By Thorsten Meyer | October 31, 2026 | ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
On Monday, 26 October, two announcements pointed the same way. In The Hague, the Dutch government told parliament that its budget cut means roughly 25,000 fewer full-time jobs in central government by 2035. In Redmond, Microsoft offered a voluntary exit to about 60,000 employees, more than a quarter of its staff. A government and one of the world’s most valuable companies, on the same day, both getting smaller.
The YouTuber and author David Shapiro has a name for this pattern: “the automation imperative”. In his view, every organization, public or private, is pushed to automate, whether it wants to or not. Budgets push governments, shareholders and competitors push companies, and once one player automates, the rest have to follow. I took his examples as a tip, checked them against the original documents, added public and private organizations from Asia, Europe and the US, and asked two questions. Where is the argument strong? And where is it thinner than it sounds? Here is what I found, and what it means for companies and for the people who work in them.
Key takeaways
- The pressure is real and written down. The Netherlands books a cut to central government running costs that grows to €3.49 billion a year by 2035. Germany plans at least 8% fewer federal staff by 2029. Hong Kong is cutting 2% of civil service posts in each of two years. Budgets, not robots, set these numbers.
- AI is more often the tool than the reason. The Dutch letter itself warns: “Do not overestimate the staffing effects of digitalisation and AI.” A Microsoft spokesperson told a local radio station that AI is not driving its offer. India’s largest tech services firm blamed skill mismatches, not AI productivity gains. Japan’s Mizuho says its AI plan “is not a headcount reduction.”
- Some companies do name AI, in their own documents. Klarna’s annual report says its AI assistant does the work of more than 850 full-time agents, while its staff fell from 4,352 to 2,831 in two years. Lufthansa links 4,000 admin job cuts by 2030 to AI. Salesforce’s CEO said AI agents let him go from 9,000 to about 5,000 support staff.
- The tax code tilts toward machines in the US. Economists at MIT and Boston University put the effective tax on labor at 25.5 to 33.5%, against about 5% for equipment and software after the 2017 tax reform.
- The evidence lags the rhetoric. US employers named AI in about one in five announced job cuts this year, yet only 4% of New York-region service firms that use AI laid anyone off because of it.
- The aggregate risk is the real story. Each company’s cut can be rational. If everyone cuts at once, paychecks, which are other companies’ revenue, shrink. That is why China, California and New York are already writing rules about AI and jobs.
Products Worth Considering
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
Amazon Kindle Paperwhite (12th Generation, 2024), 16 GB, 7-inch Display, Jade
DJI Mini 5 Pro Fly More Combo Plus with DJI RC 2
As an affiliate, we earn on qualifying purchases.
Who calls it “the automation imperative”, and what it claims
The phrase comes from David Shapiro’s recent YouTube video “Automate or DIE”. Shapiro is a YouTuber and author who writes about what he calls “post-labor economics”. This article is not a review of his work, and it does not endorse or promote his book. I use his framing because it is a clear, testable claim, and I mark his views as his.
His argument, in short:
- Every organization is under constant pressure to be efficient. What it exists for is a result, not a headcount.
- In the public sector, the pressure comes from budgets. In the private sector, it comes from shareholders and from competition.
- Two forces accelerate it: expensive labor attracts labor-saving inventions, and the US tax code favors equipment and software over people.
- Three brakes slow it down: capability (can a machine do the job?), cost (is it cheaper, counting everything?) and risk (can we live with its mistakes?). He believes AI already passes the capability and cost tests for much office work, and increasingly the risk test too.
- His conclusion: Microsoft and the Netherlands are not exceptions but first movers, and only different incentives can change where this ends.
Shapiro’s video started from the same two announcements. The transcript was auto-generated, so I treated every name and number in it as a tip and checked each one against a primary source. What held up, and what did not, is listed near the end.
