Goldman Sachs: The slowdown in tech hiring is mainly not a rate shock; post-pandemic workforce normalization matters more than AI

Overview

This report discusses the reasons behind the significant weakening in technology-sector hiring since 2022. Goldman Sachs breaks the potential headwinds into three categories: the Fed's hawkish pivot and rising interest rates, AI-driven efficiency gains, and workforce normalization after overhiring during the pandemic. Based on company- and occupation-level employment data from Revelio Labs, the report analyzes workforce changes from 2019 to 2025 across more than 300 U.S.-listed technology companies and more than 800 occupations.

Core views

The core conclusions are: higher interest rates are not the main reason for the slowdown in tech hiring; AI has indeed weighed on hiring, but the impact remains relatively small for now; and post-pandemic workforce normalization after overhiring has stronger explanatory power for the hiring slowdown, at roughly 3-4 times the effect of AI. The report estimates that identifiable channels such as AI and workforce normalization together explain only about half of the roughly 5-percentage-point shortfall in annualized employment growth in the technology sector versus its long-term trend, with the remainder more likely attributable to aggregate shocks faced jointly by all tech companies and workers.

Analysis framework

The report first observes changes in hiring at the company and company-occupation levels in 2019-2022 versus 2022-2025, and then separately constructs three tests: using changes in interest coverage to measure companies' exposure to rising rates; using the share of occupational tasks exposed to AI automation and AI displacement scores to measure occupation-level AI exposure; and identifying pandemic-period overhiring through accelerated hiring in 2020-2022 combined with declining revenue per employee. The regressions include company and occupation fixed effects to strip out, as much as possible, common company-level and occupation-level factors.

Methodology notes

• Macro labor market attributionCompany-occupation panel regression

  Identify the relative importance of different hiring headwinds through cross-sectional differences at the company and occupation levels

  The report uses employment data from 2019-2025 for more than 300 listed technology companies and more than 800 occupations, comparing changes in employment growth under different exposures to rates, AI, and overhiring.

• Interest-rate sensitivity analysisInterest coverage grouping

  Measure the extent to which companies are affected by rate hikes using changes in interest coverage

  If rates were the main drag, companies with higher rate exposure should have slowed hiring more significantly; however, the report finds that workforce changes are nearly identical across groups.

• Measurement of AI's employment impactOccupational AI exposure and AI displacement score

  Measure the extent to which tasks in different occupations can be automated or replaced by AI

  The report regresses annual occupational employment growth on occupational AI exposure and interaction terms with time, finding that in 2025, employment growth in occupations with higher AI exposure is subject to a slight negative impact.

• Post-pandemic normalization analysisOverhiring proxy variable

  Identify companies that may have overexpanded using accelerated hiring in 2020-2022 combined with declining revenue per employee

  The report argues that these companies are more likely to experience workforce normalization after 2022, and the relevant estimates show that their subsequent occupation-level employment growth is significantly lower than that of other companies.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• U.S. technology sector

  Subject of the research

  Strengths

  Still has the potential for AI-driven efficiency gains and long-term productivity improvement.

  Weaknesses

  Since 2022, employment growth has been significantly below the long-term trend, and the momentum of workforce expansion has weakened.

  Comparison

  The impact of post-pandemic normalization after overhiring is greater than that of AI-driven efficiency gains, while differences in rate exposure have the weakest explanatory power.

  Risks

  If aggregate demand or growth expectations continue to weaken, common shocks not explained by the model may continue to weigh on hiring.

• AI-related occupations

  Affected occupational group

  Strengths

  AI-related technologies are boosting productivity and explain part of corporate layoffs and job adjustments.

  Weaknesses

  Employment growth in occupations with high AI exposure has faced a slight drag since 2023.

  Comparison

  AI's impact is smaller than post-pandemic workforce normalization, but its importance has increased recently.

  Risks

  If the pace of AI deployment accelerates, employment pressure on highly exposed occupations may increase.

• Technology companies with high interest-rate exposure

  Tested channel of macro shock

  Strengths

  The report does not find their hiring performance to be significantly weaker than that of companies with low rate exposure.

  Weaknesses

  Higher rates may still indirectly affect the sector through valuations, financing costs, and growth expectations.

  Comparison

  Relative to AI and overhiring, rate exposure has the least cross-sectional explanatory power for changes in hiring.

  Risks

  If financial conditions tighten significantly again, the rate channel could become important once more.

Key data

• Tech-sector employment growth below trendabout 5 percentage points/yearSince 2022, employment growth across multiple tech subsectors has been significantly weaker than the long-term trend.
• AI's explanation for the slowdown in employment growthup to about 2 percentage pointsDifferences in occupational AI exposure explain part of the slowdown in annualized employment growth in the technology sector, but the effect exists without being dominant.
• Explanation from post-pandemic workforce normalizationup to about 2 percentage pointsCompanies that accelerated hiring in 2020-2022 and showed weaker productivity performance subsequently had poorer employment growth in 2023-2025.
• Relative importance of overhiring versus AIabout 3-4 timesThe report estimates that the contribution of workforce normalization to the tech hiring slowdown is about 3-4 times that of AI-driven efficiency gains.
• Employment growth gap for overhiring companiesabout 7 percentage points lowerSince 2022, total headcount growth at overhiring companies has been about 10%, roughly 7 percentage points lower than at companies that did not overhire.
• Share of AI-related layoffsrose above 20% in the past yearThe report says that AI-attributed layoffs were previously limited in both the overall economy and the technology sector, and only in the past year has the share of AI-attributed layoffs risen noticeably.

Impact & implications

For investment and macro judgments, the report weakens the explanation that higher rates alone caused the slowdown in tech hiring, and it also reminds the market not to simply attribute workforce contraction in the technology sector to AI replacement. The more important explanation at present is the rebalancing of workforce and cost structures after pandemic-era expansion. AI's impact on labor demand has already emerged and become more evident recently, but as of the report period it is still not the dominant factor.

Risks

• The cross-sectional approach at the company and occupation levels explains only about half of the hiring slowdown, and the remainder may come from aggregate shocks that are difficult to identify.
• There may be measurement error in the AI exposure and overhiring proxy variables, which could understate or overstate the true effects.
• The research focuses on U.S.-listed technology companies, so the conclusions may not fully apply to private companies, non-U.S. markets, or other industries.
• There may be wording bias in corporate layoff announcements; although the report believes AI-washing is not widespread, motives may still differ at the individual company level.

What to watch

• Whether the share of AI-attributed layoffs in the technology sector continues to rise.
• Whether employment growth in occupations with high AI exposure shifts from a slight drag to a broader contraction.
• Whether companies that expanded rapidly during the pandemic and faced pressure on revenue per employee continue workforce normalization.
• Whether the technology sector's share of total employment returns to its long-term trend.
• Whether common aggregate shocks, such as growth expectations, valuation adjustments, and the capex cycle, continue to affect all technology companies.