Oracle's 10-K Said What the Press Releases Didn't.
Every Friday I look at what moved in the analytics craft. Usually it’s a job posting, a comp survey, or a press release. This week it was an annual report.
Oracle filed its fiscal 2026 10-K Monday. In the section where companies disclose material business risks — the part that carries legal weight — the company wrote that “the adoption and deployment of AI technologies across our operations have resulted, and may continue to result, in reductions to our workforce.” That is not a press release. That is not a CEO town hall transcript. It is the document where the company tells the SEC what is actually happening.
Twenty-one thousand workers cut, 13% of headcount, in the past 12 months. Net income up 27%. Capital expenditure up 162% to $55.7B, directed at AI data centers. Salesforce reported cuts this week too, including data analytics roles.
Most companies have spent 18 months signaling this in earnings calls. Oracle put it in a filing. The question it leaves — which roles survive this math, and why — is the one most senior analytics ICs haven’t answered for themselves yet.
What’s actually moving in the market
Oracle’s fiscal 2026 10-K, filed June 22, is the first legally-required disclosure at this scale to name AI as a direct cause of workforce reduction. Twenty-one thousand workers cut as the company posted $3.7B in quarterly net income and $55.7B in capital expenditure — a 162% increase, directed at AI data centers. The framing matters: SEC filings don’t have a PR edit. “Adoption and deployment of AI technologies…have resulted, and may continue to result, in reductions to our workforce” is the language a company uses when it is legally required to be accurate about material risks.
Salesforce also reported layoffs this week, including data analytics and its Agentforce division. Fewer than 1,000 cuts, following the same structural pattern: orgs investing most aggressively in AI infrastructure are shedding mid-level analytics roles that don’t connect to AI output governance. The mid-level seat is clearing. The governance seat is staying open.
The Fivetran/dbt Labs merger closed June 1, and the first joint products shipped with it. dbt Core v2.0 open-sources the Fusion engine under Apache 2.0; dbt State selectively processes only changed data assets, targeting 30%+ compute cost reduction. Fivetran CEO George Fraser and dbt Labs President Tristan Handy now lead a combined company serving 100,000+ data teams. The pre-close negotiating window is gone. The next leverage point is before the combined entity’s first pricing review cycle — typically 6–12 months post-close. The merger closed. The pricing conversation didn’t.
PwC’s 2026 Global AI Jobs Barometer, released June 15, puts the wage premium for AI-skilled workers at 62% — up from 57% a year prior. The analysis covered more than one billion job advertisements across 27 countries. Ravio’s analytics-specific figure from two weeks ago: 12% for AI/ML IC comp bands. Both are real. The 62% is what the economy prices across all AI-adjacent roles; the 12% is what analytics IC bands have actually received. The gap between those numbers is the business case for moving your role description toward the right side before the next comp cycle closes.
Staff and Principal Analytics Engineer comp at major tech: $190K–$260K base; median market, $158K–$165K (Levels.fyi, Glassdoor, June 2026). The specific JD language bridging those bands in the June 2026 posting cycle: “AI output validation,” “transformation governance,” “semantic layer ownership,” “cost governance.” These phrases appear in job descriptions before they appear in comp bands — typically with a 6–8 week lag.
What I’d do this week
Write the Oracle sentence for your own role. The 10-K’s structural logic: fewer people, more infrastructure, stronger margins. That logic has a question embedded in it. Most senior analytics ICs can sense which side they’re on. Fewer have written the answer down.
- The user moment: Before your next 1:1 with your VP or skip-level. Not to discuss it — to know the answer before the conversation happens without you.
- The shape: Role audit — one sentence, written down, not shared.
- The time budget: 20 minutes, alone.
- The artifact: One sentence: the deliverable my current role produces that would require a human judgment before a VP would act on the AI version of it. If you can write it in two minutes, you're in a stable position. If you can't write it at all, that is the finding. You still have time to move.
- What success looks like: By next Friday you know which side of Oracle's 21,000 your current role would have been on. If the answer is the wrong side, the sentence tells you exactly what deliverable to move toward. One sentence is the whole exercise.
Pull the comp language before the comp conversation. Staff and Principal AE JDs from the June 2026 posting cycle are adding specific phrases as required competencies: “AI output validation,” “transformation governance,” “semantic layer ownership,” “cost governance.” These phrases appear in job descriptions before they appear in comp bands — with a 6–8 week lag. If none of them are in your current JD, you’re being priced as a non-AI role regardless of what you actually do day to day.
- The user moment: Friday afternoon. Three current Staff/Principal AE postings, your current JD, and [Levels.fyi](http://Levels.fyi) open at the same time.
- The shape: Comp audit — a gap list between what June 2026 JDs require and what your role description says.
- The time budget: 45 minutes.
- The artifact: A gap list: the specific phrases in current Staff/Principal AE JDs that describe work you already do but that your role description doesn't name. That list is the factual input for a title review or comp conversation — not a grievance, an update.
- What success looks like: A month from now you've either updated your JD language or raised it explicitly. The comp band follows the title and the JD, not the work you do in silence. The update takes 20 minutes. The failure to make it costs the differential for as long as the silence holds.
Sources
- Oracle sheds 21,000 roles over the past year amid wave of AI layoffs from tech giants — CNBC, June 23, 2026
- Oracle cut 21,000 jobs as AI buildout continues — Yahoo Finance, June 2026
- The running list: major tech layoffs in 2026 where employers cited AI — TechCrunch, June 22, 2026
- Fivetran + dbt Labs Complete Merger to Create the Data Infrastructure for Trusted AI Agents — Fivetran, June 1, 2026
- PwC 2026 Global AI Jobs Barometer — PwC, June 15, 2026
- The AI compensation and talent trends shaping the job market in 2026 — Ravio, 2026
- Analytics Engineer Salary in 2026: Real Data from 1.9 Million Job Postings — Recruiting from Scratch, 2026
Crafting runs weekly. The job is to be useful, not to sell. If a line in here is close to a move you're working on, the email reaches a person.
Hiring rather than hunting? Signal is the column for your side of the desk.
← All Crafting issues · Drafted with Claude · Edited by Paul Brown