Why entry-level white-collar collapsed faster than the models predicted.

Three forces hit at once. Each had a separate forecast. Their interaction didn't — and the interaction is what broke the bottom of the org chart.

Every macro model that mattered in 2023 was modeled in isolation. AI capex impact. Reshoring impact. Tariff impact. Each one had a "manageable" forecast. None of them modeled the others. The interaction is what ate the bottom of the org chart in 2025–2026.

The result: a 47-percentage-point divergence between what entry-level analysts make today and what AI-augmented ICs make, indexed to January 2023. The career counselor in your university's basement is still pointing at the role that lost the race.

Below: the three forces, the mechanism each one uses on hiring budgets, and the move that's right at each career stage. The forces aren't reversing. The playbook from 2010–2022 was a good playbook for a different regime. Plan accordingly.

◆ Force 1 · AI capex

The trillion-dollar capex cycle that ate hiring budgets.

The big-tech AI capex line went from ~$120B/year in 2022 to north of $400B/year. That money came out of the same operating-income pool that funded headcount growth. It's not a coincidence that hyperscalers cut hiring while announcing $50B+ infra spends — it's the trade-off, line by line.

The downstream mechanism: the model is now embedded in product, which means a team that used to need 6 ICs needs 4. The CFO sees the headcount line drop without a productivity hit and doesn't backfill. The first slot deleted is always the most junior one.

What the consensus model missed: capex and headcount aren't independent budget lines at the largest employers — they're a substitution. More compute, fewer seats.

◆ Force 2 · Reshoring

Hiring moved cities. Most graduates didn't.

The CHIPS Act, IRA-driven battery plants, and broader reshoring shifted hundreds of billions of capex into manufacturing hiring — across Arizona, Ohio, Georgia, Texas. The new jobs are real. They're skilled trades, mechatronics, plant operations, supply-chain analysts.

The friction: the marginal college grad is in a coastal metro, looking for a white-collar coordinator job that no longer exists in that volume. The mismatch isn't being modeled because it crosses two BLS categories that economists analyze separately — services-sector employment and goods-producing employment.

What the consensus model missed: "reshoring is good for jobs" is true in aggregate. It's net-negative for the specific job a 22-year-old polysci grad in Brooklyn wants.

◆ Force 3 · Tariffs & SaaS pricing

Tariffs raised input costs. Startups absorbed them by not hiring.

Tariffs raised the cost of physical inputs across the economy. For SaaS companies whose customers are SMBs in retail, logistics, and manufacturing, that meant the customer base became cost-sensitive overnight. Net retention compressed. ARR growth slowed at the median Series B.

The downstream effect on hiring: Series A–C startups have been the biggest absorber of new grads for fifteen years. With slower revenue growth and a higher cost of capital, they're hiring 30–50% fewer entry-level seats per round of funding than they were in 2021.

What the consensus model missed: tariffs were modeled as a price story. They turned into a startup-hiring story — because SMB-facing SaaS got squeezed in the middle.

◆ The interaction

Three forces, one collapsed seat.

AI capex took the seat at hyperscalers. Reshoring moved the budget to cities where the marginal grad doesn't live. Tariffs took the seat at the Series A–C startups that absorbed the rest. None of the three forces, on its own, would have produced a 14% YoY drop in entry-level openings.

Together, they did. And the forecast horizon for any of them reversing is north of 24 months.

◆ By career stage

What this means for you, by where you are.

New grad / first job

Stop hunting for the analyst program.

It's smaller than it used to be and the seat is the most exposed slot at every firm. Apply, but treat it as a backup. The primary move: get fluent enough with AI workflows that you can apply to "AI-augmented IC" jobs at Series B–D companies that don't yet have an analyst pipeline.

Ship one public artifact a month. Bias toward smaller, less brand-name companies — they pay less in the first year but the second-year compounding is 2–3× better.

Mid-career (3–8 yrs)

The most exposed cohort.

You're senior enough to be expensive, junior enough that the role is being collapsed. The move: pick a side. Go down a level into a specialist IC role with AI leverage (and a sharp raise within 18 months), or go up into a manager seat where the org's risk is in execution, not headcount.

The worst spot is "mid-IC at a mature company without an AI angle." That's the seat being deleted on every Q4 plan.

Senior (8+ yrs)

The fractional door just opened.

Companies that used to hire a full-time VP now want 30% of one. The fractional operator route is real — $4–8k/month per client, 3–4 clients, you're at $200k+ on a 25-hour week.

The trap at this level: holding onto a comfortable Director seat at a flat-growth company while your brand quietly depreciates. The longer you stay, the more the next move looks like a step down.

◆ The new playbook

Ship in public. Ship with AI. Ship work that compounds.

The 2010–2022 playbook (degree → analyst program → MBA → director) compounded credentials. Credentials work when the org chart needs more middle. The new org chart needs fewer middles and more operators.

The new playbook compounds visible output: public artifacts, AI workflows that you own, a body of work a hiring manager can verify in 90 seconds. That's why a 26-year-old with a substack and three shipped projects now outearns a 26-year-old with a top-15 MBA — the artifact pipeline is more legible than the pedigree.

None of this is permanent. It's just the regime for the next 5–10 years. Plan for the regime you're in, not the one your parents prepared for.

◆ What to do this Sunday

Write one paragraph about your next 12 months.

One paragraph, in your own words, that names the regime and your move inside it. If the paragraph could've been written in 2018, it's wrong. Read it on Monday morning before you open your inbox.

Sources & notes

AI capex figures are aggregated from Big Tech reported capex (Alphabet, Meta, Microsoft, Amazon) and SemiAnalysis estimates. Reshoring activity drawn from CHIPS / IRA award trackers and BLS manufacturing-employment series. Tariff impact on SMB-facing SaaS is editorial inference from ARR-growth disclosures and net-retention prints at public companies; the through-line is directional, not modeled. None of this is investment or career advice — it's a working theory, presented so you can reason against it.

← Back to Issue 001

Liking this? Subscribe to get next Sunday's issue.