Is IT still a good career with AI? What the 2026 job data says
BLS projects computer support jobs down 3% and security analyst jobs up 21% by 2035. This guide covers the tasks AI absorbs, which IT roles grow, degree vs certs and what to learn first.
Updated: November 2026
AI now does the password resets and first-draft scripts that used to be the way into IT. Still, the US Bureau of Labor Statistics projects computer support jobs down 3% by 2035 and security analyst jobs up 21% (!), so the field is worth studying if you aim right. Below: what AI absorbs, which roles grow and what to learn first.
TL;DR
- Worth it. IT is still a good career with AI, and BLS projects about 280,000 IT openings a year through 2035.
- What AI absorbs. Password resets, first-draft scripts, log triage and boilerplate docs.
- What grows. Security, cloud and platform work, plus anyone who can check AI output.
- Degree or certs. Support roles open with certs and hands-on work, while admin and security roles usually ask for a degree.
- Learn first. Networking, operating systems, identity and scripting, then one specialty.
What the 2026 job numbers say about IT careers
Computer support specialists: down 3% between 2025 and 2035, says the BLS Occupational Outlook Handbook. That's 24,300 fewer jobs out of 903,100. The reason, per BLS: chatbots and other automated tools now do the troubleshooting on the user-support side.
Sysadmins don't get off easy either. Network and computer systems administrator jobs are projected down 4%, and BLS notes that admins are automating their own routine tasks. So the person who scripts everything eventually scripted themselves. Probably on a Friday. Probably without a change ticket.
That's the doom-thread version: "AI is ending IT jobs." Anyone who just enrolled in an IT course lives on a constant diet of negative feedback like that. Then you read one row further down the same table.
A shrinking job still hires though. BLS expects about 48,700 support openings in each year of the decade, all from people moving on or leaving the workforce. And information security analyst jobs? Up 21% (!). BLS names the growing use of AI as one of the reasons. So AI sits in both columns. It pushes one line down and another one up.
Across all computer and IT occupations, BLS projects about 280,000 openings a year and faster-than-average growth. The median IT wage was $109,470 in May 2025. For all occupations, it was $50,980. More than double, for a career whose obituary is already drafted.
| Role | Jobs in 2025 | Projected change, 2025-35 | Openings per year | Median pay, May 2025 |
|---|---|---|---|---|
| Computer support specialists | 903,100 | -3% | 48,700 | $62,890 |
| Network and computer systems administrators | 323,600 | -4% | 13,400 | $99,130 |
| Information security analysts | 192,900 | +21% | 14,100 | $129,180 |
| All computer and IT occupations | n/a | Faster than average | 280,000 | $109,470 |
That's the decade. Now zoom in to one month. CompTIA's October tech jobs report puts tech unemployment at 2.8%, against 4.1% nationally. It counts 629,876 active tech job postings, up 18.5% on a year earlier, and 366,322 postings tied to AI skills. Ten years and one month tell the same story.
The tasks AI is absorbing and what it hands back
Imagine you're the AI agent and you're picking work from the queue. You have three tickets:
A password reset - Same steps every time. Confirm the person, reset, close. Take.
A first-draft script - Boilerplate in, boilerplate out. Take.
A weird case from an upset user that touches access - Nobody has seen it before, and somebody has to decide who gets in. Pass.
The repeatable work goes to the machine: password resets, first-draft scripts, log triage, boilerplate documentation, ticket sorting. Manual test runs and data entry too. What comes back to a person is the checking, the deciding and the owning.
Look at the humblest ticket of the three. A password reset is a small ceremony. The company checks that a person is who they say they are, and the password is the least interesting thing in the room. The agent runs the ceremony now. The suspicion stays with a person.
MSP numbers show the same split. Kaseya's 2026 State of the MSP Report surveyed more than 1,000 MSPs. 53% already use AI to automate ticketing, patching and monitoring. More than half have automated only about a quarter of their workload. So in those shops, the robot uprising covers roughly a quarter of the work. The other three quarters still go through a technician, who also supervises the uprising.
