A client in Dubai forwarded me a layoff headline on Thursday with two words attached: “good news?”. He had four open engineering roles, two of them six weeks old, and he had just read that the technology industry was shedding people at a record pace. It is a reasonable inference. It is also, in our experience of the last three of these waves, close to backwards.
What the numbers actually say
Reporting published on 11 September 2026 under the headline “From Oracle to Amazon: Tech giants drive global wave of layoffs in 2026” put the running total from the tracking platform Layoffs.fyi at 128,536 technology employees across 299 companies as of 10 September. The comparison that gives it weight is not the absolute figure but the baseline: the same tracker recorded 122,606 for the entirety of 2025. The year had passed its predecessor with nearly four months left on the clock.
The month itself was sharp rather than steady. More than 6,000 technology jobs were eliminated in the first ten days of September alone, with Uber announcing roughly 3,300 cuts — about a tenth of its workforce — and reductions also linked to PayPal, Apple, Zomato and Oracle in the same window. On a cumulative basis Oracle leads the year at around 21,000 roles, having stated in its own filing that AI adoption and deployment across its operations had resulted in workforce reductions. Amazon, Dell, Meta, Microsoft and PayPal follow.
Our expert view — read the methodology before the number
Pull two trackers for the same period and you get 128,536 and “over 180,000”. The gap is not a rounding error; it is larger than the entire 2025 total for several individual firms. It exists because trackers make different choices about whether to count non-engineering functions, contractor non-renewals, facility closures, and multi-year programmes announced once but executed in stages. We have watched hiring plans in the UAE get built on whichever figure appeared first in a newsletter. If a number is going to justify a freeze, the person quoting it should be able to say which methodology produced it. Most cannot.
The shape of the cuts matters more than the size
An aggregate number tells you almost nothing about your own requisition. The distribution does, and the 2026 distribution is unusually legible.
Cuts this year have concentrated in mid-career generalist roles at large employers — not the graduate cohort, and emphatically not specialists. The demand signal points the other way at the same time: machine learning engineer postings sit well above their pre-2020 baseline while general software engineering postings sit far below it. Two markets are being described by one headline.
This is why the arithmetic that says “128,536 people are available, therefore my senior backend role gets easier” does not survive contact with a shortlist. The people leaving hyperscalers in volume are not, in the main, the people a Dubai scale-up is trying to hire. And a meaningful share of those who are have already been absorbed: OpenAI has said it intends to roughly double its headcount to around 8,000 by the end of 2026, and Amazon has signalled plans to hire in the region of 11,000 engineers and interns this year even while cutting elsewhere.
What changed in our own pipeline
Three things move reliably in the two weeks after a wave like this, and none of them is the one employers hope for.
Application volume rises, quality density falls. A posted role in Dubai attracts a wider spread of applicants, including many whose experience is adjacent rather than relevant. The screening load goes up in direct proportion. If your process depends on someone reading every application, the layoff wave has just made your hiring slower, and the effect is largest on the roles with the loosest job descriptions.
Inbound from people needing sponsorship rises sharply. A layoff in another jurisdiction frequently starts a visa clock, and the UAE is a rational destination when that clock is running. This is genuinely useful supply, but only for employers who already have the entity, the sponsorship route and the payroll mechanics in place. If you do not, the candidate will be gone before you have them. Our notes on employer of record and WPS payroll mechanics in the UAE cover what has to exist before you can move at that speed.
Your existing team reads the same headlines. This is the effect most employers miss entirely. A record layoff year makes your current engineers more risk-averse, not more grateful, and risk-averse engineers do not raise problems, do not push back on estimates and do not tell you when an approach is failing. If you take one operational action this month, make it a round of one-to-ones that explicitly addresses whether anyone thinks their role is at risk.
Seeing more applications and fewer good ones?
We screen against the requisition rather than the market, and we can tell you within a week whether the role or the pool is the constraint.
Let’s discuss itWhy the UAE reads differently from the headline
The cuts are overwhelmingly a United States story, driven by capital reallocation toward AI infrastructure at companies whose cost base is dominated by US payroll. The UAE is not on the same cycle. Demand here is driven by government digitalisation programmes, financial services build-outs across DIFC and ADGM, and a steady flow of regional expansion by companies that were never large enough to have a restructuring to announce.
That decoupling has a cost as well as a benefit. It means the local salary benchmark does not fall when the global headline does, and a candidate who has just been made redundant in San Francisco is not benchmarking themselves against Dubai comparables — they are benchmarking against what they were paid last month. Employers who go into these conversations expecting a discount because of the news cycle tend to lose the candidate at the offer stage and conclude the market is irrational.
Our expert view — the freeze is the expensive decision
Every time this cycle runs, we watch companies with healthy pipelines and committed delivery dates pause hiring because the industry looks bad in aggregate. The reasoning is never stated plainly, because stated plainly it collapses: Oracle’s capital allocation has no bearing on whether a fifty-person company in Dubai Internet City can deliver its Q1 roadmap. The cost of the freeze is not the delayed hire, it is the six weeks of pipeline work discarded and rebuilt from zero when the freeze lifts, plus the candidates who accepted elsewhere in the interval. Pause hiring when your own demand changes. Not when someone else’s does.
What we would change this month, and what we would not
Three adjustments are worth making while the wave is running, and one popular tactic is worth avoiding.