The Hague and Redmond: same day, different reasons
The Netherlands. On 26 October, state secretary Eric van der Burg sent parliament a progress letter on a “simpler, smarter and smaller” government. The cuts to central government running costs, booked by the previous and current cabinets, rise from €901 million in 2027 to €3.49 billion a year from 2035. The letter calls the jobs translation “a rough estimate”: around 25,000 full-time positions. At the end of 2025, the ministries, inspectorates and agencies it hits employed 160,016 full-time equivalents, about 41% more than in 2018. Police and defence are exempt. Many organizations face cuts of about 20%, some up to 45% or, in a single case, 60%.
AI appears as one tool among many: simpler rules, shared desks, standard IT, joint purchasing. The cabinet promises “ambitious targets and savings plans” for AI in the first quarter of 2027. And then, in the annex, comes the line that matters most for this debate. Among the lessons from twenty years of cuts: “Do not overestimate the staffing effects of digitalisation and AI.” Government processes were digitized and automated, the letter says, but the promised savings in cost and staff never came. Instead there were new tasks, more capacity and more contractors. I covered the Dutch plan in detail in Europe’s governments are letting AI in.
Microsoft. The same day, Microsoft’s chief people officer, Amy Coleman, announced a “Voluntary Exit Offer” for about 60,000 employees in the US, Canada, the UK and India, at vice-president level and below. Microsoft had about 223,000 staff at the end of June. Eligible employees decide by early December and leave on 31 December. The terms equal Microsoft’s standard severance. A company spokesperson told the Seattle public radio station KUOW that AI is not driving the buyouts. Meanwhile, on 28 October, Microsoft cut its capital spending forecast for the year to about $170 billion from $175 billion, CNBC reported, citing delays at outside GPU suppliers; much of that money goes into AI data centers. The details of the offer are in 60,000 buyout offers and a $95 billion stock sale.
So the two stories look alike, but the stated reasons differ. In The Hague, a budget cut comes first and AI is one way to cope. At Microsoft, AI is not doing the work so much as eating the budget. That distinction runs through every case below.
Infographic 01 · The world map
Who is shrinking, and what reason they give
Sixteen public and private organizations. Budgets set most of the numbers. AI is named as the reason less often than the headlines suggest.
Europe
EU members first
- NetherlandspublicBudget
≈ 25,000 fewer full-time jobs by 2035, a rough estimate behind a €3.49B yearly cut. Letter warns not to overestimate AI.
- GermanypublicAI a tool
At least 8% fewer federal staff by 2029. AI as a means, plus missing workers.
- DenmarkpublicRedeploy
AI to free ≥ 50M hours, ≈ 30,000 full-time jobs, by 2035. For shortages and quality.
- LufthansaprivateAI named
≈ 4,000 admin jobs by 2030, citing digitalization and AI.
- KlarnaprivateAI named
Staff 4,352 → 2,831 (2023–25). AI assistant: work of 850+ agents (estimate).
Asia
public and private
- Hong KongpublicBudget
−2% of civil service posts in 2026-27 and 2027-28; > 10,000 posts this term.
- ChinapublicBrake
Cautious pace for AI in labor-intensive industries; advance warning encouraged.
- Mizuho · JapanprivateRedeploy
AI to take over the work of ≈ 5,000 of ≈ 15,000 clerical jobs in 10 years. “Not a headcount reduction.”
- DBS · SingaporeprivateAI named
≈ 4,000 temporary and contract roles not renewed over 3 years (announced Feb 2025).
- TCS · IndiaprivateNot AI
≈ 12,000 jobs, about 2% (2025). CEO: skill mismatch, not AI productivity gains.
United States
public and private
- Federal agenciespublicNot AI
≈ −256,000 staff, more than 11%, Dec 2024–Jan 2026. Directives, buyouts, hiring limits.
- CaliforniapublicNew rules
Order N-6-26: study AI’s job effects, reform layoff notices. No ban.
- MicrosoftprivateNot AI
≈ 60,000 exit offers, about 27% of staff. Spokesperson: AI not driving it.
- AmazonprivateLayers
16,000 corporate jobs (Jan 2026) after 14,000 (Oct 2025). Layers, culture.
