Systems administrators
Official title Network and Computer Systems Administrators (SOC 15-1244)
Also called SysAdmin, Network administrator, IT administrator, Server administrator, Infrastructure administrator
In brief
In May 2025, BLS counted 314,340 US systems administrators, with median pay of $99,130 a year. Of the BLS counts since 2021, the highest was 325,930, in May 2022, close to the May 2025 count. The earliest figure here is an estimate of 50,000 from 1986, not measured the same way as today's count. The last full technology era for systems administrators: AIOps first generation, 2016 to 2023.
What the sources say
Official outlook
BLS projects employment to decline 4.1% from 2025 to 2035.
AI exposure studies
- 51.4% of this job's tasks are exposed to large language models. Not a forecast of job losses. Eloundou and others, 2023
Data as of BLS OEWS May 2025; BLS Employment Projections 2025-35. BLS publishes these counts as yearly survey snapshots and cautions against reading them as a precise trend.
Last updated .
Cite this page
Future History. (2026, October 11). Systems administrators: jobs and pay since 1986 [Data set; data as of BLS OEWS May 2025; BLS Employment Projections 2025-35]. https://futurehistory.earth/timeline/15-1244
Future History. "Systems administrators: jobs and pay since 1986." Data as of BLS OEWS May 2025; BLS Employment Projections 2025-35. Last modified October 11, 2026. https://futurehistory.earth/timeline/15-1244.
@misc{futurehistory_15_1244, author = {{Future History}}, title = {Systems administrators: jobs and pay since 1986}, year = {2026}, month = oct, howpublished = {\url{https://futurehistory.earth/timeline/15-1244}}, note = {Last updated 2026-10-11. Data as of BLS OEWS May 2025; BLS Employment Projections 2025-35}, }
The access date is today's date on your device.
≈ $96,588 in 2024 prices (derived by us, CPI-adjusted)
Ken Thompson and Dennis Ritchie write Unix at Bell Labs on a discarded PDP-7. The operating system spreads through Bell Labs and then through universities on magnetic tape, creating the first communities of people who needed to administer shared multi-user systems. The occupation has no formal name yet; administration is simply a function that falls to whoever understands the machine best.
The first USENIX Large Installation System Administration (LISA) conference is held, providing the first dedicated professional gathering for systems administrators. In the same year, Evi Nemeth and colleagues at the University of Colorado publish the first edition of the Unix System Administration Handbook, noting that only three books on the subject exist. The handbook becomes the de facto professional reference for the occupation for the next two decades.
The System Administrators Guild (SAGE) is formally constituted within USENIX in January 1993, providing the first professional organization explicitly for systems administrators. SAGE grows to thousands of members over the following decade and creates formal career development resources, job boards, and ethical standards for the profession. In 2003, SAGE separates from USENIX and eventually becomes LOPSA (League of Professional System Administrators) in 2005.
The dot-com peak: Y2K remediation efforts (1998-2000) had swelled IT headcounts across every sector, and internet infrastructure investment had reached unsustainable levels. The dot-com collapse of 2000-2002 and the telecom bust that followed eliminated roughly 15-20% of IT positions across the industry. Systems administrators, more deeply embedded in ongoing operations than project-based developers, experienced a less severe collapse than programmers but still lost significant employment in 2001-2003 as companies froze IT budgets and outsourced network operations to managed service providers.
Amazon Web Services launches S3 on March 14, 2006 and EC2 in August 2006, beginning the cloud era. For the first five years, cloud adoption is primarily a startup phenomenon. The architectural implication for sysadmins is structural: every application workload that moves to a cloud provider eliminates the physical server administration work but preserves (and often expands) the network, security, and configuration management work. By 2012, enterprise migrations are accelerating, and the occupation begins its long structural contraction.
Patrick Debois coins "DevOps" at the first DevOpsDays conference in Ghent, Belgium, in October 2009, formalizing a movement that had been building since the mid-2000s. DevOps treats infrastructure as software: configuration management (Puppet, 2005; Chef, 2009; Ansible, 2012), automated deployment pipelines, and infrastructure-as-code become the new administrative practice. The philosophical impact on sysadmins is profound: the craftsperson model of knowing a system intimately through years of hands-on management is partially replaced by a software engineering model of defining systems through code, testing, and deploying them through automated pipelines.
