America’s biggest companies say they might need more people after all. For months, major employers treated hiring as an expensive last resort. Now a shift is emerging across industries, and it is running directly against two years of confident forecasting about what artificial intelligence would do to headcount.
Who is saying it
Companies ranging from railroad giant CSX to Google parent Alphabet (GOOGL) have told investors in recent days that they plan to hire to meet growth goals or to seize on new technologies. Major employers had largely held back on adding people due to economic uncertainties or a belief that AI could shoulder more tasks on the job. Some executives now say the costs and limitations of AI demand that more people be added; others simply want to hire.
The government contractor Booz Allen Hamilton (BAH) is the cleanest example of the reversal, because it cut thousands of jobs last year as the administration slashed federal contracts and asked firms to justify their costs. “The push to expand head count, at least modestly, is a reversal,” the reporting notes, and Chief Operating Officer Kristine Martin Anderson told investors on Friday: “We’re a little bit behind right now. We’re addressing that now.” On the same call the company said demand is picking up across several parts of its business and that cost-cutting is bearing fruit; it now sees healthy demand for its services, including in national security for workers who need security clearances. Shares jumped 10% to $72.53 on Friday, though the stock has lost about a third of its value over the past year.
Elsewhere: Alphabet Chief Financial Officer Anat Ashkenazi said the company expects to continue hiring in key investment areas such as AI and cloud computing. The software company ServiceNow wants to hire more “quota-bearing feet-on-the-street sales execs” to capture growth in areas like cybersecurity. Tool maker Snap-on said it plans to add employees to expand its business. CSX said its train and engine service head count will “increase modestly” in the coming months, though its head count is still lower than a year ago.
The recalibration nobody advertised
For much of the past 18 months, big employers were convinced that fewer workers meant faster growth, and U.S. public companies shrank their white-collar workforces. Now layoffs are shrinking. The most recent week of U.S. jobless claims was the lowest on record since 1969, according to federal data.
Part of the change reflects a recalibration among employers about what AI can actually do, said Sarah Franklin, CEO of the human-resources platform Lattice. Many companies stopped hiring entry-level employees, thinking AI agents could pick up the slack, and have since realized that humans are the ones who bring judgment to the work. “There’s a big thirst for that,” Franklin said. “What you have now is a realization that you need the AI-native skills. You need this entry-level workforce, which is innovative, not calcified in thought. They’re also more affordable because they are newer to the workforce.” Across Lattice’s thousands of clients, many are now back in hiring mode across a number of roles, particularly for junior positions. Building AI sales agents, she added, does not mean you stop hiring engineers — companies with AI sales agents also need salespeople.
M. Keith Waddell, CEO of staffing firm Robert Half, said AI’s impacts on the job market are proving to be “more benign than some have feared,” and that clients are recruiting again. “Hiring demand continues to improve and market conditions are increasingly more supportive of our business,” he said.
The honest caveat
Few companies are announcing major hiring sprees, and the same reporting carries the counterweight. “Do we need more people? Do we need less people?” asked Paul Osterman, a professor emeritus at MIT and the author of a new book on the transformation of employment. “We have no idea. No one has any idea.” Osterman noted that many companies have treated employees as dispensable — cutting them when convenient, or downgrading them into contractor or part-time roles — and expects that trend to continue in this moment of flux. “AI introduces so much uncertainty with employers, not sure knowing what they need or don’t need,” he said.
Hiring plans are announced in press releases and rescinded in footnotes. We will believe the payroll data over the intentions.
Why this is a portfolio input and not a trade
For two years the bear case on the American consumer had a clean sequence: AI layoffs arrive, spending falls, credit deteriorates, the cycle turns. The order appears reversed. The spending arrived first — the data centers, the power contracts, the capex — and the spending needs people to build and operate it. Employed people pay their bills.
That is quietly supportive of exactly the unglamorous sleeves we lean on when we go defensive: staples, discount retail, and the deposit bases of regional banks. It is also the reason the recession-portfolio triggers we set going into the year remain unmet. This is not a signal to add risk. It is a signal that the specific bear case most widely held in 2025 has not happened on schedule, and that a defensive book does not need to become a fearful one.
For two years the bear case on the American consumer was that AI layoffs would arrive before AI productivity. The order appears reversed: the spending arrived first, and the spending needs people. That is quietly bullish for exactly the unglamorous sleeves we lean defensive with — staples, discount retail, regional banks’ deposit bases — because employed people pay their bills. Action: no trade; this is a thesis input. It supports the consumer-discretionary tilt we flagged going into Q1 and keeps recession-portfolio triggers unmet. The honest caveat: hiring plans are announced in press releases and rescinded in footnotes. We will believe the payroll data over the intentions.
Booz Allen (BAH) · Union Pacific (UNP) · Dollar General (DG)
Tickers are named to identify the businesses discussed. They are not recommendations, and a name appearing here does not mean it is held in any model.