Career Intel with Dan 📊 | Same Boom, Different Bill

DEFINING PATTERN:

This week's developments don't point to a single, dramatic collapse in the job market. They point to something more diffuse: money and risk shifting to different places at once. Employers are redirecting capital toward AI infrastructure while trimming headcount elsewhere. The clearest wage and hiring damage is landing on the newest entrants to the labor market, even as demand explodes for the skilled trades workers building the AI boom's physical infrastructure. And in workplaces everywhere, employees are adopting AI faster than employers can govern it, which quietly shifts risk onto workers who have to decide, largely on their own, when and how to disclose it. None of this is one story. Together, it's a pattern worth naming.

🔊 1. Oracle Begins a New Layoff Round While Its AI Infrastructure Spending Keeps Climbing

Oracle started terminating employees again in mid-September, its third distinct round of workforce cuts in 2026. Some employees reported receiving termination notices by early-morning email, with system access cut off the same day. Oracle has not confirmed an official number for this round; outside estimates range from roughly 2,500 to 10,000 positions, and the company has repeatedly declined to comment on specifics.

This follows a fiscal 2026 in which Oracle's total headcount fell by about 21,000, or 13 percent, to roughly 141,000 employees, a figure that blends direct layoffs with unreplaced attrition. Oracle's own securities filings acknowledge that AI deployment has already contributed to workforce reductions and could continue to. At the same time, the company is pursuing an AI infrastructure buildout reported at tens of billions of dollars in commitments, and posted negative free cash flow of $23.7 billion for the fiscal year as it funds that expansion.

🔹 India's IT workforce is directly exposed: employees in Bangalore and Hyderabad have described the same 6 a.m. termination emails seen in the US, and reporting frames this round as a horizontal budget sweep across divisions rather than cuts targeted at one underperforming unit.

🔹 Oracle disclosed in its own SEC filing that AI deployment is a contributing factor in workforce reductions, one of the more direct corporate admissions of the AI-to-layoffs link so far this year.

🔹 The company added roughly $700 million to its restructuring budget this month, on top of an already-confirmed $2.8 billion in 2026 restructuring expense, tying the cuts explicitly to funding AI data-center capacity rather than a broader business downturn.

💡 The number that matters here isn't the headcount, which Oracle still won't confirm. It's the pattern: a profitable, growing company choosing to shrink its workforce specifically to redirect cash toward AI infrastructure, and saying so in its own regulatory filings.

Impact: Immediate for affected employees in the US and India; emerging as a template other AI-infrastructure-heavy companies may follow.

🔊 2. New Census Bureau Data Puts a Number on the AI-Era Entry-Level Squeeze, and It's Larger Than Expected

A working paper from the US Census Bureau's Center for Economic Studies, examining administrative records covering roughly 29 percent of bachelor's degrees awarded between 2016 and 2024, found that graduates from the college majors most exposed to AI saw their odds of holding a job one quarter after graduation fall by 5 percentage points, and their initial earnings fall by 13 percent, following the release of ChatGPT in late 2022. The three most-exposed majors, computer science, computer and information systems, and computer engineering, saw the largest declines. The researchers wrote that the earnings decline is comparable in magnitude to graduating into a severe recession.

About half of the lost earnings comes from lower pay within the industries that traditionally hire these graduates; the rest comes from graduates shifting into lower-paying sectors like retail and food service. The gap narrows over time, to about 5 percent two years out, and job-switching rates are higher, suggesting some catch-up, though the authors say there isn't yet enough data to know whether these graduates fully recover.

🔹 This directly challenges the assumption that a computer science or software degree guarantees an early-career wage premium, a message aimed squarely at career advisors, high school counselors, and university program planners currently recommending those majors.

🔹 The researchers explicitly frame this as underemployment, not unemployment: graduates are finding work, just in retail and food service rather than their field, which changes how advisors should talk about "landing a job" as a success metric.

🔹 Separate data from the Federal Reserve Bank of New York, cited alongside this study, put the unemployment rate for recent college graduates at 5.7 percent as of June, almost double the 2.9 percent rate for all college graduates, reinforcing that this is an entry-point problem specifically.