BLS publishes its 2024-2034 employment projections showing -4% decline for network and computer systems administrators (from approximately 331,500 to approximately 318,300). The same projections show the computer systems design sector growing 11% overall, confirming the bifurcation: the high-skill cloud-architecture and DevOps tier is absorbing the industry's growth while traditional on-premises sysadmin roles contract. AIOps tools are not yet fully modeled in the 2024 projections; a second inflection may be visible in the 2026 data as agentic platforms like Datadog Bits AI SRE, Dynatrace Davis AI, and ServiceNow ITOM with Now Assist compound the productivity multiplier.
Will AI replace systems administrators? What the research says
Each figure is stated by the named source; we publish no forecasts of our own. Where sources disagree, we show the disagreement rather than an average.
Each figure is stated by the named source; we publish no forecasts of our own. Where sources disagree, the spread is the honest signal.
What's changing day to day for systems administrators
Three parts of your work where AI is already doing real lifting, and what stays yours.
AI is sitting alongside you hereMonitor infrastructure health using AIOps dashboards (Datadog, Dynatrace), reviewing AI-prioritized alert queues, validating autonomous root-cause diagnoses, escalating confirmed incidents, and closing false positives with documented rationale.
Monitor infrastructure health using AIOps dashboards (Datadog, Dynatrace), reviewing AI-prioritized alert queues, validating autonomous root-cause diagnoses, escalating confirmed incidents, and closing false positives with documented rationale.[3],[4],[5]
Shift from passive dashboard watching to active AIOps governance: learn to tune alert thresholds and suppression rules so the AI surfaces genuine signal, and build fluency in reading agent-generated RCA traces to validate or override machine conclusions confidently.
AI is sitting alongside you hereWrite and maintain automation scripts and Infrastructure-as-Code (PowerShell, Bash, Ansible playbooks, Terraform modules): using AI coding assistants to generate first drafts, then reviewing, testing, and owning the resulting runbooks and IaC pipelines.
Write and maintain automation scripts and Infrastructure-as-Code (PowerShell, Bash, Ansible playbooks, Terraform modules): using AI coding assistants to generate first drafts, then reviewing, testing, and owning the resulting runbooks and IaC pipelines.[7],[15],[8]
Own the automation architecture, not just the scripts: design idempotent, version-controlled runbooks, validate AI-generated Ansible YAML against your environment's specific constraints, and build test harnesses (Molecule, Terratest) so automated code is trustworthy in production.
AI is sitting alongside you hereManage cloud infrastructure provisioning and cost optimization (AWS, Azure, GCP): using AI copilots (Azure Copilot Optimization Agent, Pulumi Neo) to identify rightsizing opportunities, enforce tagging policies, and automate resource lifecycle management.
Manage cloud infrastructure provisioning and cost optimization (AWS, Azure, GCP): using AI copilots (Azure Copilot Optimization Agent, Pulumi Neo) to identify rightsizing opportunities, enforce tagging policies, and automate resource lifecycle management.[9],[8],[2]
Build cloud financial operations (FinOps) and architecture skills: AI agents surface cost anomalies and recommend instance type changes accurately, but committing to reserved instances, negotiating enterprise agreements, and making architecture decisions about multi-cloud vs. single-cloud deployments require human judgment and financial accountability.
Where systems administrators go next
Natural next steps for someone with your foundation: not exits, evolutions.
- Computer and Information Systems Managers
- US median pay
- $175,140 (May 2025)
- BLS outlook 2025-35
- +15.8%
- Computer Network Architects
- US median pay
- $134,050 (May 2025)
- BLS outlook 2025-35
- +7.7%
- Computer Systems Engineers/ArchitectsNo timeline for this role yet
What each move takes
Computer and Information Systems Managers
Senior sysadmins who accumulate production-change authority, vendor relationship management, budget ownership, and the ability to communicate infrastructure risk to non-technical stakeholders are well-positioned for IT management. The 2026 market specifically demands IT managers who can govern AIOps tool adoption (Datadog Bits AI, Dynatrace, ServiceNow ITOM), set automation strategy, and translate infrastructure investments into business outcomes: all skills that emerge naturally from a senior SysAdmin background.
How has the work of systems administrators changed since 1969?
The tools that defined the work, one line per era. Select an era to see how it reshaped the work.