- SalesforceprivateAI named
Support staff ≈ 9,000 → 5,000, CEO says, as AI agents handle half of chats.
- KrogerprivateTechnology
≈ 3,000 corporate jobs over 3 years. Memo: “Technology is reshaping what’s possible.”
Reasons as stated by each organization or, where marked, as reported. A category is a summary, not a verdict on the real cause.
Sources: Dutch letter to parliament, 26 Oct 2026, German Digital Ministry, Danish AI taskforce, Lufthansa, 29 Sep 2025, Klarna Form 20-F, Hong Kong Civil Service Bureau, 16 Mar 2026, State Council, gov.cn, 23 Oct 2026, Mizuho via Bloomberg, DBS via Fortune, TCS statement, GAO-26-108583, California EO N-6-26 (Akin Gump summary), Microsoft via KUOW, Amazon via CNBC, Salesforce via Fortune, Kroger via Fortune. Translations by Thorsten Meyer.
Powered by Thorsten Meyer AI
Europe: budgets, shortages, and two companies that name AI
Germany (public). The federal cabinet’s modernization agenda, adopted in October 2025, says: “By 2029, we will reduce the federal workforce by eight per cent and cut material costs in the federal administration by ten per cent.” The coalition agreement covers the ministerial and Bundestag administration and certain subordinate agencies, at least 8%. AI shows up as a means, not as the reason: the agenda wants AI applications expanded across the federal administration and calls modern IT a “prerequisite” for getting by “with fewer staff”. It also names a second driver: demographic change “demands innovative approaches to compensate for missing workers”.
Denmark (public). The Danish government, the municipalities (KL) and the regions share a goal that AI should free up at least 50 million hours, the equivalent of at least 30,000 full-time jobs, across the public sector by 2035. The official target says freed capacity should ease labor shortages, raise service quality and, in some cases, allow money to be moved to other priorities. Layoffs are not mentioned. It is an AI-efficiency target, not a cut.
Lufthansa (private). At its capital markets day on 29 September 2025, the airline group announced a “reduction of 4,000 administrative jobs by 2030 through digitalization, automation, and process consolidation”, mostly in Germany. Its statement gives the reason: “the increased use of artificial intelligence will lead to greater efficiency in many areas and processes.” The cuts are to be made in consultation with unions and employee representatives.
Klarna (private). The Swedish payments company is the clearest case of a company putting AI in its legal filings. Its 2025 annual report on Form 20-F says the AI assistant handled 80% of customer service chats and did “the work equivalent of over 850 full-time agents”, an estimate, saving about $59 million. Full-time employees fell from 4,352 at the end of 2023 to 2,831 at the end of 2025, a decline of about 35%, which the filing attributes to “our strategic decision to reduce our overall headcount and drive operational efficiency by leveraging AI”. There is a twist: in May 2025, chief executive Sebastian Siemiatkowski told Bloomberg that cost had been “a too predominant evaluation factor” and that the result was “lower quality”, and Klarna began recruiting human agents again in a flexible, remote setup. The risk brake, in other words, is not theoretical.
Outside the EU, Britain’s finance minister said in March 2025 that the civil service could shrink by 10,000 jobs as departments cut running costs by 15% by the end of the decade, partly through technology and AI.
Asia: shrinking, redeploying, and one government on the brakes
Hong Kong (public). According to a Civil Service Bureau paper for the Legislative Council from 16 March 2026, the government is reducing the civil service establishment by 2% in each of 2026-27 and 2027-28, under its Productivity Enhancement Programme. With posts deleted under a zero-growth policy since 2021-22, it expects a cumulative cut of more than 10,000 posts this term, to about 188,000 posts. The paper names “digitalisation” and “application of technology” as means and adds AI training for civil servants. The reason is fiscal and productivity-driven; AI is a tool.
