NASA Cut Its Civil-Service Workforce 22%. Now It Wants Core Technical Work Back In-House.

NASA entered 2026 with a workforce problem that was no longer abstract. The agency had lost more than 4,000 civil servants in a year, and 25 of its 36 major projects told the Government Accountability Office that the reductions had affected them. The Space Launch System program lost almost one-fifth of its civilian workforce. Orion lost roughly one-tenth. The DAVINCI mission to Venus lost key program-management and engineering leaders and reduced some risk-reduction work.

Then NASA changed direction.

In a February directive, the agency said its dependence on contractors for core functions had eroded internal capability, increased program risk and reduced flexibility. NASA ordered its centers to identify technical work that should be performed in-house, restore expertise in engineering, operations and science, and expand the civil-service workforce. A later workforce message said the agency employs more than 40,000 contractors and sees potential civil-service conversions “in the thousands.”

The shift is larger than a recruiting campaign. It is an attempt to redraw the boundary between government and contractor work immediately after NASA removed roughly one-fifth of its own workforce. That boundary matters because a contractor can supply exceptional expertise without carrying the government's ultimate responsibility for technical judgment, mission acceptance, program direction or continuity across contracts.

Federal workforce records show both the scale of the loss and an early response. OPM-covered NASA employment fell from 18,078 in January 2025 to 13,914 in January 2026, a decline of 4,164, or 23.0%. It reached 14,248 in June after 695 accessions that month. That hiring is real. It is not yet proof that NASA has converted thousands of contractor positions, filled its most consequential skills gaps or rebuilt the institutional memory that departed with experienced employees.

A reversal after a long retreat

NASA's civil-service workforce was already much smaller than it had been at the end of the space-shuttle buildup. An agency personnel report counted 24,566 civil servants in 1990. OPM counted 19,207 in September 1998 and 18,150 in September 2024. The January 2026 contraction took the covered workforce below 14,000 before the June rebound.

The two series in the chart are deliberately separated. NASA's historical report and OPM's later employment files do not form one perfectly continuous measurement system. They do establish the long-run scale: the agency approached 25,000 civil servants in 1990, operated near 18,000 for much of the past decade, and entered 2026 with fewer than 14,000 in the OPM monthly file.

GAO's July 2026 review describes what changed inside programs as the latest reduction unfolded. Officials reported lost expertise, heavier workloads, less oversight of contractor work and difficulty maintaining schedules. SLS officials said fewer civil servants reduced NASA's ability to support development and oversee contractors. Orion officials cited losses in systems engineering, safety and mission assurance. DAVINCI officials said departures included critical leaders whose knowledge was difficult to replace quickly.

NASA's fiscal 2027 budget request presents the new approach as budget-neutral: change the contractor-to-civil-servant mix rather than simply add cost. The agency's own planning language goes further. A published core-competencies document says NASA intends to convert contractor roles to civil-service positions “in the thousands” where in-house capability is required.

That is a significant institutional reversal. It does not mean every function should move inside government. It means NASA is acknowledging that the mix itself can create risk when too much mission knowledge, technical authority or operational continuity sits outside the civil-service workforce.

Every center entered 2026 with fewer people

The contraction was agency-wide. Every NASA center represented in OPM's agency-subelement data had fewer covered employees in January 2026 than in January 2025.

Goddard had the largest numerical and percentage decline in the OPM comparison: 1,027 employees, or 33.3%. Headquarters fell by 565, or 26.4%, and Langley by 468, or 26.0%. GAO, using agency records and a different grouping date, separately reported a 34% workforce reduction in Maryland and a 17% reduction in Alabama. Those figures should not be treated as exact substitutes, but they reinforce the center-level pattern.

By June, the recovery was uneven. Kennedy had risen from 1,673 to 1,889 and Headquarters from 1,576 to 1,652. Johnson, Marshall, Armstrong and Stennis also added people. Goddard fell another 49 from its January level, while Ames, Glenn and Langley were roughly flat or lower. A single agency-wide total therefore conceals substantially different local rebuilding paths.

