October 13, 2026
Day91 Labs Brief — Issue 015: Special edition · AI & workforce
Oracle is the only large vendor whose filings say AI is cutting jobs. Most companies say nothing, and several deny it. Meanwhile revenue per employee is up 16–21% at big tech and flat at IT services, and the people AI is keeping out are the newest workers, not the experienced ones.
1. Oracle is the only large vendor whose filings say AI is cutting jobs.
Evidence tier A — 10-Q note and 10-K risk factor opened. Plan total is our calculation.
Oracle's 10-Q for the quarter ended 31 August 2026 (Note 4) describes its 2026 Restructuring Plan, budgeted at up to $2.1 billion, as working "including through the adoption and integration of artificial intelligence technologies across certain functions." After the quarter, management added about $700 million of further actions, taking the plan to about $2.8 billion (calculated). Oracle has spent $1,971 million on the plan so far. Revenue was $19,345 million against $14,926 million a year earlier, up 29.6% (calculated). Its FY26 10-K reports about 141,000 employees at 31 May 2026 and says AI deployment has resulted, and may continue to result, in workforce reductions.
Most companies cutting staff this year describe it as focus or simplification. Oracle's lawyers wrote AI into the reason, in two filings, while revenue grew almost 30%. That is a growing company choosing to do more of the work with software, not a struggling one cutting costs.
If Oracle is a partner, competitor or vendor of yours, expect changes in its account teams and support contacts over the next two quarters. Confirm who owns your relationship now, before the next renewal or joint deal.
Sources
2. Who says it's AI, and who says it isn't.
Evidence tier A — Company documents opened for HubSpot, Block, Cloudflare and Cognizant. Intuit's call quote and Etsy's letter are B.
Reading each company's own document on its 2026 cuts: Block's shareholder letter (26 Feb) ties a cut of more than 40%, from over 10,000 staff to just under 6,000, to a smaller team using its own AI tools. Cloudflare's founders (7 May, more than 1,100 roles) say the company must be built for "the agentic AI era" and that it is not a cost-cutting exercise. Cognizant's Project Leap 8-Ks describe building an "AI-enabled" workforce, with $230–320 million of charges, while headcount still rose 3.8% year on year. HubSpot's CEO memo (about 7%) says the cut "is not driven by AI-related efficiencies." Intuit's 8-K for its 17% cut says nothing about AI; on the call its CEO said none of it had to do with AI (B). Etsy's CEO letter says the same of its 12% cut (B).
Two companies blame AI, two deny it, and the rest say nothing. The denials and the silences don't mean AI played no part; they mean the company chose not to say so. Block and Cloudflare are the cleanest test cases because they put the reason in writing and will have to show the results.
When a vendor or competitor cuts, read the reason it gave in its own document, not the coverage. A company that blamed AI is telling customers it expects to serve them with fewer people, so ask how support and account coverage will change.
Sources
3. Revenue per employee is up 16–21% at big tech and flat at IT services.
Evidence tier A — Revenue and headcount from company releases and filings. Every revenue-per-employee figure is calculated.
Calculated from each company's latest reported revenue and period-end headcount, annualising quarters. Microsoft: 223,000 staff, down 2.2%, revenue up 18%; about $1.49 million per employee, up 21%. Alphabet: 198,933 staff, up 6.3%, revenue up 24%; $2.41 million, up 17%. Amazon: 1,595,000 staff, up 3.2%, revenue up 20%; $503,000, up 16%. Accenture: about 814,000 staff, up 4.5%, revenue up 6.5%; $91,100, up 1.9%. Cognizant: 356,700 staff, up 3.8%; $61,500, up 0.7%. Wipro: 243,044 staff, up 4.2%, IT services revenue up 1.0%; $43,000, down 3.1%.
The platforms are growing revenue with flat headcount. The services firms are still adding people at the same pace as revenue, and in Wipro's case faster. If AI is making services delivery more efficient, the savings are going to clients at renewal, not to the firms' margins. Infosys's CFO described it on the July call as "AI led deflation."
If you sell through or alongside systems integrators, expect them to ask for lower prices or more scope at renewal as their own pricing comes under pressure. If you buy services, ask your SI to show where AI has reduced the hours it bills you.
Sources
4. Salesforce keeps engineers flat, lets agents take support, and hires 1,200 more sellers.
Evidence tier A — Salesforce's own published transcripts and Q2 release.
On its Q1 FY27 call, Salesforce said it had about 15,000 engineers, mostly flat for about two years, and that its help agent had handled four million inquiries, double what human agents handle. On the Q2 call it said agents had held more than five million conversations, 64% resolved without a person. Its sales chief said the company is reaching 15,000 account executives and still has to hire 1,200 more before the end of the year. The Q2 release puts Agentforce ARR above $1.5 billion, up more than 240%.
This is the clearest picture of where one large vendor is moving its people: fewer in building and support, more in selling. The work AI absorbs first is the work that doesn't face a buyer.
Expect more Salesforce sellers in your accounts in the fourth quarter, many of them new and carrying an Agentforce number. If you compete with Salesforce, get ahead of the agent conversation in your accounts before they do.
Sources
5. AI is the top stated reason for job cuts this year, but monthly AI-cited cuts are falling.
Evidence tier A — Challenger's own monthly reports opened. Shares are calculated.
Challenger, Gray & Christmas counted 573,195 announced job cuts from January to September 2026, down 39% on the same period of 2025. Employers cited AI for 120,136 of them, 21.0% (calculated), more than any other reason. Technology companies announced 165,925 cuts, up 54%. But AI-cited cuts were only 3,462 in August and 3,961 in September (9.2% of September's total, calculated), and in August AI fell to the fourth most-cited reason, ending five months at the top.
