In August 2026, employment in technology occupations across all industry sectors grew by 86,000 workers, while tech companies cut about 14,700 positions. Both figures came from the same analysis of the same month. If your AI hiring plan still assumes you are bidding against other software companies, it is calibrated to a market that no longer exists.
Two numbers published on September 4, 2026 tell opposite stories about the same four weeks.
CompTIA's analysis of the August jobs report found that employment in tech occupations, which includes tech professionals working across all industry sectors, increased by 86,000 workers. The same analysis found that tech companies reduced staffing by about 14,700 positions, inclusive of both technical and non-technical roles.
Technology work grew. Technology companies shrank. That is not a contradiction, and it is not a rounding error. It describes where technology jobs now live, and it changes who shows up on the other side of your offer.
Worth carrying the caveat: CompTIA prints itself: monthly occupation-level data from the Bureau of Labor Statistics runs hotter on variance than most series and is subject to backward revision. One month is a signal, not a settled trend. It points the same direction as everything else in the 2026 data.
What Do the August Hiring Numbers Actually Say?
Tech employment is growing outside the tech industry. In August 2026, tech occupation employment across all industry sectors rose by 86,000 workers while tech companies cut roughly 14,700 positions. Active job postings requiring AI-related capabilities passed 320,000, up 4.5% month over month, inside a pool of nearly 600,000 active technology occupation postings, 42% of them newly advertised.
That 320,000 counts active postings, meaning new listings plus openings carried over from earlier months. It measures the size of the queue you are competing in rather than the flow of fresh demand, which is the more useful number when you are opening a requisition. The queue grew.
Dice's August 2026 Tech Job Report, built on Lightcast data covering more than seven million United States tech job postings, puts a sharper edge on the AI slice: postings for AI and machine learning roles grew 101% year over year, August 2026 against August 2025, while overall tech postings rose 18%. AI demand is not keeping pace with the market. It is running at roughly five times the market.
Why Is It Harder to Hire AI Engineers If Tech Companies Are Still Cutting?
Because the roles being eliminated and the roles you are trying to fill are mostly not the same roles, and because the segment you are hiring into was already the most senior-skewed part of the American labor market before AI arrived.
Indeed's Hiring Lab found that in the first quarter of 2026, senior-level positions accounted for 69.3% of job postings in software development, the highest concentration of any sector it tracks. Entry-level postings in software development sat at 4.5%, the bottom of a range that reaches 91.3% in personal care and home health. Nationwide, by contrast, only about 14% of postings are for senior positions, while almost half (46%) are entry-level.
Software development is not participating in the same labor market as the rest of the economy. It is running an inverted version of it. Nearly seven in ten openings want someone who has already done the job, in a field where the mechanism that produces experienced people has been narrowing for four years. As of May 2026, senior-level postings were up 14.7% year over year, while entry-level postings, which have been trending down since their 2022 peak, fell 7.5%. We covered the downstream cost in the hidden costs of cutting junior developers: the pipeline problem and the senior-scarcity problem are two sides of the same problem.
The Bureau of Labor Statistics projects the pressure will continue. Between 2024 and 2034, it expects data scientists to grow 33.5%, adding 82,500 jobs; information security analysts to grow 28.5%, adding 52,100 jobs; and software developers to grow 15.8%, adding 267,700 jobs, against 3.1% growth across all occupations. BLS states that adoption of AI technologies, including generative AI tools, is expected to fuel strong job growth among computer and mathematical occupations in the coming years. These are ten-year projections rather than observed change and deserve the skepticism any projection deserves. The direction matters, and it aligns with the monthly data.
Who Are You Competing With Now?
Employers outside the technology industry. That is what CompTIA's figure counts, and it is where the growth landed. In our own hiring pipeline, it shows up as manufacturers, health systems, banks, insurers, retailers, and logistics operators, all now drawing from the same pool you are.
That changes three practical things for a talent function.
Your compensation benchmarks are pointed at the wrong peer set. If you are benchmarking AI engineering comp against software companies, you are benchmarking against the segment that is cutting. The employers actually clearing your offers may be a regional health system with no equity component and a much higher tolerance for cash.
Your pitch has to survive a different comparison. An engineer weighing your offer against a bank's is not comparing your mission to that of another startup. They are comparing scope, ownership, and whether the AI work is real or a labeled pilot.
Your time-to-fill assumptions are stale. A senior AI engineer in a market where 69.3% of postings want senior people has options that did not exist when your last benchmark was set.
What Are Your Four Options, and What Does Each Really Involve?
Hire senior AI engineers directly in the United States. The right answer for roles that must be in the room: the person who owns the architecture, the person accountable to your board. The wrong answer for volume, because you are bidding into the most contested segment of the market against employers from every industry.
