AI Adoption High, but 'Incredibly Shallow' Says Dr. Rebecca Homkes

AI adoption is accelerating across industries, but its impact on productivity and efficiency remains uneven and difficult to measure. Dr. Rebecca Homkes of London Business School says companies are using AI to their advantage in varied ways, making the broader economic effects unclear. The article is largely interpretive commentary rather than a market-moving event.

Analysis

The key market implication is not that AI is already boosting output everywhere, but that the dispersion in realized productivity is widening. That favors vendors that sit closest to workflow automation, data plumbing, and model orchestration, while penalizing “AI-washed” software that adds cost without measurable labor substitution. In the near term, the winners are likely to be the picks-and-shovels layer; over 6-18 months, the more interesting alpha may come from firms that can translate AI into fewer headcount additions rather than headline usage metrics.

Second-order effects are important: if companies broadly deploy AI but fail to capture efficiency, margins may not expand and hiring may only slow at the margin. That creates a subtle negative for sectors where investors are already underwriting AI-driven operating leverage, especially large-cap software and IT services names priced for immediate monetization. Conversely, firms with large repetitive back-office labor pools could see a step-function improvement once managements hardwire AI into processes, but only if governance and change management are strong enough to prevent productivity leakage.

The contrarian view is that the market may be overestimating the speed of substitution and underestimating the integration bottleneck. Adoption can be rapid while payback remains slow because the binding constraint is not model access, but workflow redesign, data quality, and employee behavior. If macro weakens, management teams will be pushed from experimentation to cost takeout faster than expected, which could create a 2-4 quarter lag before benefits show up in reported margins; until then, the theme risks becoming a capex and software expense story rather than an earnings story.

Catalysts to watch are earnings calls that quantify AI-linked productivity gains in hours saved, customer resolution rates, or SG&A reduction, not just pilot counts. A negative catalyst would be rising evidence of AI-driven churn or quality issues, which would force companies to slow deployment and delay ROI. The best trading setup is likely a relative-value trade between AI infrastructure beneficiaries and application-layer names whose monetization is still aspirational.

Key Decisions for Investors

• Long AI infrastructure and workflow-enablement names vs short application-layer software with weak proof of ROI; use a 3-6 month horizon and expect outperformance if management teams keep highlighting pilots but not margins.
• Buy call spreads on semiconductor and cloud infrastructure leaders into any post-earnings weakness over the next 1-2 months; the setup improves if commentary emphasizes capex persistence despite uncertain end-user monetization.
• Short high-multiple IT services or outsourcing names that rely on labor-arbitrage models over a 6-12 month horizon; risk/reward improves if clients start demanding AI-enabled cost cuts without corresponding pricing power.
• Hold off on buying broad ‘AI productivity’ beneficiaries until companies report measurable operating leverage; the better entry is after one or two quarters of evidence, not on adoption headlines.
• For event-driven upside, buy medium-dated calls on firms that can directly quantify AI-driven SG&A reduction in upcoming earnings; target 2:1 to 3:1 payoff if management provides credible KPI improvement.