NetworkChuck asks the same question about networking in "I'm worried about Network Engineers":
Which IT roles are most and least exposed
| Role | What AI absorbs | What stays with a person | BLS outlook, 2025-35 |
|---|---|---|---|
| Scripted tier 1 help desk | Password resets, how-to answers, ticket sorting | Odd cases, upset users, anything touching access | -3% (computer support specialists) |
| Systems administration | Routine patching, script drafts, log review | Change decisions, architecture, recovery | -4% |
| Security analysis and incident response | Alert triage and enrichment | Containment calls and accountability | +21% |
| Cloud, platform and automation engineering | Boilerplate config and code | Design, guardrails and review | No separate BLS line |
One caveat, and it's the kind that ruins a clean chart. "Infrastructure" as a job title is splitting in two. The classic sysadmin seat shrinks on paper, and employers still struggle to fill it. The same job is disappearing and hard to hire for, at the same time.
Robert Half's 2026 tech hiring outlook lists systems administrators and network and cloud engineers among the roles shaping hiring plans. 65% of the tech leaders it surveyed say skilled people are harder to find than a year ago. The work moves toward cloud, platform and security. And the people who did the old job well get hired for the new one.
If the tier labels are new to you, this guide to IT support tiers explains what each one does.
The entry-level floor is rising
The strongest evidence comes from payroll data. Stanford's Digital Economy Lab tracked millions of US workers through ADP records up to June 2026. Employment of 22 to 25 year olds in AI-exposed occupations sits 19% below where it would be if it had kept pace with less-exposed peers. The gap comes from fewer hires. Layoffs didn't go up. Experienced workers show no comparable gap.
The authors add two caveats. First, the study covers AI-exposed occupations in general, with IT support included but not singled out. Second, they call the findings early indicators, with no proof of cause yet.
MSPs show the same shape. ConnectWise's Service Leadership found that 43.4% of IT solution providers cut headcount in 2025, up from 36.4% the year before, as Channel Dive reported. Peter Kujawa, who runs Service Leadership, expects the MSP staffing pyramid to end up closer to a diamond: a narrower level-one base and more level-two techs. In essence, a diamond is a pyramid that hired fewer juniors and still expects level-two techs to come from somewhere.
So there are fewer seats at the bottom, and more is expected from whoever sits in them. The door isn't shut though. Those 48,700 support openings a year are still there. In Kaseya's survey, the share of MSPs struggling to hire skilled technicians went from 9% to 16% in a year. And Kujawa still calls MSPs an "unbelievably great training ground for techs."
Expect less time on the scripted layer than someone who started five years ago. They got years of password resets as a warm-up. You get the abridged edition.
Accountability does not automate
Production breaks, or a client tenant gets compromised. A named person owns it. An agent can draft the fix. It can't sign for it.
Picture 02:14 on a weeknight. The agent has a fix and full confidence. It has no pager, no manager and no name on the change ticket. The on-call admin has all three.
So checking AI output is turning into a core skill. In Action1's 2026 survey of sysadmins, reported by Help Net Security, 53% said they wouldn't let AI deploy patches across production while they were away. 14% said they would. Admins use AI to suggest causes and commands, then verify what it found. They treat the model's answer like a config a vendor's sales engineer sent over: thanks, looks nice, checking it anyway. Nobody in IT was born yesterday.
Security teams report the same need. ISC2's 2025 Cybersecurity Workforce Study surveyed 16,029 people. 59% reported critical or significant skills needs on their teams, up from 44% a year earlier. AI was the top skill need at 41%, ahead of cloud security at 36%. So the shortage is in people who understand what the tools are doing.
Day to day, this shows up as approval gates. Someone decides which actions an agent may take alone and which need sign-off. That someone has to know the system well enough to judge. Our guide to human in the loop AI for IT teams walks through that decision action by action.