💡 Career Intel has covered the entry-level AI squeeze before, and this theme has run often enough that it needs a genuinely new reason to earn space again. This is that reason: the first time we've seen it quantified with this level of rigor, using actual administrative wage and employment records rather than survey responses or job-posting counts. The magnitude, on par with a recession, is what makes a familiar theme worth a second look this week.

Impact: Immediate for the current graduating cohort; long-term for lifetime earnings trajectories and how universities market STEM programs.

🔊 3. UK Workers Are Buying Their Own AI Tools Because Employers Haven't Caught Up

Deloitte UK's inaugural GenAI Workforce Survey, conducted with Ipsos among 25,000 workers and published September 15, found that two-thirds of UK workers have tried generative AI for work and nearly a quarter use it daily. Seventeen percent pay for at least one AI tool themselves, a collective bill Deloitte estimates at roughly £958 million ($1.4 billion) a year. Nearly a third of users, 31 percent, say they use generative AI without their employer's knowledge, so-called shadow AI. About half of workers using generative AI have had no formal training on safe use, and 65 percent say leadership has given them little guidance. Workers use it mostly for information search, drafting, and summarizing, reporting an average time savings of 70 minutes a week.

🔹 Sixty-four percent of weekly AI users told Deloitte they worry their manager might see their AI use as evidence their job could be automated, a fear that directly discourages the disclosure organizations would need to govern AI use responsibly.

🔹 Deloitte UK's chief AI officer, Hayley McKelvey, characterized this as workers simply not waiting for permission, which reframes "shadow AI" less as employee misconduct and more as an employer governance gap.

🔹 With half of AI-using workers reporting zero formal training, IT security and compliance teams are the ones absorbing the resulting data-handling and quality-control risk, not the workers making individually reasonable choices to save time.

💡 What caught my attention here isn't the adoption number, it's the stigma number: workers are using AI to be more productive while quietly worrying it will make them look replaceable. That tension is going to shape how honestly employees report their own AI use for a long time, unless employers change how they measure and reward it.

Impact: Immediate. The tools are already in daily use; the governance, training, and performance-evaluation catch-up is still emerging.

🔊 4. The AI Boom Has a Blue-Collar Labor Shortage, and It's Getting Cheaper to Fill

AI data-center construction is now competing directly with housing and other building projects for electricians, engineers, and construction crews, according to Reuters Breakingviews, which cited an Associated Builders and Contractors estimate that the US construction sector is short roughly 439,000 workers, with electricians and project managers in shortest supply. Separately, ZipRecruiter data shows job postings tied to AI data centers up 118 percent year over year, led by physical trades: electrical postings alone are up 293 percent and control-system technician postings up 134 percent. Median advertised pay for these roles, however, fell 7.8 percent to roughly $91,600, as hiring volume shifts from scarce senior engineers toward more standardized construction and operations trades.

🔹 Community colleges, apprenticeship programs, and workforce boards focused on electrical, HVAC, power, and data-center-operations pathways are looking at a training pipeline with real near-term demand, distinct from the software-side AI skills conversation that dominates most coverage.

🔹 The falling median pay figure is a specific, checkable warning sign for job seekers being recruited into this space: the earliest wave of hiring paid a premium for scarce expertise, and that premium is already compressing as employers standardize roles.

🔹 This demand is regionally concentrated near hyperscale construction zones, which means workforce organizations should track where AI data-center buildouts are actually happening locally rather than assuming this is a uniformly available opportunity nationwide.

💡 For someone already established as a licensed electrician or project manager, this is a seller's market. For someone entering an apprenticeship now hoping to ride the wave, the pay trend suggests they should ask what the job looks like in three years, not just what it pays this quarter.

Impact: Immediate hiring demand; emerging risk that pay and role standardization normalize faster than training pipelines can adjust.

🔊 5. Adecco Puts Generative AI in Front of 27,000 of Its Own Recruiters

Global staffing giant Adecco Group has expanded its enterprise generative AI deployment to more than 27,000 internal recruiters and staff across more than 40 countries, targeting automated candidate sourcing, initial screening, and matching. It's among the largest production rollouts of assistive and agentic AI inside a recruitment organization to date.

🔹 The stated goal is to shift recruiters from manual resume review toward exception handling and relationship management, meaning the entry-level recruiter task of first-pass screening is one of the first recruiting jobs being restructured by the tools recruiters themselves use to screen others.