China (public). On 23 October 2026, the State Council’s General Office published guidelines on AI and employment. They tell authorities to put AI first into jobs with labor shortages or dangerous conditions and to manage “steadily and cautiously” the scope, intensity and pace of AI in labor-intensive industries. Companies are encouraged, not required, to explore an “unemployment buffer period” in which they warn affected workers in advance and help them move. AI projects that need state approval and may cause a large employment shock get an employment risk assessment. This is a government leaning against the imperative, not with it. (Translations of the official Chinese text are mine.)
Mizuho (private). Japan’s third-largest bank plans for AI to take over the work of about 5,000 of its roughly 15,000 clerical positions in Japan over ten years, Bloomberg and Nikkei reported in February. AI will check documents for account openings and transfers and register customer data. The bank’s own statement: it wants to boost earnings “by shifting human resources to our focus areas by fully utilising AI.” And: “It is not a headcount reduction.” Affected staff move to other roles. In a country with a shrinking workforce, AI looks more like a replacement for workers who are not there than for workers who are.
DBS (private). In February 2025, the then chief executive of Singapore’s largest bank said its workforce would shrink by about 4,000 over three years. The bank clarified that AI could reduce the need to renew about 4,000 temporary and contract roles, through attrition, and that permanent staff would not be affected (Reuters, Fortune).
TCS (private). India’s largest tech services company said in July 2025 that it would release about 2% of its global workforce, around 12,000 people, mostly middle and senior grades, as part of becoming a “Future-Ready organisation”. Its chief executive told Moneycontrol it was “not because of AI giving some 20 percent productivity gains” but because of skill mismatches. Reuters noted that TCS did not answer how many layoffs were tied to AI adoption.
United States: cut first, automate later
The federal government (public). Across the 22 largest federal agencies, the workforce fell by nearly 256,000 people, more than 11%, between December 2024 and January 2026, according to the Government Accountability Office. The reasons GAO lists are presidential directives, incentives to resign or retire, reductions in force and hiring limits. AI is not among them. The order is the reverse of the imperative story: the cuts came first, and AI comes later to cope. The Office of Personnel Management, which lost 35% of its own staff, proposed in its 2027 budget request “using AI tools and modernizing its IT systems in offices with fewer staff”, GAO reports.
California (public). On 21 May 2026, Governor Gavin Newsom signed Executive Order N-6-26. It does not restrict AI. It orders state agencies to study AI’s effect on jobs, build a dashboard from unemployment insurance data, and recommend updates to the state’s mass layoff notice law, due around mid-November. It also asks the state’s business office to look at ways for workers to share in AI-driven gains through ownership, according to summaries by the law firms Akin Gump and A&O Shearman.
Amazon (private). In June 2025, chief executive Andy Jassy wrote to staff that “in the next few years, we expect that this will reduce our total corporate workforce” as AI spreads through the company. When Amazon cut 14,000 corporate jobs in October 2025, he told analysts the move was “not really AI-driven, not right now at least. It really is culture.” In January 2026, Amazon cut another 16,000 corporate roles; its official post spoke of “reducing layers, increasing ownership, and removing bureaucracy.”
Salesforce (private). In a podcast interview in August 2025, chief executive Marc Benioff said: “I’ve reduced it from 9,000 heads to about 5,000, because I need less heads,” referring to support staff, as AI agents now handle about half of customer conversations. The company told Fortune it no longer needed to backfill support engineer roles and had redeployed hundreds of employees.
Kroger (private). In October 2026, chief executive Greg Foran told staff in a memo that the grocer would cut about 3,000 corporate jobs over three years, about 37% of its corporate workforce, Business Insider and Fortune reported. “Technology is reshaping what’s possible,” he wrote. Headlines called it an AI move; the memo, as quoted, speaks of technology, duplicated work and lower prices.
Why everyone moves the same way: the argument, piece by piece
Budgets. For governments, the pressure is plain. The Dutch letter, the German agenda and the Hong Kong paper all start with money and staff numbers, then ask how to get there. Shapiro’s point that an agency exists for its outcomes, not its payroll, is a fair description of how these documents are written.