Unix command-line tools and shell scripting (PDP-7, PDP-11, VAX)
The first generation of systems administrators worked almost entirely through terminal emulators and shell scripts. Unix provided a remarkable toolkit: the file system, pipes, grep, sed, awk, cron, and shell scripting let a single administrator manage a multi-user machine with dozens of simultaneous users without a graphical interface. Administration was fundamentally a craft of scripted text manipulation: editing /etc/passwd, managing disk quotas with df and du, monitoring processes with ps, and scheduling jobs with cron. The job was oral: knowledge passed between administrators through personal networks, mailinglists, and the nascent culture of Unix documentation (manual pages) that Ritchie and Thompson built into the system from the start.
PC networks: Novell NetWare, TCP/IP, and the LAN administrator era
The arrival of Novell NetWare 2.0 (1985) and the proliferation of IBM PC-compatible hardware brought a new job class into existence: the LAN administrator. Every organization that acquired more than a handful of PCs needed someone to run the file server, set up user accounts, and manage shared printers. These administrators were often not Unix practitioners at all; many came from accounting or operations backgrounds and learned Novell's proprietary tools (SYSCON, FCONSOLE, NWADMIN). The parallel growth of TCP/IP networking after the National Science Foundation opened the internet backbone to commercial use in 1991 created demand for a second skill set: routing, DNS, DHCP, and email server administration. By 1993, the System Administrators Guild (SAGE) was formally constituted within USENIX as the profession's first dedicated professional organization, reflecting a community that had grown from academic practitioners into a mainstream corporate function.
Windows NT/2000/2003 Active Directory and the enterprise server era
Microsoft's Windows NT 4.0 (1996) and Windows 2000 with Active Directory (2000) transformed corporate IT administration. Active Directory centralized identity and policy management across an organization's entire Windows estate: user accounts, group policies, software deployment, and security settings could all be managed from a single console. This moved the systems administrator from machine-by-machine configuration toward domain-wide administration. It also created a parallel credential ecosystem: Microsoft Certified Systems Engineer (MCSE), a certification that by the late 1990s had become the most widely held IT credential in the US. Windows NT also brought proprietary event-log monitoring tools (Microsoft Operations Manager, later System Center) that began systematizing what had been informal, ad-hoc monitoring. The administrator's toolkit was increasingly graphical, less scripted, and more vendor-specific.
Cloud infrastructure (AWS S3 and EC2, 2006) and the beginning of cloud absorption
Amazon Web Services launched S3 on March 14, 2006 and EC2 in August 2006. For the first several years, cloud adoption was primarily a startup phenomenon: companies like Airbnb and Dropbox built on AWS infrastructure rather than buying servers. The impact on the systems administrator workforce was initially indirect: organizations still ran their own data centers, but the growth rate of on-premises infrastructure investment began to slow. By 2010-2012, large enterprises began serious cloud migrations, and the nature of the sysadmin role began to shift: administering a virtual machine fleet on EC2 was architecturally similar to administering physical servers, but the physical provisioning and hardware maintenance work was gone. Amazon's early pitch to IT departments was exactly this: "you keep the administration, we handle the hardware." The sysadmin workforce peaked around 2012-2013 as this transition began in earnest.
Effect on the workThe BLS employment series for this occupation peaked around 2013 and has been in gradual structural decline since, with cloud absorption of on-premises workloads cited by BLS as one of the primary drivers of the projected employment decline through 2034.
Infrastructure as Code: Puppet (2005), Chef (2009), Ansible (2012), Terraform (2014)
The DevOps movement, formally named at the first DevOpsDays conference in Belgium in 2009, reshaped the systems administrator role more profoundly than any prior technology shift. The core idea was that infrastructure should be defined in code, version-controlled, tested, and deployed through automated pipelines, just like application software. Puppet (released 2005) was the first tool to make configuration management declarative and repeatable at scale. Chef (2009), Ansible (2012), and HashiCorp Terraform (2014) followed with different philosophies but the same goal: eliminate manual, undocumented, drift-prone configuration from the administrator's daily work. For the profession, this was both an opportunity and a displacement. Administrators who learned IaC and joined DevOps teams could multiply their impact dramatically; those who did not found their work progressively automated away. The line between systems administrator and software developer blurred substantially, with "Site Reliability Engineer" (a Google term from around 2003) becoming the aspirational title for the elevated tier.
Effect on the workDevOps adoption did not directly reduce headcount in the short term but did bifurcate the market: IaC-fluent administrators became more valuable while traditional click-through-the-console administrators became more substitutable. TechTarget's 2026 networking job market analysis confirmed this split explicitly: entry-level manual-configuration roles are shrinking while architecture and automation roles are growing.