🔹 Job seekers interacting with Adecco's pipelines will increasingly be filtered and surfaced by algorithmic tools before a human recruiter sees their application, which raises the same disclosure and bias questions that have followed AI hiring tools at individual employers, now at staffing-industry scale.

🔹 Career coaches helping clients through Adecco or similar staffing-agency pipelines should treat resume and application formatting for algorithmic screening as a practical, teachable skill, not a hypothetical one.

💡 There's a specific irony worth sitting with: the industry whose entire business model is matching people to jobs is now among the most aggressive adopters of AI to do that matching. Whether that improves outcomes for candidates or just speeds up the funnel is the open question.

Impact: Immediate for Adecco's own recruiting staff and the candidates in its pipelines; emerging as a signal for how the broader staffing industry moves next.

🔊 6. 2026 Tech Layoffs Pass 200,000, and Economists Are Increasingly Skeptical of the "AI Did It" Explanation

Independent trackers including Layoffs.fyi put 2026 tech-sector layoffs at roughly 209,000 through mid-September, with close to half of tracked layoff events citing AI or automation as a contributing factor. More than 5,000 positions were cut in early September alone across major employers including Oracle, Uber, and PayPal. But economists tracking these announcements increasingly flag what they call "AI-washing," where a company cites AI adoption as the explanation for cuts that are, on closer inspection, driven by ordinary cost-cutting, restructuring, or a return to a more conservative hiring posture after years of overhiring.

🔹 For workforce organizations advising laid-off workers, this distinction matters practically: a layoff genuinely driven by task automation calls for different reskilling advice than a layoff driven by financial restructuring where the same role may simply reappear at a competitor.

🔹 Employers have a real incentive to cite AI specifically, since "we're adopting cutting-edge technology" reads better to investors and the public than "we overhired and are now correcting."

🔹 Not every AI-cited layoff is AI-washing, and not every layoff should be waved off as AI-washing either. The honest answer requires looking at a specific company's hiring history, margins, and stated AI usage rather than taking either claim at face value.

💡 This is analysis, not a verified fact pattern for any single company: "AI-washing" is a real, evidence-supported phenomenon in some cases, but it's becoming a reflexive explanation people reach for anytime a layoff and an AI announcement land in the same month. Treat it as a hypothesis to check case by case, not a conclusion to assume.

Impact: Immediate for the workers affected; ongoing analytical challenge for anyone trying to measure AI's actual employment effect using layoff announcements as a proxy.

🔊 7. Blizzard's New Union Contract Makes AI a Subject Employers Have to Bargain Over

Blizzard's union reached a new contract that does not ban AI tools outright, but requires the company to bargain with the union before changing how AI is used in the workplace. It joins a growing list of union contracts, including recent agreements from the NewsGuild-CWA, SAG-AFTRA, and UFCW, that establish negotiated guardrails on workplace AI in the absence of federal legislation.

🔹 This gives unionized game-industry workers a formal seat at the table for decisions that, at most non-union employers, are made unilaterally by management with no worker input or advance notice.

🔹 It's a concrete, referenceable template other entertainment and tech unions can point to in their own negotiations, which matters for workforce organizations tracking how labor-side AI protections are actually being written, as opposed to proposed.

🔹 For non-union workplaces, this is a preview of the kind of AI-use policy workers are winning through collective bargaining rather than waiting for, since no comparable federal standard currently exists.

💡 Absent federal AI-at-work legislation, contracts like this one are quietly becoming the de facto rulebook. Career professionals working with unionized clients should know these provisions exist, and that they increasingly differ from the largely informal, unilateral AI policies at non-union employers.

Impact: Immediate for covered workers; long-term as a precedent other unions negotiate toward.

🔊 8. Europe's Auto Industry Is Shedding Tens of Thousands of Jobs, and AI Isn't the Reason

Bosch worker representatives are calling for EU action as the auto supplier plans to eliminate 13,000 jobs by the end of the decade. Separately, Volkswagen's broader restructuring plan reportedly includes up to 50,000 additional layoffs and as many as four potential plant closures starting in the early 2030s unless alternatives are found, with Porsche facing roughly 4,100 further cuts under the same plan.