Shareholders. Shapiro cites Milton Friedman’s 1970 essay in the New York Times Magazine, “The Social Responsibility of Business Is to Increase Its Profits.” Friedman argued that executives work for the owners, who generally want as much money as possible. What usually gets left out is his condition: within the basic rules of society, “both those embodied in law and those embodied in ethical custom”. In Friedman’s own terms, how a company cuts, and what it tells people, is part of the job.
Competitors. In Lewis Carroll’s Through the Looking-Glass, the Red Queen tells Alice that “it takes all the running you can do, to keep in the same place.” Economists use the image for competition: once one firm lowers its costs, rivals must match, find a better product or lose share. Shapiro compares this to a ratchet: once a process is automated somewhere, the rest of the industry follows. That is his claim; it is plausible for processes where costs are easy to compare and weaker where quality and trust decide, as Klarna found.
Expensive labor. In The Theory of Wages (1932), the economist John Hicks wrote that “a change in the relative prices of the factors of production is itself a spur to invention”, directed at saving the factor “which has become relatively expensive”. Expensive labor attracts labor-saving machines. That is why well-paid office work, not only factory work, is now in the frame.
Taxes. In a 2020 paper for the Brookings Papers on Economic Activity, Daron Acemoglu, Andrea Manera and Pascual Restrepo estimated effective US tax rates of 25.5 to 33.5% on labor, against 10% on software and equipment in the 2010s and 5% after the 2017 tax reforms, down from about 20% in 2000. Their conclusion: the US tax system has “promoted levels of automation beyond what is socially desirable.” This is a model-based estimate for the US, not a law of physics, and other countries tax differently. But it supports Shapiro’s point that policy, not only technology, sets the incentives.
The brakes. Shapiro names capability, cost and risk. The evidence says all three still bite. An MIT study from January 2024 found that at the costs of the time, only 23% of the wages paid for tasks that computer vision could do were attractive to automate. Klarna’s return to human agents shows the risk brake. And the Dutch government’s own lesson shows a fourth brake that Shapiro does not list: organizations that digitize without simplifying their rules tend to create new work.
Infographic 02 · The forces
Why the pressure points one way, and what still slows it
Three pressures push every organization toward automation. Two forces speed it up. Three brakes, plus one Shapiro leaves out, slow it down.
Three pressures
David Shapiro’s framing, with the primary sources behind it
Budgets
The Dutch cut grows to €3.49B a year. Germany: −8% staff. Hong Kong: −2% posts, twice.
Shareholders
Friedman, 1970: business should increase its profits, within the rules of law and “ethical custom”.
Competitors
The Red Queen: “all the running you can do, to keep in the same place.” Rivals copy cost cuts.
Two accelerants
Expensive labor invites labor-saving inventions
John Hicks, 1932: a change in relative prices is “a spur to invention”, aimed at the factor “which has become relatively expensive”. Well-paid office work is now in the frame.
The tax code tilts toward machines
Effective tax rates, model estimates
Three brakes, and a fourth
Shapiro names capability, cost and risk. The evidence says all of them still bite.
Capability
Can a machine do the whole job? Often only parts of it.
Cost
At 2024 costs, only 23% of the wages paid for computer-vision tasks were worth automating (MIT).
Risk
Klarna recruited human agents again in 2025 after its CEO said cost focus had brought “lower quality”.
Sources: Milton Friedman, New York Times Magazine, 13 Sep 1970; Lewis Carroll, Through the Looking-Glass (1871); John Hicks, The Theory of Wages (1932); Acemoglu, Manera, Restrepo, Brookings Papers 2020; Svanberg et al., MIT, Jan 2024; Bloomberg (8 May 2025, as reported); Dutch letter to parliament, 26 Oct 2026. Framework: David Shapiro, “Automate or DIE” (YouTube), his view.
Powered by Thorsten Meyer AI
Where the argument is strong, and where it is thin
Strong: the pressure itself. The budget targets are official and dated. The tax tilt is measured. Competition is real, and the companies that name AI in their filings, like Klarna, show what happens when a process can be measured and automated end to end.