AIOps first generation: Datadog (2010), PagerDuty (2009), Splunk ML (2016)
The first generation of AIOps tools automated the most repetitive part of the sysadmin's day: monitoring dashboards and alert triage. Datadog (founded 2010, IPO 2019) built a unified observability platform that correlated metrics, logs, and traces and used machine learning to surface anomalies and group related alerts. PagerDuty's ML-based noise reduction cut alert volumes by reducing duplicate pages for the same incident. Splunk's machine learning toolkit (2016) enabled behavioral baselining and anomaly detection that had previously required dedicated data-science resources. For systems administrators, these tools reduced the hours spent staring at Nagios dashboards or manually correlating events across disparate systems. The daily rhythm shifted: less reactive page-and-fix, more proactive capacity planning and architecture review.
Agentic AIOps: Datadog Bits AI SRE (2025), Dynatrace Davis AI, Azure Copilot Agents, Ansible Lightspeed
The second generation of AIOps, arriving at scale in 2024-2025, crossed from "assisted human" to "autonomous agent." Datadog Bits AI SRE (generally available June 2025) investigates alerts end-to-end without human prompting, reaching root-cause conclusions in several minutes and executing triage actions including Slack notifications, Jira ticket creation, and PagerDuty incident management. Dynatrace Davis AI, expanded in February 2025, uses causal AI to both diagnose active incidents and predict issues before they surface; IDC independently validated 56% faster mean time to resolution for critical incidents. ServiceNow ITOM with Now Assist AI agents processes incoming IT alerts end-to-end and cuts MTTR by 45% on average across deployments. Red Hat Ansible Lightspeed generates complete, testable YAML automation playbooks from natural-language descriptions. The human administrator's role is shifting from executing the operations cycle to governing and directing the automated agents that now run it: setting thresholds, validating machine-generated root-cause analyses, authorizing production changes, and making the architecture-level decisions that agents cannot.
Effect on the workBLS projects a decline for this occupation through 2034, driven by cloud absorption and DevOps automation. The AIOps acceleration visible in 2024-2025 is not yet fully reflected in those projections, which are based on 2024 baseline data. The practical impact is a rising productivity bar: an AI-fluent sysadmin with full AIOps tooling can credibly manage infrastructure that required a few additional headcount in 2020.
See the same long-arc view for your own profession.
Browse the directory by industry, or search by title or SOC code.
Browse all rolesNetwork and Computer Systems Administrators employment and pay by year
In other countries
The latest figure each country's statistics agency publishes for this job. Each uses its own occupation definition, survey basis and currency, so the numbers are not directly comparable with each other or with the US figures.
Canada (NOC 22220)
- Employment, 2020
- 66,285, Statistics Canada Census 2021 Official statistic Report an errorStatistics Canada Census of Population 2021: number of persons with employment income in NOC 22220 Computer network and web technicians, Canada, reference year 2020. A recipients basis (anyone with employment income during the year), slightly broader than a point-in-time employed count. Statistics Canada Open Licence.
- Median pay, 2020
- C$73,500, Statistics Canada Census 2021 Official statistic Report an errorStatistics Canada Census of Population 2021 (Table 98-10-0586): median employment income for NOC 22220 Computer network and web technicians, Canada, reference year 2020, full-year full-time workers (the basis comparable to the US OEWS full-time median). Statistics Canada Open Licence.
View all 10 years of cited US data, each linked to its source
| Year | US employment | Annual pay (median unless marked) | Source | Report |
|---|---|---|---|---|
| 1986 | 50,000 | n/a | Est., USENIX LISA conference history Researched context | Report |
| 1993 | 200,000 | $34,000 | Est., Wikipedia (employment) Researched contextEst., Bureau of Labor Statistics (pay) Researched context | Report |
| 2000 | 234,040 | $51,280 | BLS OEWS 2000 Official statistic | Report |
| 2019 | 354,450 | $83,510 | BLS OEWS May 2019 Official statistic | Report |
| 2020 | 339,560 | $84,810 | BLS OEWS May 2020 Official statistic | Report |
| 2021 | 316,760 | $80,600 | BLS OEWS May 2021 Official statistic | Report |
| 2022 | 325,930 | $90,520 | BLS OEWS May 2022 Official statistic | Report |
| 2023 | 323,020 | $95,360 | BLS OEWS May 2023 Official statistic | Report |
| 2024 | 318,570 | $96,800 | BLS OEWS May 2024 Official statistic | Report |
| 2025 | 314,340 | $99,130 | BLS OEWS May 2025 Official statistic | Report |