Center headcount also does not reveal which missions gained or lost capability. Goddard supports Earth science, astrophysics, heliophysics and communications work. Johnson's roster spans human-spaceflight operations and engineering. Kennedy's June increase may reflect its own hiring and conversion needs, but OPM data do not identify the project, work package or previous employer attached to each action.

The technical bench narrowed with the total

NASA did not lose only administrative capacity. Its two largest engineering series accounted for 1,501 fewer employees between January 2025 and January 2026. Aerospace Engineering fell from 4,400 to 3,564. General Engineering fell from 3,510 to 2,845.

Electronics Engineering declined by 183, Information Technology Management by 182, Computer Engineering by 177 and Contracting by 162. General Physical Science lost 106 employees. Engineering Technicians fell from 257 to 168, a 34.6% decline, before recovering to 196 in June.

These occupational series line up with several gaps NASA identified. GAO reported needs in aerospace, mechanical, electrical and computer engineering, information technology and cybersecurity. But the apparent match has limits. OPM series are broad classifications. They do not show whether an aerospace engineer supports Orion, aeronautics research or another mission, and they do not identify the contractor role that NASA might convert.

The June occupation totals offer a mixed first reading. General Engineering rose by 30 from January. IT Management gained 48, Materials Engineering gained five, and Engineering Technicians gained 28. Aerospace Engineering, Electronics Engineering and Computer Engineering were still slightly lower than in January. Rebuilding is not moving at one speed across disciplines.

There is also a depth problem that headcount alone cannot measure. In June, 29.0% of NASA's covered workforce was age 55 or older, while 7.5% was under 30. Among selected technical series, the 55-and-older share was 34.0% in Computer Engineering, 40.6% in General Physical Science, 40.5% in Computer Science and 47.4% among Engineering Technicians. Age is not retirement eligibility, and it should not be converted into a departure forecast. It does show why hiring without mentorship and knowledge transfer may fail to replace capability.

Tenure points in the same direction. OPM records show 3,375 NASA employees with at least 25 years of service in June and another 4,747 with between 15 and 24.9 years. A new civil servant can fill a position; rebuilding judgment accumulated across missions takes longer.

Contract dollars are not contractor headcount

NASA's contractor relationship is extensive and heterogeneous. USAspending records $15.8 billion in NASA contract obligations in fiscal 2025, compared with $12.5 billion in fiscal 2015, in nominal dollars. GAO separately said NASA spent more than $18 billion on contractor services in fiscal 2024.

Those numbers use different definitions. The USAspending series covers the contract award group under NASA as the awarding top-tier agency. GAO's figure describes contractor services. Neither is a contractor payroll, and neither can be divided by an assumed salary to estimate people. Contract obligations also buy hardware, research, mission operations, software, integration and specialized services.

NASA Administrator Jared Isaacman drew a narrower boundary in a May workforce message. He said the conversion focus was on people working in long-term, exclusive NASA roles tied to launch, operations and mission-critical engineering. He excluded specialized contractors serving multiple customers and described facilities maintenance, IT, cybersecurity, medical support and non-core research as examples of work that generally would remain contracted.

That distinction is essential. Contractor dependence is not synonymous with contractor failure. A specialized supplier can provide technology and expertise that would be costly or impractical to reproduce internally. Competition can produce innovation. Flexible capacity can help a program manage peaks in demand. The risk arises when NASA no longer retains enough informed civil servants to set requirements, challenge technical assumptions, accept work, manage safety and preserve knowledge across contract boundaries.

The Jet Propulsion Laboratory illustrates why no single contractor ratio can govern the agency. JPL is NASA's only federally funded research and development center and is managed by the California Institute of Technology. Its roughly 5,500 employees and on-site subcontractors are not NASA civil servants. JPL is a deeply integrated, specialized institution, not simply a set of staff-augmentation positions that can be relabeled.