Thin: the leap from pressure to inevitability, and from announcement to cause.
- What companies say is not what they measure. US employers named AI in 120,136 announced job cuts from January to September 2026, about 21% of the total, according to Challenger, Gray & Christmas. Yet only 4% of service firms in the New York region that use AI said they laid off workers because of it in the six months to August, the New York Fed found. I took this gap apart in Is AI really behind the layoffs?.
- Several of the biggest cuts are not AI cuts by their own account. Microsoft’s spokesperson, TCS’s chief executive and Amazon’s chief executive all said so, at least for the cuts at hand. You can doubt them, but then the evidence for the AI story has to come from somewhere else.
- Where AI is the tool, the reason is often labor scarcity. Denmark, Germany and Japan’s Mizuho all describe AI as a way to cope with missing workers. That is a different story from “AI replaces you”, with different consequences.
- Rules can change the gradient. Shapiro argues that neither shaming nor voting will stop the imperative; only different incentives will. But China’s State Council, California’s governor and New York’s layoff form show governments starting to change the costs of cutting jobs for AI. Whether those rules work is open. That they exist shows the outcome is not fixed.
My reading, clearly as opinion: the pressure to automate is real and will grow, but “imperative” overstates how fast and how evenly it works. Budgets and competitors push. Evidence lags the headlines. And rules, from tax codes to layoff notices, decide who pays.
Infographic 03 · Said vs. measured
The pressure is real. The evidence for AI as the cause is thinner.
What organizations announce, what surveys and studies measure, and where the automation-imperative argument holds.
Where the argument is strong
- Budget targets are official and dated: €3.49B a year (NL), −8% staff (DE), −2% posts twice (HK).
- The US tax tilt is measured: labor 25.5–33.5% versus about 5% for equipment and software.
- Where a process can be measured end to end, firms do cut and say so in filings (Klarna).
Where it is thin
- The said-versus-measured gap: 21% of announcements name AI, 4% of AI users cut because of it.
- Several big cuts are “not AI” by their own account: Microsoft, TCS, Amazon in 2025.
- Where AI is the tool, labor scarcity is often the reason: Denmark, Germany, Mizuho.
- Rules can change the incentives: China, California and New York are writing them.
The four numbers measure different things and cannot be subtracted from each other. They show the size of the gap between what is said and what is measured.
Sources: Challenger, Gray & Christmas, 1 Oct 2026, New York Fed, 1 Sep 2026, Svanberg et al., MIT, Jan 2024, Autor, Chin, Salomons, Seegmiller, QJE 2024, Dutch letter to parliament, 26 Oct 2026, Klarna Form 20-F, Acemoglu, Manera, Restrepo, Brookings Papers 2020.
Powered by Thorsten Meyer AI
The part that should worry everyone: paychecks are someone’s customers
Shapiro’s strongest point is the one he borrows from Martin Ford. In The Lights in the Tunnel (2009), Ford argued that jobs are the main way purchasing power reaches consumers. One company cutting its payroll is rational. If every company does it at once, the paychecks that disappear are other companies’ customers. The same logic applies to the state: a smaller civil service also means fewer household incomes.
This is a risk, not a law of nature. Economists David Autor, Caroline Chin, Anna Salomons and Bryan Seegmiller found that about 60% of US employment in 2018 was in job titles that did not exist in 1940. New work does appear. The problem is timing and distribution: new jobs do not arrive on schedule, in the same places, or for the same people. Who pays for the gap is exactly the question behind Larry Ellison’s “universal high income” and South Korea’s basic income pilot.
What this means for companies
- Measure before you announce. If AI is the reason, show the process it took over and the result. If it is a bet on what AI might do, say so. The Dutch government’s own lesson is that digitizing without simplifying brings more work, not less.
- Tell one story. Staff, investors and regulators now compare the press release, the filing and the layoff notice. Klarna’s filing and Amazon’s earnings call are public records.
- Decide where freed time goes. Denmark wrote down that freed hours go to shortages and service quality. Companies that make that choice early keep the people and knowledge they will need for the next product.
