The technology jobs everybody thinks disappeared were never here
Years of headlines about tech layoffs have created a widely held belief that computer and technology jobs are vanishing from the U.S. economy. The occupation-level data says otherwise — and the reason for the gap between perception and reality turns out to be about where those jobs are actually classified, not whether they exist. Data: BLS Occupational Employment and Wage Statistics (OEWS), SOC 15-0000 “Computer and Mathematical Occupations,” 2004–2024; DOL/OFLC H-1B LCA disclosure data by SOC code, 2010–2024.
This is every worker BLS counts in computer and mathematical occupations nationwide — software developers, systems analysts, IT support, data scientists, network engineers, and the rest of SOC 15 — regardless of what industry employs them. The only down years on record are 2009–2010, during the financial crisis. Every other year, including 2022–2024, the years most associated in the press with mass tech layoffs, employment grew.
The same years show sustained, large H-1B activity specifically in this occupation. This is LCA filing volume — it includes renewals, extensions, amendments, and employer transfers, not just new hires, so it should not be read as “this many new workers.” What it does show is the program’s scale and persistence in exactly the occupation this report is about, running alongside the employment trend above rather than in place of it.
This is the two charts above overlaid year by year: the actual net change in national SOC 15-0000 employment that year (blue, can go negative — it does in 2010), against H-1B/E-3/H-1B1 LCAs filed for SOC 15 that same year (orange). H-1B volume is larger than net job growth in every single year shown, sometimes by a wide margin — 2024 is the starkest, 567,000 applications against just 15,000 net jobs added, roughly 37 to 1. That gap is expected and shouldn’t be read as “H-1B replaced 37 native jobs for every 1 created” — most LCA filings are for workers already in the country (renewals, extensions, employer transfers), not brand-new positions, so the two bars aren’t counting the same thing. What the comparison does show plainly: the program’s activity in this occupation has been consistently large relative to the occupation’s actual net growth for a decade and a half, including in the one year — 2024 — where that growth nearly stalled.
This is the part that explains the gap between perception and reality. When people picture “tech jobs,” they picture NAICS 51 — Information: software publishers, internet companies, telecom, media — the sector most mass-layoff headlines come from. But that sector accounts for only 14.0% of the nation’s computer & math occupation workforce. More than twice as many — 35.4% — sit in NAICS 54, Professional, Scientific & Technical Services: IT consulting, computer systems design services, engineering services, staffing firms. Those companies rarely make headlines because they’re vendors and contractors serving every other industry, not household names. The rest are scattered across finance, manufacturing, government, education, and virtually every other sector on the list — because by 2024, writing code and running systems is not a “tech industry” job, it’s a function every industry does in-house.
Two things are both true at once, and they don’t contradict each other. Real, visible layoffs happened at real, recognizable companies — mostly ones classified under NAICS 51, Information, which is exactly the slice of the economy that generates headlines when it cuts staff. And at the same time, national employment in computer & mathematical occupations grew in 19 of the last 21 years, including the exact years the layoffs made news, because that sector is only 14% of where these jobs live. The other 86% — consulting firms, finance, manufacturers, hospitals, universities, government agencies, retailers — mostly kept hiring, they just don’t put out press releases when they do. The jobs people think disappeared didn’t disappear. Most of them were never concentrated in the industry people were watching in the first place.