NASA therefore has to classify work, not merely count badges. The public test should be whether civil servants retain technical authority, enough hands-on competence to exercise it and continuity across programs, while contractors and partners continue to supply specialized capabilities where that model is stronger.

June produced the first measurable hiring wave

The clearest sign of a changed workforce trajectory appears in OPM's personnel-action data. NASA recorded 695 accessions in June 2026 and 67 separations. The underlying category extract includes 544 competitive-service new hires, 147 excepted-service new hires and four transfers.

That wave followed extraordinary separation activity. OPM recorded 1,178 separations in December 2025 and 1,934 in January 2026. The categories include several types of departures, and they should not be reduced to a single program without additional records. The monthly headcount series and action files are also different systems: accessions minus separations do not have to equal the change between two employment snapshots.

Still, June was not a statistical rounding error. Covered headcount rose by 286 from May to June. Kennedy recorded 250 June accessions, Marshall 136, Stennis 123 and Johnson 107. Headquarters had 23; Langley 19; Glenn 13; Armstrong 12; Ames 10; and Goddard two.

The sequence is consistent with a post-directive hiring response, but causation and completion remain open. Public OPM data do not say how many June hires were former contractors, which positions were converted, whether the same work remained under contract or which competencies were restored. NASA has said early conversion pilots generated more than $100 million in annual savings, but the reviewed public materials do not provide a center-by-center ledger that would let outsiders reproduce that calculation.

NASA's NASA Force program provides one mechanism. It uses term appointments for mission-critical roles, with the possibility of conversion to permanent civil service under stated conditions. Term appointments can move faster and align staffing with mission duration. They also create a different retention question than permanent appointments if knowledge is supposed to remain inside the agency for decades.

The existing early-career pipeline matters for the same reason. NASA says its Pathways internships can lead directly to civil-service careers. The FederalHiringData archive found no 2025 or partial-2026 NASA announcements with Pathways, Student Trainee, Recent Graduate or Intern in the title, but that narrow title match is not a complete inventory of early-career recruiting. Programs can use different titles, shared announcements or hiring channels. The defensible finding is that the public USAJOBS title signal is currently weak, not that NASA has no early-career pipeline.

The public recruiting record shows the size of the reset

FederalHiringData's historical USAJOBS archive contains 15,718 distinct NASA announcements opened from March 13, 2017 through August 14, 2026. Of those, 8,955 included at least one selected technical series used in this analysis.

Related research

Put this finding in context

All articles

Data investigations/August 21, 2026

DOD Cut 78,375 Civilian Jobs. The Work Did Not Disappear.

The 2025 reduction reached IT, contracting, engineering, logistics and weapons testing. Public records do not show a clean transfer of the work or its full cost.

Data investigations/August 20, 2026

OPM's 35% Cut Was Its Deepest Since It Transferred an Entire Bureau

The federal government's HR agency now has fewer employees than at any reviewed point since 1998 while directing agencies to rebuild hiring and workforce planning.

NASA opened 2,236 distinct announcements in 2018, 2,252 in 2020, 1,993 in 2022 and 1,673 in 2024. The count fell to 396 in 2025. The partial 2026 archive contains 238 through August 14; 179 opened from the February 6 directive through that date, including 103 in the selected technical series.

Announcements are not vacancies, offers, selections or hires. One announcement may support multiple appointments, and an announcement may close without a selection. The annual bars also have unequal partial periods in 2017 and 2026, which are labeled directly. The archive is best read as a measure of visible recruiting activity, not a replacement for OPM employment and action data.

The series nevertheless captures the sharpness of the reset. The June accessions demonstrate that NASA can add hundreds of employees even when public announcement volume is far below earlier years, likely because recruitment and appointment pathways do not map one-for-one to announcements. That makes an auditable conversion ledger more important, not less. Without it, observers cannot connect the directive, public recruiting, personnel actions and contractor changes at the level of a center or skill.

