US tech work expanded outside a shrinking tech-company boundary
CompTIA estimated 86,000 more technology workers across the economy while technology companies cut about 14,700 positions. The apparent contradiction is a measurement lesson, not proof of an AI jobs boom.
What happened
CompTIA released its analysis of August U.S. labour data, combining BLS employment estimates with Lightcast job postings.
Why it matters
Employers and training providers may misread industry layoffs as falling demand for technical work, or job postings as completed hiring.
The August U.S. labour data produced two apparently incompatible headlines. CompTIA's release estimated that technology occupations across the economy increased by 86,000, while companies in the technology sector reduced employment by about 14,700. Both can be true because occupation and industry are different boundaries.
A software developer at a bank, hospital or manufacturer counts as a technology worker outside the technology industry. A salesperson, lawyer or facilities worker at a software company counts inside the technology industry but may not hold a technology occupation. When technical capability moves into user industries, occupational demand can rise even while technology producers restructure.
CompTIA also reported more than 320,000 active U.S. postings requesting AI-related capabilities in August, up 4.5% from July. That is a demand signal, not a hiring total. One vacancy can be posted on several sites, remain open across months or never be filled. The result also depends on how Lightcast identifies AI language and deduplicates advertisements.
Three measures answer three questions
Industry payroll asks where people work. Occupation estimates ask what work they do. Postings ask what employers say they want. None alone shows which skills were used after hiring, whether a new role replaced another task or whether a position delivered value. Workforce planning becomes unreliable when the three measures are blended into one “tech jobs” line.
The wider labour market was stronger in August than the technology-sector decline suggests. The Bureau of Labor Statistics reported a preliminary increase of 162,000 nonfarm payroll jobs and an unchanged unemployment rate of 4.1%. However, initial monthly estimates are revised, seasonal adjustment matters and a gain after weak months does not establish a durable trend. Independent coverage described the market as slow to hire and slow to fire, retaining the revision risk.
The data do not identify AI as the cause of either movement. Technology companies can cut because of investment cycles, consolidation, demand, margins or reorganisation. User industries can add technical roles for cloud migration, cybersecurity, data engineering and conventional software as well as AI. A posting that names an AI capability may seek a specialist, or it may attach a generic requirement to a broader job.
Plan for destinations, not only suppliers
For talent leaders, the useful question is where technical work is migrating. Split demand by employer industry, occupation, seniority, location and contract type. Within postings, separate model development from data engineering, security, product integration, change management and domain-facing implementation. A single AI keyword count cannot tell which pipeline to build.
Training providers should connect curricula to destination industries. A technical worker entering healthcare or finance needs sector regulation, data constraints and operational context alongside tools. Employers hiring from outside their industry need to test whether candidates can translate technical choices into domain consequences, not only whether they recognize product names.
The next check is persistence. Compare three-month moving averages and later BLS revisions before reallocating a programme. Follow postings through to hires and retention where possible. If occupations rise across user industries for several months while supplier payrolls shrink, that would support a diffusion story. One August estimate is only an early signal.
The Skills Atlas can help separate technical and complementary capabilities. The labour data supply a more basic discipline: always label whether a number describes an occupation, an industry or an advertisement.