Skip to Content Skip to Menu Skip to Footer

By Pollyanna De Lima, Alexander Johnston, and Sophie Malin

Highlights

AI adoption remains focused on productivity and revenue gains, not explicitly on head count reduction. Among enterprise AI objectives, process efficiency (64%) and employee productivity (59%) are much more commonly prioritized than head count reduction (24%). Job cuts remain a secondary consequence, rather than the primary objective of investment.

AI's employment impact has turned modestly negative, reversing the more positive picture in 2025. In our prior report, Generative AI and the workforce: More redistribution than reduction, AI's employment effect was characterized as neutral to slightly positive. In contrast, the latest S&P Global Purchasing Managers' Index survey shows a global net impact of -5 percentage points over the past 12 months (percentage of businesses increasing workforce due to AI adoption minus percentage decreasing), with a further -2 points net impact forecast for the coming year.

AI adoption is broad, but most deployments remain augmentative rather than autonomous. Across 38 AI use cases, the average current adoption rate is 50%, and the average planned adoption rate in the next year is 37%. Widely adopted use cases include summarization (71%), translation (62%) and data management (61%). However, only 22% of AI projects target a fully autonomous end state, highlighting the continued need for human oversight and aligning with the modest projected pace of AI-driven workforce displacement.

Automation pressure is real, but constraints are likely to limit progress. AI’s capacity to replace labor is constrained by delivery challenges related to trust, data and skills. The technology may be advancing faster than organizations can reliably deploy it: Only 46% of AI initiatives launched in the past year are deemed on track to achieve positive ROI within 12 months, and only 37% are assessed as live and delivering value.

Building on our previous analysis in Generative AI and the workforce: More redistribution than reduction, this report provides an updated view of AI and labor market dynamics from the latest Purchasing Managers' IndexTM (PMI®) special survey on AI investment, along with recent data from 451 Research’s Voice of the Enterprise survey lines.

Businesses worldwide continue to increase AI adoption and planned capital allocation across all tracked geographies and sectors. However, a notable shift has occurred in the AI and labor landscape. While our prior report indicated a neutral to slightly positive net employment impact from AI adoption, the latest findings show a negative global net impact for the past year (-5 points) and a marginal decline forecast for 2026 (-2 points).

We calculate these net impact figures by deducting the percentage of survey respondents reporting lower employee counts due to AI adoption from the percentage reporting higher employee counts. The change reflects ongoing task reallocation, as declines in more "automatable" roles are only partially offset by growth in AI-related functions.

Employment consequences vary across company sizes and sectors, with larger firms increasingly anticipating net negative impacts, while small and medium-sized enterprises still forecast positive effects. Targets of automation include streamlining manual processes for small firms and optimizing recruitment and customer acquisition for larger companies.

However, persistent challenges remain, including concerns about data privacy, accuracy of AI outputs and employee resistance. These dynamics are central to understanding AI's impact on jobs in 2026 and the broader trajectory for AI and the future of work.

Negative employment impact reported in the past 12 months

The PMI special survey indicates firms’ primary objective in adopting AI is to achieve AI-enabled productivity gains, rather than deliberate workforce reduction. However, on a global basis, the net past-12-months employment effect appears negative, with the balance of private-sector firms reporting job losses coming in 5 percentage points higher than the share reporting job gains.

This reflects ongoing shifts in AI adoption and the labor market, as reductions in roles such as administration and office support, translation, manufacturing, production and customer service were partially offset by new positions to manage AI initiatives, including technical, software and digital functions.

A pain point on our data center operations team is getting enough staff. … Everybody agrees that we're understaffed.

— United States, healthcare, 10,000-49,999 employees.

Source: 451 Research’s Voice of the Enterprise in-depth interview, February 2025.

Individual countries reported varying AI impact on employment:

  • In Germany, the influence on total employment was mild, with a net balance of -2 points, indicating a slight tilt toward workforce reductions linked to AI. New jobs included AI developers and implementation managers, while administrative and marketing roles were cut.
  • In Italy, the net balance was +9 points, reflecting more firms adding jobs. New roles included marketing, graphic design, analyst and IT jobs.
  • Japan showed a broadly neutral impact, as growth in data and machine learning specialist roles offset reductions in administrative, call center, translation and website content creator positions.
  • UK businesses reported a net negative workforce impact (-6 points), with cuts in administration, customer service and finance roles.
  • In emerging markets, Brazil registered a positive net impact on employment, respondents in Mainland China reflected more companies reducing staffing than adding, and India recorded a neutral impact.

Globally, goods producers and service providers reported productivity gains and revenue growth as the main benefits of AI investment. The net impact on employment was slightly more pronounced among manufacturing firms (see dashboard figure above).

Among large enterprises, the trend shifted toward job losses from AI over the past year: In the latest survey, the proportion reporting AI-related job reductions was 8 points higher than the share reporting gains. However, overall head count reduction appears to have been far more widespread.