Rebuilding requires a definition of “core”

The February directive asks centers to assess expertise that has been outsourced or lost, identify work that should return in-house and build implementation plans. That is the right sequence. It is also where the hardest choices begin.

A useful public definition of core work would address at least four questions. Does the role exercise inherently governmental authority? Does NASA need enough internal expertise to evaluate a contractor's work independently? Would a contract transition create unacceptable mission or safety risk? Does the capability need to persist across programs even when near-term demand falls?

The answers will vary. Flight operations, systems engineering, mission assurance and technical authority may require a durable internal bench. Specialized manufacturing, multi-customer research and commercial services may remain better outside. Some functions need both: contractors doing substantial work and civil servants capable of directing, testing and challenging it.

GAO's project findings show why numerical replacement is insufficient. Losing an experienced engineering leader and adding an entry-level engineer are not equivalent events, even if headcount returns to its prior level. Converting a long-serving contractor may preserve mission knowledge but still require conflict, pay, appointment and supervisory decisions. Hiring from outside may widen the talent pool but increase onboarding time.

The next evidence should therefore be more specific than an agency total. NASA can report which competencies each center designated as core, target and completed conversions by occupation, time-to-hire, retention, contractor positions or work packages changed, and outcome measures such as technical-authority staffing, oversight capacity and schedule risk. It can protect sensitive project information while still publishing an aggregate implementation ledger.

June's hiring wave is a credible first checkpoint. It is not the finish line. NASA had 14,248 covered civil servants in June, still 3,830 below January 2025. The agency's challenge is to show that the people it adds restore the ability to understand, direct and accept mission-critical work, rather than only moving costs between contract and payroll accounts.

Readers can browse current federal jobs, explore federal workforce statistics, compare agency records and read more FederalHiringData investigations.

Methodology and limitations

Workforce history: NASA's 1981-1990 civil-service totals come from an official agency personnel report preserved by the NASA Technical Reports Server. FederalHiringData used OPM legacy FedScope September snapshots from 1998 through 2013 and monthly Federal Workforce Data for later years. The source break is shown rather than interpolated. January 2025, January 2026 and June 2026 are monthly OPM observations. OPM-covered employment excludes contractors and does not identify project assignments.

Centers, occupations, age and tenure: Center totals use OPM agency-subelement codes. Occupational series are broad classifications and may include employees working across missions. Age 55 or older is not retirement eligibility. Length of service is not a measure of performance or unique expertise. Percent changes use unrounded headcounts.

Personnel actions: Accessions and separations come from OPM monthly action files. Action categories and employment snapshots have different timing and processing rules, so their arithmetic does not reconcile one-for-one. A June accession is not necessarily a contractor conversion or a response to the February directive.

Contract spending: Contract obligations come from the USAspending API's contract award group for NASA as the awarding top-tier agency. Values are nominal and are not adjusted for inflation. GAO's fiscal 2024 contractor-services figure uses a different definition and is presented separately. Neither series measures contractor headcount.

Recruiting archive: FederalHiringData counted distinct USAJOBS control numbers with a NASA agency label or NASA agency code. Coverage begins March 13, 2017 and runs through August 14, 2026. Selected technical announcements include specified engineering, science, IT and program-related occupational series. Announcements are not vacancies, applications, selections or hires. The early-career check searched titles for Pathways, Student Trainee, Recent Graduate or Intern and should not be treated as a complete program inventory.

Primary documents: Program effects and the description of NASA's February directive come from GAO-26-108556. NASA's conversion strategy comes from its fiscal 2027 budget, core-competencies plan, May workforce message, NASA Force page and careers materials. Those documents describe intentions and early pilots; they do not provide a complete public ledger of conversions by center, occupation or prior contractor.

The hero photograph is an official NASA image published with the NASA Force workforce initiative. It shows personnel monitoring science operations in a control room and illustrates the kind of mission work discussed here; it does not identify the employment status of every person shown.