August 2026 tech jobs report: who's hiring engineers

The August 2026 tech jobs report says tech occupations added 86,000 workers while tech companies cut about 14,700. Hiring moved to manufacturers, consultancies and retailers. Prepin exists for exactly that search.

A few fits you'd never browse for by name.

Nearly 600,000 tech occupation postings were active in August, and 42% of them were newly advertised. Plenty of roles. The catch is that the ones worth a senior engineer's time increasingly sit at employers whose name you've never typed into a search box.

What did the August 2026 tech jobs report actually say?

Two numbers pulled in opposite directions. Tech occupation employment rose by 86,000 in August while tech companies reduced staffing by about 14,700 positions, according to CompTIA's analysis of BLS data and Lightcast postings. Both are true at once, and together they're the whole story.

Tech occupation employment is a labor measure that counts tech workers across every industry sector, tech companies included. The 14,700 figure covers technical and non-technical roles at tech companies specifically. So engineers got hired in August. Just not, on net, by the firms most engineers apply to.

Beyond tech, the economy stayed steady. Total nonfarm payrolls rose by 162,000 and the unemployment rate held at 4.1%. The information sector, a BLS industry group that includes computing infrastructure, data processing, web hosting, publishing and broadcasting, lost 23,000 jobs against a prior 12-month average of 8,000 per month.

Dice's August report, built from over 7 million US tech postings pulled on September 3, showed postings down 2% month over month and up 18% year over year. Two data sets, one direction. Fewer new postings than July, far more than a year ago, and a thinner slice of them at tech companies.

Figures above are current as of September 9, 2026.

Why are tech companies cutting while tech jobs keep growing?

Because most tech jobs aren't at tech companies, and August's net growth came from everywhere else. CompTIA's 14,700 figure counts everyone a tech company let go, engineers and otherwise. The 86,000 figure counts engineers, support and infrastructure staff wherever they work, and in August that meant professional services, manufacturing and administrative support.

CompTIA's framing was that "employers remain focused on strengthening core technology functions while expanding emerging capabilities." Growth showed up in tech support, infrastructure, project management and cybersecurity alongside AI. Think of a platform team at a manufacturer that finally got budget.

The information sector losses came from three places.

• 8,000 at computing infrastructure providers, data processing, web hosting and related services
• 7,000 in publishing industries
• 5,000 in broadcasting and content providers

Those are the employers a backend engineer defaults to when the search starts. They're shrinking.

Where did the senior backend, platform and ML roles go?

Into companies whose main product isn't software. CompTIA found the most new tech postings in professional, scientific and technical services, manufacturing, and administrative support services, with month-over-month gains in retail, wholesale trade and educational services. Dice's industry cut points the same way.

• Retail postings up 18% month over month
• Consulting up 7%
• Finance and banking up 6%
• Software up 5%
• Manufacturing up 4%

Geography shifted too, with Illinois postings up 11% and California up 9%. If your saved searches are a list of 30 software companies you admire, none of this reaches you. It also explains a lot about why 200 applications produced no replies. Many of them went to the shrinking column.

Does a 2% dip matter if AI and ML postings doubled?

Not much if you're an ML or platform engineer. Dice recorded AI and machine learning postings up 101% year over year, and CompTIA counted more than 320,000 active postings requiring AI-related capabilities, a 4.5% rise from July. A 2% monthly dip across all of tech is rounding error next to that.

The definition of an AI skill is getting narrower. Dice flagged year-over-year growth above 200% in Responsible AI, AI Agents, Agentic AI, Artificial Intelligence Infrastructure and Enterprise Integration. CompTIA's version is employers who want people to "apply AI tools, manage AI-enabled workflows, and support AI-driven initiatives." Read that as a request for the workflow you actually shipped.

A resume line reading "familiar with LLMs" doesn't clear that bar. Skill checks and assessment evidence beside your profile do more work than a claim. Prepin keeps them next to the profile for that reason, instead of relying on resume lines alone.

Why does volume applying get worse in a market like this?

Because the roles that fit are scattered across industries you don't follow, and spraying applications at the ones you recognize just hits the wrong targets faster. Nearly 600,000 active postings with 42% new is a lot of noise. The signal is the small share at a consultancy or manufacturer that needs your exact platform background and posts a range above your floor.

Auto-apply tools optimize for count. In a month where retail tech postings jumped 18% and tech-company headcount fell, count is the wrong thing to optimize. The math in mass applying vs targeting 10 roles a week gets worse when the good roles hide in unfamiliar industries. And the application process itself keeps getting harder in 2026.

How does constraint-based matching find roles you'd never browse for?

It starts from what you've already decided instead of from a company name. Constraint-based matching is a search method that filters open roles by fixed inputs like location mode and a compensation floor, then shows only what clears them. Nobody has to know a manufacturer's name to get matched to its platform team.

Prepin does this through a candidate agent. A candidate agent is an AI job search tool that builds your profile, surfaces the few roles that fit and contacts an employer only after you approve the introduction. There's a fuller walkthrough in what an AI job search agent does in 2026. Here's the actual workflow, no forms involved.

• One voice conversation about goals, strengths and dealbreakers, combined with your LinkedIn and resume, becomes a single structured profile a recruiter can actually read.
• Search setup takes a location mode (remote, hybrid, onsite or specific cities) and a compensation minimum. You set both once.
• Matched roles arrive in an inbox showing company, location and posted compensation range, so a consultancy you've never heard of shows up on the same terms as a software firm.
• A profile weakness review runs before any recruiter sees it.
• Nothing reaches an employer until you press approve. There's no auto-apply switch to find, because there isn't one.

Prepin's site positions the best use as finding 3 to 5 strong fits, which is the right shape for a month like August. Fit is the number that matters. That's the whole argument in a better alternative to auto-apply bots, and it gets stronger when the fits are at employers you'd never have searched for. If you're employed and searching quietly, per-intro approval also means your profile doesn't drift to your current employer's competitor by accident.

What should you do with this report this week?

Widen the industries you'll consider and tighten everything else. The report says roles moved to manufacturers, consultancies, retailers and professional services. Your constraints, a compensation floor and a location mode, are what keep that wider net from turning into noise.

• Write down the floor and the location mode before you look at a single posting.
• Drop the company shortlist. Let industry be an output of the search.
• Get assessment evidence for the AI skills employers name, since 320,000-plus postings ask for them.
• Run a system design and behavioral practice loop before the first real one, while the pipeline is still small enough to prepare for.
• Approve intros one at a time and decline the rest.

If that shape sounds right, prepin.ai runs the conversation, the matching and the approvals.