Of the S&P Global 1200 index (a composite of seven headline indices), 994 index participants (83%) had a lower head count in January 2026 compared to January 2025, with just 153 (13%) experiencing an increase. This implies that recent head count reductions cannot be attributed solely to AI, and in many cases, other factors were likely equally or more important, underscoring the complexity of AI's impact on jobs in 2026.

A lot of companies are now utilizing AI where they would have recruited a young person to come into the company. … What we'll see in a few years' time is quite a big employment gap where we'll lose those skills because enterprise businesses have gone down the AI route. … We're going to skip that start point for people.

— United States, education and training, 2-9 employees.

Source: 451 Research’s Voice of the Enterprise in-depth interview, January 2026.

Modestly negative employment impact forecast for the next 12 months

Global respondents to the PMI survey forecast a marginal decline in net employment due to AI investment in the year ahead, with the net balance at -2 points. This indicates a relatively stable employment outlook amid ongoing AI integration.

AI investment is reshaping employment across firms of all sizes — affecting the number of new hires, the roles being hired for and how employees are retrained. But while companies across all size bands report that AI is improving productivity, the employment consequences are more uneven.

Small firms continue to forecast a net positive employment effect from AI investment into 2026 (+3 points). AI is often used to extend capacity, allowing small teams to take on more work so hiring can focus on growth needs. Workforce change tends to mean incremental job creation tied to growth — hiring generalists who can use AI, rather than dedicated AI specialists. Reskilling and training in small firms is typically pragmatic and internal: AI is embedded in day-to-day processes and learned on the job, sometimes supplemented by lightweight internal courses.

Medium-sized firms also forecast a positive employment effect from AI in 2026, with a net balance of +2 percentage points, whereas large companies forecast a net negative employment impact of -13 points (see dashboard figure above).

Larger organizations appear better positioned to strategically operationalize AI, in part because they are more likely to have formal AI structures. In 451 Research's Voice of the Enterprise: AI & Machine Learning, Use Cases 2026 survey, respondent organizations with 10,000+ employees were significantly more likely to have a formal, internal AI discipline, with 44% citing a "clear, documented AI strategy" aligned with core business goals, including dedicated roles and career paths for specialized AI professionals.

There is a direct relationship between such a discipline and the success rates of AI projects, their return on investment, and the recognition of AI as an integrated, organization-wide capability. Formal discipline may give these companies a clearer view of where AI can be applied, and a stronger ability to align those insights with operating models and workforce planning.

The presence of structured career paths for specialized AI roles at larger enterprises also indicates that large firms are the most active in creating new AI-related roles, even if these roles do not fully offset reductions elsewhere. Their workforce strategy commonly includes dedicated AI teams that deliver projects and support enterprise-wide adoption.

We are creating a dedicated team within the purchasing department that will handle all AI-related topics, support its implementation, and train and prepare employees for the changes

— Germany, electrical manufacturer, 250+ employees.

Source: S&P Global Purchasing Managers' Index special survey on AI investment.

Administrative tasks are commonly automated, but autonomy remains limited

Organizations are already applying AI across diverse tasks, reflecting the breadth of adoption and suggesting the extent of planned expansion. Across 38 AI use cases surveyed in the 451 Research study cited above, the average current adoption rate is 50%, and the average planned adoption rate in the next year is 37%. Uses such as summarization (71%), translation (62%) and data management (61%) are among the most widely implemented.

The marketing department uses AI for internal and external communication. It is used for debriefings and transcripts of video meetings. We have adapted it to identify and avoid discrepancies between purchase orders and invoices, on new contracts that still require some editing. It is used to identify billing discrepancies, for example overly generous commercial gestures.

— France, renting & business activities, 1-19 employees.

Source: S&P Global Purchasing Managers' Index special survey on AI investment.

Importantly, the breadth of investment does not directly translate to widespread AI-driven workforce displacement. Less than a quarter (22%) of AI projects target a fully autonomous end state, where AI operates without human intervention. Use cases viewed as highly automatable include financial transactions, financial planning and analysis, and asset valuation, but these are less widely adopted compared to AI-augmented, human-led tasks such as summarization, meeting productivity and content creation.

 

Risk and security use cases stand out both for the level of investment and for their higher assessed automatability. Just over half of survey respondents (51%) reported investing in AI for identity verification and access assurance, with 29% targeting full automation, and a further 28% targeting "predominant automation with human oversight." Fraud detection and audit and regulatory reporting show a similar combination of broad adoption and high automation potential. This may place pressure on review-heavy roles in risk, fraud, identity and compliance operations.

AI-driven productivity gains may indirectly fuel job loss, accelerated by agentic autonomy

All tracked geographies and sectors reflect notable increases in AI investment and planned capital allocation compared to the 2024 PMI survey, indicating a growing commitment to AI.