These are announced cuts with the reason the employer gave, not measured job losses. The year-to-date total is driven by a heavy first half. The monthly figures suggest the wave of cuts publicly blamed on AI has slowed, even as tech cuts keep running well above last year.
Don't build a plan or a pitch on "AI is replacing jobs" as a steady trend. The data shows a spike in early 2026 and a slowdown since. Use the dated monthly figure, not the year-to-date total.
Sources
6. The information sector has 120,000 fewer jobs than a year ago.
Evidence tier A — BLS releases opened. Year-on-year changes are our subtraction from BLS levels; August and September are preliminary.
The September jobs report showed payrolls up 29,000 and unemployment at 4.2%, with July and August revised down by a combined 60,000. From September 2025 to September 2026, information sector employment fell by 120,000 to 2,739,000 (down 4.2%) and computer systems design fell by 32,200 (down 1.3%), while professional and business services overall grew by 123,000 (calculated from BLS Table B-1). In August, job openings in professional and business services fell to 1,186,000 from 1,305,000 (JOLTS).
The losses are concentrated in the sectors that build and run software, while the wider white-collar services economy is still adding jobs. That is consistent with tech companies doing more with fewer people, but the BLS data can't say why.
If you sell to tech companies, budget for fewer seats per account and buyers who must show more output per person. Price and package around outcomes or usage where you can, not headcount.
Sources
7. The youngest workers in AI-exposed jobs are the ones not being hired.
Evidence tier A — Stanford paper opened. The authors call the patterns descriptive, not causal.
Stanford's Digital Economy Lab updated its ADP payroll study in August 2026, covering 3.5–5 million workers a month to June 2026. Employment of 22–25-year-olds in the most AI-exposed occupations is about 19% below where it would be had it kept pace with the same age group in less-exposed jobs, up from 15% a year earlier. From November 2022 to June 2026, their employment fell about 11%, while 22–25-year-olds in the least-exposed jobs grew about 10%. Workers aged 35–49 in the most-exposed jobs grew about 10%. The decline is linked to jobs where AI is used to automate tasks, not where it is used to assist.
AI is not removing experienced people; it is closing the entry door. Fewer junior hires today means fewer mid-level people in three to five years, in the same functions.
Look at your own entry-level pipeline in sales development and support. If AI is absorbing the work juniors used to learn on, decide now how your next account executives and managers will be trained, because the old path is narrowing.
Sources
8. Job postings have tilted to senior roles and AI titles.
Evidence tier A — Indeed Hiring Lab and LinkedIn Economic Graph pages opened. Hiring Lab pages did not show a publication date.
Indeed Hiring Lab reports senior job postings up 14.7% year on year and entry-level postings down 7.5% (as of May 2026). Software development postings are up almost 15% since February 2025, while all postings fell 7%, but they remain about 27.5% below February 2020; 71% of the year's increase came from senior roles. LinkedIn's August 2026 update shows US hiring down 6.9% year on year, entry-level hiring down 7.6%, and AI Engineer as the fastest-growing role, up 64%. LinkedIn says US sales hiring weakened noticeably against overall hiring in Q1 2026, without giving a figure.
Companies are still hiring, but for experience and for AI skills. The same pattern as the payroll data: the senior market holds, the junior market thins.
Expect your buyers' teams to be more senior and more technical, with fewer junior evaluators. Shorter, more technical evaluations will win over long education-led cycles.
Sources
9. Not everyone is cutting: Infosys and Alphabet are hiring into AI.
Evidence tier A — Infosys 6-K transcript, Alphabet's Q2 release and Q1 call transcript opened.
On its Q1 FY27 call, Infosys said more than 80,000 delivery staff use AI coding tools, that it has done no staff restructuring, and that it plans to hire 20,000 graduates in FY27, more than 4,000 of them already in Q1. Its CEO said the same work can be done "maybe with fewer people, but there is more work." Alphabet's headcount rose 6.3% to 198,933 at 30 June. On its April call, its CFO said Alphabet expects to keep hiring in AI and Cloud, and management said its engineers now orchestrate autonomous agents.
The companies betting most on AI demand are not shrinking yet. They are moving people toward the work AI creates. That makes the cuts elsewhere a choice about where to grow, not an inevitable effect of the technology.
When a buyer says AI means they need fewer of your seats, ask what new work AI has created in their team. The answer often points to the next thing you can sell.
Sources
10. AI labs' own data: sales outreach is among the fastest uses to automate.
Evidence tier A — AI labs' own research pages opened. These measure exposure and usage, not job losses.
Anthropic's March 2026 Economic Index found that API use for business sales and outreach automation at least doubled in three months. Its labour-market study (March 2026) found no systematic rise in unemployment among highly exposed workers since late 2022, but the rate at which 22–25-year-olds find jobs in exposed occupations fell about 14%, a result the authors call barely significant; computer programmers were the most exposed, with 75% of tasks covered. OpenAI's April 2026 framework classifies about 152 million US jobs: 18% at higher near-term automation risk, 24% likely to be reorganised, 12% that could grow with AI and 46% facing less immediate change.
Usage data shows where companies are pointing AI, a step ahead of the jobs data. Automated outreach doubling in a quarter is the signal most relevant to revenue teams.
If your pipeline depends on outbound volume, assume your buyers' inboxes are filling with automated outreach. The advantage moves to relevance and proof, so lead with a sourced, specific reason to talk.
Sources
Every claim links to its source. Spotted something wrong? It will be fixed and logged on the corrections page. Also published on Beehiiv.