Degree vs certs vs hands-on experience
You'll probably want more than one. The three stack on top of each other.
BLS spells out the entry bar by role. User support jobs typically ask for some college, and a high school diploma plus IT certifications can qualify. Network support typically asks for an associate's degree. Systems administrator and security analyst roles typically ask for a bachelor's.
| Path | What it gets you | Where it falls short | Who it suits |
|---|---|---|---|
| Degree | Depth in the fundamentals, and a pass through HR filters for admin and security roles | Slow, and it won't teach you a ticket queue | People aiming at security, architecture or large employers |
| Targeted certs (Security+, CCNA, Azure or AWS) | Proof of a specific skill a hiring manager can check | A cert without practice shows in the first interview | Career changers and anyone adding a specialty |
| Hands-on work (home lab, help desk seat, MSP job) | Judgment, the part AI doesn't supply | Hard to show on paper without a cert or degree beside it | Everyone, as early as possible |
Each path has a blind spot. Take the degree. Four years of computer science, a capstone project and a graduation speech about shaping the future. Then the first ticket: the printer is "doing the thing again." Nothing in the syllabus covered the thing.
The ISC2 study has a useful number here. Among security professionals aged 21 to 29, 38% got in through an IT role and 23% through a cybersecurity degree. So general IT is still the main on-ramp to the fastest-growing corner of the field.
In MSP hiring, a targeted cert plus a year of real tickets often beats a degree with no hands-on work. For the service desk track, our ITSM certifications guide compares ITIL, HDI and KCS.
What to learn, in order
Fundamentals first. You can't check AI output if you don't know what right looks like. Then pick one specialty where AI creates work.
- Networking. Learn DNS, DHCP, routing and VLANs well enough to find where a fault sits. The running joke in IT is that it's always DNS. The skill is proving it. This network troubleshooting method is a good drill.
- Operating systems. Windows and Linux internals: services, processes, permissions and logs.
- Identity. Who can sign in to what, and why. Entra ID vs Active Directory covers the two systems you'll meet first.
- Scripting. Enough PowerShell or Bash to read a script line by line before it runs. These PowerShell commands sorted by ticket are a practical start.
- One specialty. Security, cloud or automation.
- Reviewing AI output. Read every generated script before it touches a machine. Make it your default setting.
Each step is what lets you judge the output of the next one.
If you're heading into MSP work
An MSP is still the fastest way to see a lot of environments in a short time. The level-one seat there is changing shape. Fewer pure ticket-takers. More techs who supervise automation and move to level two sooner. Basically, the level-one seat without the level-one tickets.
That's the model we build for at Flamingo. (Yes, this is the paragraph where the vendor blog mentions the vendor. It's short.) In OpenFrame, AI agents close routine tickets and flag issues nobody reported. The technician approves, checks and owns the result. Same team, more done.
For the owner's side of the same shift, see what AI changes in the MSP business.
So, is studying IT worth it?
Yes, if you plan to move past the scripted layer. The numbers back the field: 280,000 openings a year and pay at more than double the all-jobs median. They back the direction too, with security up 21%. Planning to stay on scripted tier 1 work for years? Think twice. That's the layer BLS expects to shrink.
If automation is the specialty you're leaning toward, how to build an AI agent for IT shows what the work looks like.
If it's security, start with an incident response plan a small team can run.
Learn the fundamentals well enough to tell the machine it's wrong. Ideally not on a Friday, and with a change ticket.

Content Marketing Lead
Ohayo! I run content, SEO, social, and community at Flamingo. Before IT, I worked as a correspondent for Ukraine's Public Broadcasting Company and have a Master's in journalism.
Related Content
Frequently Asked Questions
/blog/is-it-a-good-career-with-ai
Still deciding?
Ask Mingo answers from our docs and customer stories.