- Budget for the brakes. Quality, error rates and customer trust have a cost. Klarna’s return to human agents is a cheap lesson to learn from someone else.
- Watch the rules. China’s guidelines, California’s order and New York’s layoff form point the same way: more disclosure about AI and jobs. Plan for being asked.
What this means for workers and users
- Ask which kind of cut it is. Did AI do the work, did AI eat the budget, or is AI the story? The answer tells you whether the job may come back and what to learn next.
- Read the first signals. Voluntary exit offers, hiring freezes and “not backfilling” roles usually come before layoffs. Microsoft’s offer and Salesforce’s unfilled support roles are examples.
- Learn the tools your employer actually uses. Mizuho and DBS pair their AI plans with retraining. Employers that redeploy prefer people who already know the tools.
- As a customer, check the human route. Where service moves to AI, ask how to reach a person. Klarna’s chief executive called that “so critical” for trust.
- As a citizen, follow the budget, not the buzzword. In public services, cuts are set in budgets first. Parliamentary debates, like the Dutch one on 1 December, decide which services shrink.
What to watch next
- Early November: Challenger’s October job cuts report; California’s recommendations on its layoff notice law (due around mid-November).
- 1 December: the Dutch parliament debates the “decisive government” plan.
- Early December and 31 December: the decision window and leaving date for Microsoft’s exit offer. Microsoft expects take-up well below the roughly 30% of its spring retirement offer.
- First quarter of 2027: the Dutch cabinet’s AI targets and savings plans; within six months, its multi-year plan for central government.
- 2027 onward: whether Germany reports progress on its 8% target, and whether Hong Kong’s second 2% cut proceeds.
Transcript check: what held up, and what didn’t
- Held up: the Dutch numbers (25,000 full-time jobs as a rough estimate, about 160,000 staff, +41% since 2018, €3.5 billion a year, police and defence exempt, cuts of 45% or 60% in a few organizations), Microsoft’s 60,000 offers out of about 223,000, the spring retirement offer taken by about 30% of roughly 8,750 eligible staff, the July cut of about 4,800 jobs, Denmark’s 30,000 full-time equivalents by 2035, Amazon’s 16,000 corporate jobs, Kroger’s roughly 3,000 corporate jobs, Friedman, Hicks, Carroll and Ford.
- Needed precision: the US tax figure. Labor is taxed at 25.5 to 33.5%, not “about 25%”. Equipment and software fell to about 5% after the 2017 reforms (Acemoglu, Manera, Restrepo, 2020).
- Overstated: “HSBC is cutting 20,000 jobs.” Bloomberg reported in March 2026 that HSBC was weighing cuts of up to about 20,000 roles over several years, with no final decision. In October, its finance chief said AI “will impact some roles, it will transform other roles”.
- Garbled in the transcript: “Wisec” is WiseTech Global, an Australian logistics software company that announced about 2,000 job cuts, nearly a third of its workforce, in February 2026, citing AI (Reuters, ABC). “Block cut nearly half”: Block cut more than 40%, from over 10,000 to just under 6,000.
- Timing: Britain’s 10,000 civil service cut was announced in March 2025. Microsoft’s capital spending forecast is about $170 billion for the year, after a cut on 28 October.
- Not a count by Reuters: Reuters keeps a running factbox of companies that tied job cuts to AI. The number of companies on it depends on when you read it, so I do not quote one.
Honest limits
- Stated reasons are what organizations say. Several companies here have reasons to emphasize or play down AI. Where possible I used filings and official documents; media reports are marked as reported.
- The Dutch, German and Hong Kong figures are targets and plans, not results. The 25,000 Dutch figure is the government’s own rough estimate.
- The US tax estimates come from one well-known study and depend on its model. They describe the US, not Europe or Asia.
- The MIT cost study is from January 2024 and covers computer vision tasks only. AI costs have fallen since, so the share worth automating today is likely higher.
