AI adoption expands, but hiring impact remains limited: Report
ADVERTISEMENT
AI adoption is accelerating, but remains far from broad-based, with uptake concentrated among larger companies and sectors such as information, professional services, finance, and education, according to a report by Panorama, 360 ONE Asset.
The report finds that AI is affecting the labour market primarily through slower hiring rather than widespread layoffs. Job openings in AI-exposed sectors have declined more sharply than in the broader market, suggesting that companies are becoming more cautious about hiring as they look to automate tasks. However, layoffs in sectors such as information and professional and business services show no discernible divergence from the broader market, indicating that AI has yet to trigger large-scale job losses.
At the same time, companies are increasingly citing potential AI-driven productivity gains on earnings calls, signalling a growing intent to automate production processes and substitute labour. Despite the investment push, corporates are becoming more selective in their AI spending.
Debt-funded AI investment raises financial risks
Capital expenditure by hyperscalers and other AI companies continues to grow at a rapid pace and is increasingly being financed through debt. This raises the possibility that vulnerabilities in the AI investment cycle could spill over into credit markets.
The report warns that this risk is amplified by circular transactions, which account for a sizeable share of AI labs' spending and hyperscalers' forward revenue. Such arrangements could increase the sensitivity of the AI investment boom to a slowdown in demand or a reassessment of expected returns.
Chinese models close the gap at lower cost
Competition in AI is also intensifying. Chinese models, many of them open-weight, are rapidly narrowing the performance gap with US models and are reaching comparable capabilities with a lag of only around six months, according to the report.
Chinese models are also gaining an advantage on cost-effectiveness. On a cost-per-task basis, which accounts for both actual token consumption and headline per-token pricing, they operate at a fraction of the cost of leading Western frontier models.
The cost of AI usage is consequently falling. The blended price paid per million tokens has declined by roughly 45% since May 2026, partly as users shift from frontier models towards cheaper open-weight alternatives.
Semiconductor rally raises echoes of the dot-com era
The sharp rally in semiconductor stocks is beginning to resemble the late stages of the 1990s technology cycle, when gains were concentrated in hardware while software stocks failed to participate to the same extent.
A similar divergence is emerging today. Hyperscalers have remained largely rangebound in recent months even as semiconductor stocks have surged. While price-to-earnings multiples have not undergone a major re-rating because of strong earnings growth, the report notes that price-to-book valuations are already at or above dot-com-era peaks. That leaves relatively little room for disappointment if earnings fail to meet elevated expectations.
AI boom faces the classic technology-cycle risk
The scale and pace of the current AI investment boom, driven by expectations of substantial productivity gains, also echo previous technology cycles.
The report points to a recurring pattern in which genuine technological breakthroughs attract capital well in excess of their eventual commercial returns. Such periods have historically been followed by sharp investment reversals, which in some cases contributed to economy-wide recessions.
The Gartner Hype Cycle similarly highlights the risks that can emerge as new technologies move from the “Peak of Inflated Expectations” towards the “Trough of Disillusionment.”
Markets that have benefited heavily from the AI investment theme could be particularly vulnerable if enthusiasm begins to fade. Taiwan and South Korea, where AI-related companies have contributed significantly to recent market gains, could see those gains unwind if the AI cycle loses momentum.
India's equity market is relatively less concentrated and therefore more insulated from a reversal in the global AI trade. However, the report cautions that India's rapidly expanding data-centre buildout could slow if global AI capital expenditure pulls back.