Amazon's Seattle Postings Fell 80%. Milan and Pisa Took Them.
Amazon Seattle job posts dropped 80% while Milan and Pisa became its #2 and #3 hiring cities. The AI layoff wave is offshore arbitrage. Source accordingly.
If you have been reading the Q1 2026 layoff coverage and feeling like the math does not add up, you are right. Amazon cut 16,000 corporate roles, most under an "AI efficiency" banner, while AWS posted its fastest growth in 13 quarters. The receipts are in the job postings, and they do not point at a robot. They point at Italy.
The 80% number is not what it looks like
JobsPikr's spring analysis pulled Amazon's posting volume by city and found something the press releases glossed over. Seattle postings dropped from 22,700 in H1 2025 to 4,540 in Q1 2026. Over the same window, Milan and Pisa rose to become Amazon's #2 and #3 largest hiring cities globally.
That is not a workforce being automated away. That is a workforce being relocated. Italian software engineers with Python and AWS skills earn a fraction of Seattle comp, and the established employer base in Milan, Turin, and Pisa (Capgemini, Microsoft, Prometeia, the CNR ecosystem around Scuola Superiore Sant'Anna) is deep enough to absorb a real hyperscaler buildout.
Forrester predicted exactly this in its 2026 Future of Work note: half of AI-attributed layoffs would be quietly reversed, with jobs returning offshore or at lower wages once the "AI-washing" collided with operational reality. Gartner went further in February 2026, projecting that 50% of companies attributing headcount reductions to AI will rehire staff to perform similar functions by 2027, often under different titles.
The Register reported the leading indicator a quarter earlier: 55% of employers already regret laying off workers because of AI, and AI-investment leaders expecting headcount to increase outnumber those expecting it to decrease nearly four to one (57% vs 15%).
How to tell arbitrage from genuine contraction
The useful insight from the JobsPikr report is not the Amazon headline. It is the heuristic. Pull a company's posting volume by city for the last 18 months and look at the ratio, not the total.
- Arbitrage signature: US hub down sharply, named offshore hub up. Amazon is the textbook case. Block shows the same shape, with San Francisco contracting while Melbourne sustains hiring.
- Contraction signature: Decline distributed proportionally across all cities. Atlassian fits this pattern. The business is genuinely smaller; nobody is being replaced by a Pisan.
The PR will read identically in both cases. The job postings will not. This is the single most actionable thing a sourcing lead can do this quarter: stop reading layoff press releases and start pulling a company's posting volume by metro before you build the candidate list.
This is also where most ATS-grounded tools fail you. LinkedIn Recruiter will not tell you that Amazon's Pisa headcount is climbing. JobsPikr will, but you still have to translate that into a candidate list. Plain-English search across GitHub, LinkedIn, and the open web is the missing layer, which is why we built Refolk: you describe the engineer you want and the geography you want them in, and you get a ranked shortlist instead of a 1,000-result cap.
The two-sided window
The arbitrage pattern creates two distinct sourcing opportunities, and they are on opposite continents.
Side one: senior US engineers, mispriced for 6 to 12 months
Q1 2026 tech layoffs hit 81,747, roughly half of all of last year's total in a single quarter. Seattle metro alone absorbed 16,590 Amazon and Microsoft cuts. San Francisco took 9,395. Menlo Park ate Meta's 1,500.
These engineers are being told, publicly, that their jobs were eliminated by AI. Most of them know that is not true. They watched the Pisa req go up the same week their team got the all-hands. They are angry, and they are mispriced, because the broader market has bought the AI-efficiency framing and assumes their skills are devalued.
They are not. Forrester's own data shows AI-investment leaders adding headcount, not cutting it. The 275,000 open AI jobs the laid-off workers cannot cross the skills divide to fill are real, but the population that was cut was not, on the whole, the population those reqs were waiting for. They were senior backend, platform, and infra engineers whose teams now exist in EU time zones.
For a US founder who needs senior ICs, this is the cleanest window in years. The candidates are available, the comp expectations are reset, and the conventional wisdom (that they were "AI roadkill") is wrong in a way that is easy to verify with a posting-data check.
Side two: the offshore hubs that just became real
Milan and Pisa are not theoretical anymore. Refolk's index of Italian software engineers with Python and AWS skills returns about 2,176 candidates, with Capgemini, Microsoft, and Prometeia among the top employers and Pisa, Turin, and Rome among the top regions. That is not a deep pool by Bay Area standards, but it is deep enough that Amazon, Block-equivalents, and US scale-ups will be competing for the same names within two quarters.
If you are a US company that has decided to open a European engineering presence (and a lot of Series B and C companies will, once they see what their Big Tech competitors just did), the move is to be in Milan and Pisa before Amazon's second wave of reqs goes up. The same applies in Melbourne if you are tracking Block.
The Klarna asterisk
Before anyone reads this as "AI cannot do the work," the Klarna story is worth keeping in mind, because it cuts both ways.
Sebastian Siemiatkowski cut roughly 700 customer service roles between 2022 and 2024, replacing them with an OpenAI-built chatbot that he claimed matched 700 agents. By early 2025, CSAT had collapsed, he publicly admitted the company "focused too much on efficiency and cost," and Klarna began rehiring human agents under an Uber-style flexible remote model. That is the canonical AI layoff boomerang, and it is real.