🔹 These cuts are being driven by trade competition, tariffs, high production costs, and the industry's transition to electric vehicles, not by AI adoption, which makes this a useful direct counterexample whenever a layoff gets attributed to AI by default.

🔹 The scale here dwarfs most AI-cited layoffs covered this year, a reminder that AI is one driver of workforce disruption in 2026, not the only one, and workforce programs need sector-specific retraining strategies rather than a single "AI skills" response applied everywhere.

🔹 Skilled-trades and manufacturing workers in Germany's auto supply chain face a very different transition challenge than a laid-off software engineer, and conflating the two risks mismatched retraining investment.

💡 It's worth including this story precisely because it isn't about AI. Treating every layoff as an AI story, when the real driver is trade policy or a technology transition unrelated to generative AI, would be its own kind of distortion.

Impact: Emerging to long-term. These are multi-year restructuring programs, though the labor and policy debate around them is current.

🔊 9. Singapore Pairs Free AI Tool Access With Curated Training, a Model Worth Studying

Singapore's national workforce agency rolled out more than 200 AI courses alongside a six-month free subscription to premium AI tools, aimed at helping workers build both AI skills and general AI readiness.

🔹 The pairing of subsidized tool access with structured coursework, rather than training alone, directly addresses the access gap visible in the UK shadow-AI data above: workers without employer-provided tools either go without or pay out of pocket, and Singapore's approach removes that barrier at the policy level.

🔹 For US state and regional workforce boards, this is a concrete comparison point: most domestic AI-workforce initiatives fund training, but few also fund the tool access that makes the training immediately usable on the job.

🔹 This is worth tracking for outcomes over the next year, since a six-month free subscription window creates a natural point to measure whether workers who used it actually changed their AI usage habits once the free period ends.

💡 Training without tool access is a real gap in a lot of US workforce programming, and Singapore's model is a useful reference point for what closing that gap can look like at national scale.

Impact: Emerging. The program just launched; its effect on actual skills and job outcomes will take time to measure.

🔊 10. Slower Pay, Not Sudden Layoffs, Is How AI Is Showing Up in Many Knowledge-Work Paychecks

A New York Times analysis published September 16, drawing on research across 321 occupations, found slower wage growth in AI-exposed roles over the past three years, even as net employment in those occupations stayed broadly unchanged. The article also reported a pullback in job openings, especially for younger workers, in fields like journalism and economics.

🔹 This complicates a simple "AI is destroying jobs" narrative: the occupations studied mostly kept their headcount, meaning employers appear to be capturing AI-driven productivity gains through wage restraint and slower hiring rather than mass firing.

🔹 It reinforces a distinction useful for job seekers in these fields: the risk right now looks less like losing an existing job and more like a new opening not materializing, or an annual raise quietly shrinking, which is a harder thing to point to and advocate against.

🔹 Recruiters and career advisors working with knowledge-work clients should treat this as evidence that demonstrating judgment, domain expertise, and responsible AI use is what actually protects pay and advancement now, not just AI fluency on its own.

💡 The headline risk of AI at work keeps getting framed as layoffs. This data suggests the more common, quieter risk for now is a job that doesn't pay more this year than it did last year, in a role that technically still exists.

Impact: Emerging. This reflects a three-year trend, not a single event, but it's the clearest occupation-level wage data on this question so far.

BOTTOM LINE:

Put the pieces together and a specific kind of labor market comes into view: one where the total number of jobs isn't necessarily collapsing, but where entry points are narrowing, risk is moving downward onto workers and new entrants, and oversight is lagging well behind adoption. For career advisors, the practical takeaway isn't "AI is coming for jobs" in the abstract. It's that judgment, disclosure, verified outcomes, and adaptability are becoming the actual job requirements, on top of whatever technical skill got someone in the door.

It's also worth holding two competing facts at once this week: some of the most dramatic-sounding layoffs, like Europe's auto industry restructuring, have nothing to do with AI, while some of the quietest developments, like wage growth simply slowing across 321 occupations, may be where AI's real labor-market fingerprint shows up first. Both belong in the same weekly briefing.

Stay curious, stay current

Dan Lopez | danscareercorner.com