Organizational AI investment is driven primarily by efficiency and productivity objectives, which can support greater automation across sectors. Among respondents to 451 Research’s Voice of the Enterprise: AI & Machine Learning, Use Cases 2026, process efficiency (64%) and employee productivity (59%) were the top-cited objectives in AI initiatives. Head count reduction was less commonly cited (24%), but AI-enabled productivity gains can still contribute to lower labor requirements.

 

 

Investment in AI is broad-based, spanning various AI classes (see figure below). Notably, organizations expect to mature rapidly in their investments in AI agents, suggesting a growing role for agentic AI in reshaping the workforce. Agents are associated with high automatability — they are designed to plan, act and complete multistep workflows with significant autonomy. This may reduce the amount of human intervention required in AI-enabled workflows, further accelerating AI-driven workforce displacement.

An AI agent is your model employee: Show up every day and work consistently. And so, the risk is over-relying on them. But when they are doing their thing correctly and applying a certain degree of intelligence to the decisions and the outputs that it's producing, then they could be very valuable, especially when you string agents together to complete more of an end-to-end process.

— United States, Financial Services, 250-499 employees.

Source: 451 Research’s Voice of the Enterprise in-depth interview, January 2026.

Skills gaps, trust and ROI concerns could limit automation

While AI adoption is widespread, several factors may limit automation and, by extension, the pace of AI-driven workforce displacement. For example, worker concerns may lead to caution in implementing widespread automation, especially in politically sensitive areas and highly regulated sectors. With firms forecasting significant productivity boosts and future hiring pullbacks, governments and businesses may face pressure from organized labor groups to implement stronger worker protections and transition measures.

Although AI investment is increasing across all company sizes, large companies continue to lead the way due to their greater resources, ability to scale projects quickly and specialized teams. This suggests that smaller firms face limitations in funding and implementing extensive automation projects, which accordingly limits head count reduction by these businesses.

Further, a mixed picture regarding returns may limit future investments. In 451 Research's Voice of the Enterprise: AI & Machine Learning, Use Cases 2026 survey, respondents estimate that, on average, just 46% of AI initiatives launched in the past year are on track to achieve ROI within 12 months. This figure is even lower in France (43%) and Germany (38%). In addition, just 37% of AI initiatives over the past 12 months were classified as live and delivering value, with many projects stuck in development or partial deployment.

We cannot use AI for specific projects, which are usually subject to confidentiality, as the information requested from ChatGPT, for example, becomes the property of ChatGPT, and the same applies to the drafting of manuals, which would obviously contain confidential information. We cannot yet use it for software development, as it would not give us the right answers; it is not yet capable of doing so. In essence, we have not yet found an application that fits well with our reality to help us simplify or improve our work.

— Italy, electrical manufacturer, 1-19 employees.

Source: S&P Global Purchasing Managers' Index special survey on AI investment.

AI initiatives continue to face various constraints, helping to explain the mixed returns. Respondents most frequently cite concerns about data privacy and security (51%) as limitations of generative AI models, followed by response accuracy and quality (46%) and data quality (38%).

These issues are often more acute in specialist domains such as energy and manufacturing, where large language models and agents can struggle to interpret domain-specific operational context, particularly when the use case relies on data types that are underrepresented in the training data of foundation models. This suggests AI’s automation opportunity may extend less readily to specialized operational use cases than to more general enterprise tasks.

Organizations also face numerous skills gaps that may need to be addressed through hiring or reallocating staff for AI-related roles. Technical functions are particularly prominent. Cybersecurity is the most acute shortage, with 64% of respondents saying gaps had a moderate or severe impact on AI initiatives, followed by machine learning and AI development (59%), software development and engineering (58%), and data management and governance (57%).

 

According to 451 Research’s Voice of the Enterprise: AI & Machine Learning, Use Cases 2026 survey, confidence in outputs from both third-party and in-house AI models has fallen markedly since 2023. In 2026, just 16% say they completely trust third-party AI models, while 30% mostly trust them, down from 24% and 42%, respectively, in 2023. Autonomous deployment will remain impractical unless organizations confine AI to lower-risk use cases or materially improve the reliability of model outputs

Looking forward

While AI investment continues to accelerate, with increasing focus on autonomous agents and the emerging agentic AI workforce, the recent softening in employment should not be interpreted as evidence of large-scale AI-driven workforce displacement. In the near term, practical constraints continue to limit AI's impact on employment. Most deployments still require human oversight, and concerns about accuracy, reliability and security constrain full automation.

As a result, the dominant pattern remains one of task reallocation, with shifts in labor demand rather than outright workforce reduction. That said, over a longer horizon, sustained AI-enabled productivity gains will no doubt translate into reduced reliance on labor resources, potentially creating a gradual but persistent downward pressure on head count, even if workforce reduction is not an explicit objective for most firms.

Contributors: Matt Tompkins and Carla Donaghey

Content Type

Special Reports