Upskill the engineers you already have. Structurally, the best option and the one most often assumed to be free. The Conference Board found that more than half of workers (55.1%) use generative AI or AI agents daily or weekly, but only 33.3% have used organization-provided AI training in the past six months, and 28.3% say their organization does not provide AI training at all.
Only 48% agree their organization provides sufficient time during work hours for AI skills development, and 47.6% agree they have sufficient tools, access, and resources to build AI capabilities. The survey covered nearly 1,300 workers globally and included interviews with 35 enterprise leaders, but it does not disclose its fieldwork window, so treat the precision loosely. The pattern is not loose. Most upskilling plans are an aspiration with no time, no budget, and no tooling behind them. If you intend to choose this option, fund it like a program. That is what our AI fluency program is designed to do, and our guide to preparing your workforce for AI covers the adoption gap in greater detail.
Contract offshore for capacity. Works when the specification is stable, and the work is genuinely separable. Struggles with AI work, which is iterative by nature and depends on tight feedback loops with the people who understand the data. A twelve-hour time difference turns a two-hour clarification into a two-day one, and AI projects generate a lot of clarifications.
Build a dedicated nearshore team. The argument here is not primarily cost, and it stopped being primarily cost some time ago. It is that AI work needs overlapping hours, shared standups, and engineers who accumulate context about your systems rather than closing tickets against a spec. We laid out the engagement structures in AI staffing models, and the capability-not-cost argument in why US companies hire LATAM engineers for AI.
What Does Nearshore Actually Solve, and What Does It Not?
An honest disclosure first, because it affects how you should read this section. Almost every published statistic about Latin American engineering supply comes from a company that sells Latin American engineering, and that includes us. We looked for independent 2026 data on LATAM talent supply, salary differentials, and time-zone overlap outcomes from development banks, government statistics agencies, or surveys with disclosed methodologies. It does not exist in any form we would be willing to cite. Anyone quoting precise LATAM market figures at you is quoting their own book of business.
Near, a LATAM staffing firm, announced an analysis of more than 2,000 placements across 411 roles in January 2026, reporting that 98% of its engineering hires were mid-level or senior, that 84% of all its LATAM hires were mid-level or senior, and that many roles were filled in 7 to 28 days. Those are real numbers about Near's own placements, not measurements of the LATAM market, and the distinction is the whole thing.
What is verifiable is the arithmetic on the United States side, and it is enough to make the decision. You are hiring into a market where 69.3% of software development postings want senior people, where AI role postings doubled year over year, and where the employers competing with you now come from every industry. Against that, a model offering overlapping working hours, engineers who stay on your systems long enough to build real context, and a hiring motion that does not bid directly into the most contested segment of the US market is worth evaluating on its structure rather than its rate card.
What it does not solve is worth stating just as plainly. Nearshore does not fix unclear specifications, and an ambiguous AI project fails in Costa Rica exactly as it does in Austin. It does not remove your obligation to run your own technical interview loop, because seniority claims deserve the same scrutiny wherever they come from. And it does not eliminate the need for a senior internal person who owns the architecture. What it does is let that person lead a team rather than be the team. Our dedicated AI teams model is built around exactly that split, and the forward-deployed engineer post covers the role that usually sits at the seam.
Common Questions About Hiring for AI in 2026
Is tech hiring recovering in 2026?
Employment in tech occupations is growing, but not within tech companies. CompTIA's analysis of August 2026 data found tech employment across all industry sectors up by 86,000, while tech companies cut about 14,700 positions. The demand moved to employers outside the technology industry rather than disappearing.
Why is it hard to hire AI engineers if tech layoffs are still happening?
Because the two groups barely overlap and AI roles are concentrated in the most contested part of the market. Senior-level positions made up 69.3% of software development postings in the first quarter of 2026, compared with about 14% economy-wide. Dice reported that AI and machine learning postings grew 101% year over year in August 2026, while overall tech postings grew 18%.
Should we upskill our current engineers instead of hiring?
Often yes, but fund it properly. The Conference Board found that 55.1% of workers use generative AI or AI agents daily or weekly, while only 33.3% have received organization-provided training in the past six months, and 28.3% reported none was available. Upskilling fails when it is assumed rather than resourced with time, tooling, and a curriculum.
How does nearshore hiring compare with offshore for AI work?
The difference that matters is working-hour overlap rather than rate. AI projects are iterative and require constant clarification, so a team in adjacent time zones resolves in hours what a team twelve hours away takes days to resolve. Be skeptical of precise LATAM market statistics from any source, including staffing vendors, since almost all of it is drawn from the vendor's own placements.
If you are planning 2027 headcount against a market that changed shape this year, the useful conversation is not about rate cards. Talk to the Golabs team about which AI roles genuinely need to sit inside your building, which need to sit in your time zone, and how to structure the rest.