- The framework comes from David Shapiro’s video “Automate or DIE”. I used its auto-generated transcript only as a tip and checked every fact against primary sources or named outlets. Apart from the term itself, no quote in this article comes from the transcript.
- Translations from Dutch, German and Chinese are mine. Nothing here is investment, legal or career advice.
Sources
Dutch Ministry of the Interior (BZK), Voortgangsbrief Slagvaardige Overheid with annex “Taakstelling; cijfermatig overzicht en geleerde lessen”, letter to the Tweede Kamer, 26 October 2026 (tweedekamer.nl).
GeekWire, Microsoft makes new “voluntary exit” offer (26 October 2026); CNBC, Microsoft voluntary buyouts (26 October 2026); KUOW (26 October 2026); CNBC on Microsoft’s capital spending (29 October 2026).
German Federal Ministry for Digital Affairs and State Modernisation, Government modernisation and the Modernisierungsagenda für Staat und Verwaltung (Bund) (October 2025).
Danish Digital Taskforce for AI, Målbillede (June 2025).
Lufthansa Group, Lufthansa Group is consistently pursuing its strategy (29 September 2025).
Klarna Group plc, Form 20-F for 2025 (filed 26 February 2026); Bloomberg, Klarna Turns From AI to Real Person Customer Service (8 May 2025), as reported; CX Dive.
Civil Service Bureau, Hong Kong, Civil Service-related Matters Featured in the 2026-27 Budget, LegCo Panel on Public Service (16 March 2026).
General Office of the State Council of China, Opinions on adapting to the development of AI technology to support employment-friendly development (Guo Ban Fa [2026] No. 29, dated 16 October, published 23 October 2026).
Bloomberg via The Star, Mizuho plans to replace 5,000 clerical jobs with AI in 10 years (March 2026); UPI (27 February 2026).
DBS: Reuters and Fortune (24-25 February 2025).
TCS: company statement via Business Standard (27 July 2025); Moneycontrol interview with K. Krithivasan (21 July 2025), as reported.
US Government Accountability Office, GAO-26-108583 (17 June 2026) and GAO-26-108916 (20 July 2026).
California Executive Order N-6-26 (21 May 2026), summarized by Akin Gump and A&O Shearman.
Amazon, Andy Jassy on generative AI (June 2025); Fortune on the October 2025 call; CNBC (28 January 2026).
Salesforce: Fortune (2 September 2025), on Marc Benioff’s interview on The Logan Bartlett Show.
Kroger: Fortune (14 October 2026), citing Business Insider.
HSBC: Bloomberg (19 March 2026), as reported; AFP via Malay Mail (27 October 2026). WiseTech Global: ABC News (25 February 2026). UK: Civil Service World (March 2025).
Milton Friedman, A Friedman Doctrine: The Social Responsibility of Business Is to Increase Its Profits, New York Times Magazine (13 September 1970). Lewis Carroll, Through the Looking-Glass (1871). John Hicks, The Theory of Wages (Macmillan, 1932). Martin Ford, The Lights in the Tunnel (2009).
Daron Acemoglu, Andrea Manera, Pascual Restrepo, Does the US Tax Code Favor Automation?, Brookings Papers on Economic Activity (Spring 2020).
Maja Svanberg, Wensu Li, Martin Fleming, Brian Goehring, Neil Thompson, Beyond AI Exposure: Which Tasks are Cost-Effective to Automate with Computer Vision? (MIT, January 2024).
David Autor, Caroline Chin, Anna Salomons, Bryan Seegmiller, New Frontiers: The Origins and Content of New Work, 1940–2018, Quarterly Journal of Economics (2024).
Challenger, Gray & Christmas, September 2026 job cuts report (1 October 2026); Federal Reserve Bank of New York, Businesses Are Using AI to Transform Work, Not Cut Jobs (1 September 2026).
Framework: David Shapiro, “Automate or DIE”, YouTube video on the Dutch and Microsoft announcements (October 2026). Used as a tip; his views are marked as his.
Claims by companies are quoted as claims. Media reports are marked as reported. Nothing here is legal, career or investment advice.
