to enhance engagement of domestic PhD students and to ensure sufficient faculty to teach the next generation of computing leaders.

This page intentionally left blank.