Tighten the job description before you touch the pipeline. Wider funnels expose loose requisitions. If your posting lists eleven technologies and one outcome, you will now receive several hundred applications you cannot rank. Rewrite it around what the person will own in their first ninety days, which also has the effect of filtering on the only signal that predicts anything.
Shorten the loop, not the standard. Genuinely strong displaced candidates are typically off the market within a month. If your process runs five stages over four weeks, you are structurally unable to hire them, and the people still available at week five are available for a reason. Compress the calendar and keep the bar.
Ask what the person did after the layoff. This is the highest-signal question available right now and almost nobody asks it. Redundancy says nothing about capability — it is usually a decision about a function, not a person. What they did with the following six weeks says a great deal.
What we would not do is price on distress. The softening in compensation expectations after a public layoff is real, modest and short-lived, and it never reaches the specialisations that were scarce to begin with. Hiring someone materially below market because their circumstances are temporarily weak is a retention problem with a twelve-month fuse: the discount becomes visible the moment they benchmark against a colleague, and you lose them at exactly the point they have become useful. If budget is the constraint, our method for building a developer salary benchmark in Dubai is a more durable answer than opportunism.
The pattern across our other markets
The same wave produces different local effects, and comparing them is the fastest way to see that the headline is not the mechanism. Our Singapore team sees the sponsorship-driven inbound earlier and more strongly, because regional mobility is higher and the notice periods are shorter. Our US practice sees the opposite problem: enormous supply at the mid-career level and no relief at all on the senior and ML-specialist roles that clients are actually trying to fill.
Three markets, one set of headlines, three different hiring realities. That is the argument against managing your requisitions from an aggregate number, stated as plainly as we can put it.
Our expert view — “AI did it” is doing a lot of work in these announcements
Oracle’s own filing attributes workforce reductions to AI adoption across its operations, and it is far from alone in framing cuts that way. We would treat the framing as partly real and partly convenient. Automation genuinely removes some work, but a restructuring attributed to AI reads better to investors than one attributed to over-hiring in 2021, a margin target, or capital being redirected into data centres. For an employer in Dubai the distinction is not academic: if you believe the AI explanation wholesale, you conclude that engineering headcount is structurally falling everywhere and you under-hire. What the postings data actually shows is redistribution — machine learning roles up, general software roles down — which is a very different instruction.
The bottom line for a Dubai employer
128,536 is a real number, it is a record, and it is close to irrelevant to whether you can fill your senior backend role this quarter. The layoffs are concentrated where your demand is not, the trackers disagree by more than fifty thousand people, and the local salary benchmark has not moved. What has changed is that your funnel is wider, your screening is slower, and the genuinely strong people in it will be gone in four weeks.
Treat the wave as an operational event rather than a strategic one. Tighten the requisition, compress the loop, ask candidates what they did with the last six weeks, and leave the compensation strategy where it was. And before you consider a freeze, check whether anything about your demand has actually changed. In our experience, it usually has not.
Frequently asked questions
How many tech workers have been laid off in 2026?
It depends entirely on which tracker you read, and the spread is wide enough to matter. Layoffs.fyi, cited in reporting on 11 September 2026, counted 128,536 employees across 299 companies by 10 September, which already exceeds the 122,606 it recorded for the whole of 2025 with nearly four months of the year still to run. Other trackers covering the same period report totals above 180,000, because they count differently: some include non-engineering functions, contractor terminations and multi-year programmes announced once and executed in stages. Neither number is wrong, and quoting either without saying which methodology produced it is how a hiring conversation ends up built on a statistic nobody can reproduce.
Does a record layoff year make it easier to hire engineers in Dubai?
It makes it easier to get applications and harder to fill a specific role. Volume rises immediately after a large public layoff, and the ratio of applications to qualified candidates gets worse rather than better, because the pool widens with people whose experience does not match the requisition. The roles that are hard to fill in Dubai were hard to fill before the layoffs and remain hard afterwards: senior engineers with production ownership, anyone with genuine machine learning deployment experience, and people willing to relocate. Layoffs concentrated in mid-career generalist roles at large employers do not produce those candidates in any volume.
Should we slow down hiring because of the layoff headlines?
Only if your own demand has changed. The most expensive decision we see in a month like this one is a freeze justified by somebody else’s restructuring, taken by a company whose roadmap has not moved and whose team is already under-resourced. Layoffs at hyperscalers reflect their capital allocation and their AI infrastructure commitments, not the staffing needs of a fifty-person company in Dubai Internet City. The correct input to your hiring decision is your own pipeline of committed work, your runway, and whether the role in question has a named owner waiting on it.
Are laid-off engineers cheaper to hire right now?
Marginally, briefly, and less than employers expect. Compensation expectations do soften in the first weeks after a public layoff, particularly among candidates with a visa clock or a notice period running out. That window closes quickly for strong candidates, who typically have offers within a month, and it never opens at all for the specialisations that are still scarce. Hiring on the assumption that a displaced engineer will accept materially below market is also a retention problem you are scheduling for twelve months out, because the discount becomes visible the moment the market recovers and they benchmark themselves against a colleague.
Trying to read this market before you commit a headcount?
We will tell you what your specific role looks like in the current UAE pool — including when the honest answer is that the requisition, not the market, is the problem.
Let’s discuss it