Eighty-Three Per Cent Of The World's Largest Companies Shrank Last Year

Fewer than half of their AI initiatives are on track for a return, and the research is careful not to say the two are connected.

Sign up for my Substack daily AI newsletter here.

See my AI Training course portfolio for corporate Business Leaders here.

Article link: https://johanosteyn.substack.com/p/eighty-three-per-cent-of-the-worlds?r=73gqa&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Follow me on LinkedIn: https://www.linkedin.com/in/johanosteyn/

The most interesting thing in the latest S&P Global research is a caution rather than a finding. Employment across the world’s largest listed companies is falling. Returns on artificial intelligence are not arriving. The analysts decline to join those two facts together, and the reason they decline is the part every board should read.

CONTEXT AND BACKGROUND

Of the S&P Global 1200 index, 994 constituents, or 83 per cent, had lower headcount in January 2026 than a year earlier, with just 153 recording an increase. Over the same period the net employment impact of AI adoption turned negative globally at minus five percentage points, having previously been neutral to slightly positive, with a further minus two forecast for the year ahead and large companies expecting minus thirteen. Against that, only 46 per cent of AI initiatives launched in the past year are on track to achieve positive return within twelve months, and only 37 per cent are assessed as live and delivering value, with many stuck in development or partial deployment.

INSIGHT AND ANALYSIS

The report’s own reading of the headcount figure is the useful part. It states that reductions on that scale cannot be attributed solely to AI, and that other factors were likely equally or more important in many cases. Its survey evidence supports the caution, since headcount reduction was cited as an objective by only 24 per cent of respondents, well behind process efficiency at 64 per cent and employee productivity at 59 per cent, and fewer than a quarter of AI projects target a fully autonomous end state. Trust is falling rather than rising. Sixteen per cent completely trusted third-party models in 2026 against 24 per cent in 2023, and 30 per cent mostly trusted them against 42 per cent.

Confidence in the technology is declining among the people deploying it, at the same moment investment in it accelerates. What emerges is not a technology replacing labour but organisations reducing headcount for ordinary commercial reasons while attributing it to a programme that has not yet demonstrated a return.

IMPLICATIONS

That attribution has consequences here. South African firms are reducing staff through unfilled vacancies rather than retrenchment, avoiding the consultation Section 189 of the Labour Relations Act requires, with one labour specialist describing attrition as the employer’s least risky route to its operational objectives. One respondent in the S&P research put the effect plainly, observing that companies now use AI where they would have recruited a young person, and that the result will be a skills gap because the starting point is being skipped. In a country where unemployed youth aged 15 to 34 rose by 264,000 to 5.0 million in the second quarter, pushing the youth rate to 47.4 per cent, that starting point is not a career convenience.

CLOSING TAKEAWAY

Three questions follow for any board approving the next phase. First, what proportion of our AI initiatives is live and delivering measured value, expressed as a number rather than a narrative, given that the global figure is 37 per cent. Second, where headcount has fallen, can we demonstrate the reduction was caused by capability the technology delivered rather than justified by capability we expect it to deliver? Third, if the honest answer to either is uncomfortable, are we prepared to say so before the market says it for us. The research declines to draw a causal line between the cuts and the technology. Companies that have already drawn that line in their own communications should consider whether they can defend it.

Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence’s impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan’s work at https://www.aiforbusiness.net