But Klarna is also the exception that proves the JobsPikr point. The customer service rep cuts were AI substitution. The engineering cuts at Amazon, Meta, and Oracle are something else. Customer service rep postings dropped 24.9% across 18 months before AI was widely cited as the cause. Meanwhile, AI engineer postings grew 654% from H1 2024 to H2 2025 at the same companies making the cuts. AI is replacing some functions. It is being used as cover for offshoring others. Treating those as the same story is how recruiters miss the window.
Treat every AI-layoff press release as an offshore-hiring leading indicator and start sourcing in the named geos within 30 days.
Sourcing senior engineers in 2026: a concrete playbook
Here is the sequence we have been recommending to founders and heads of talent who want to act on this.
1. Build a layoff-to-geography map for your top 20 target companies
For each company on your "would love to hire from" list, pull the H1 2025 posting volume by city and the Q1 2026 volume by city. Anything over a 40% delta with a corresponding rise in a named offshore hub is an arbitrage candidate. Anything that declined proportionally everywhere is a contraction candidate. Treat them differently.
2. Source the displaced US ICs in the first 30 days
The Bay Area senior hire cycle was already running 38 to 67 days. The newly displaced cohort is not going to last 90. The senior ex-Amazon platform engineers who got the Q1 2026 letter are taking calls right now, and they have not all updated LinkedIn yet, which is the actual problem. They are findable on GitHub, on conference talk pages, on the open web. They are not findable in your ATS or in LinkedIn Recruiter title search alone. This is the gap Refolk closes: ask for "senior infra engineers who left Amazon Seattle in the last 90 days" in plain English and skip the Boolean gymnastics.
3. Pre-position in the named offshore hubs
If Milan and Pisa are Amazon's #2 and #3 right now, they are someone else's #2 and #3 in nine months. Build a pipeline of 50 to 100 named senior ICs in each of the relevant European hubs before the next funding round forces you to do it on a deadline. Italian engineering hires from Politecnico di Milano and the Pisa CNR ecosystem still convert at sane comp ratios, but the curve will steepen.
4. Calibrate the AI-productivity claim privately
Executives are quoting AI productivity numbers in board decks that the underlying studies do not support. METR-style and GitClear-style analyses circulating in the industry coverage suggest, in some cohorts, that developers are slower with AI, not faster, and that AI-generated code contains substantially more defects than human-written code. We are not citing a specific number here because the methodologies vary and the writer who hand-waves a stat is the writer who gets it wrong. The point is: when a hiring manager tells you "we do not need to backfill, AI handles it," your prior should be that the backfill is already happening in another time zone.
5. Watch the Duke CFO number
The Duke CFO Survey in March 2026 polled 750 CFOs and projected roughly 502,000 AI-attributed job cuts in 2026, a roughly 9x increase over 2025's ~55,000. If Forrester and Gartner are right that half of those reverse, and the Amazon pattern holds, you are looking at the largest senior-IC supply shock and the largest offshore-hub buildout to land in the same 18 months. Plan staffing accordingly.
What to actually tell the candidate
Do not pitch the displaced ex-Amazon engineer as "AI roadkill" or as a layoff survivor. They know the Pisa req went up. Tell them you saw the geography move, you know what kind of engineer they are, and you are hiring in the city they live in for the work they were already doing. That is a 60-second pitch the offshore hubs cannot match, and it is the one thing the AI narrative has accidentally made cheap to deliver.
FAQ
Is the AI layoff wave real or fake?
Both. Some functions, notably customer service and parts of HR/recruiting, are being genuinely substituted by AI. Klarna and Meta's recruiting cuts (35 to 40% of Meta's announced 8,000) fit that pattern. But the senior engineering cuts at hyperscalers are largely offshore arbitrage with AI framing. The JobsPikr data on Amazon's Seattle-to-Milan shift is the cleanest evidence. Forrester and Gartner both project that roughly half of AI-attributed cuts will be reversed by 2027.
How do I tell which companies are doing arbitrage versus genuine contraction?
Pull their job postings by city for the last 18 months. Arbitrage looks like Amazon: US hub down 80%, a named offshore hub up sharply. Genuine contraction looks like Atlassian: declines distributed proportionally across all cities. The PR will read the same in both cases. The posting data will not.
Should I be sourcing in Milan and Pisa right now?
If you are building a US-only team, no. If you are considering a European engineering presence in the next 12 months, yes, urgently. The Italian Python/AWS pool is roughly 2,176 candidates by our index, deep enough to staff a regional team but not deep enough to survive a second hyperscaler entering. Pre-position before Amazon's next req wave, not after.
What is the single highest-leverage move this quarter?
Build a 50-name shortlist of senior ICs displaced from Amazon Seattle, Microsoft Redmond, and Meta Menlo Park in Q1 2026, before the market reprices them. The 38 to 67 day Bay Area senior hire cycle does not apply to a cohort that is angry, available, and currently mispriced. Move in the first 30 days or the window closes.