The world’s largest health and beauty retailer says AI will make shopping ‘more human, not less’

AS Watson CEO Malina Ngai said the retailer is deploying AI to enhance employee productivity and customer service rather than reduce headcount, reporting materially higher employee engagement and rising customer-love scores since launching its human-AI strategy last October. At its automated Foshan warehouse, robots have shifted workforce composition to 62% female by replacing heavy-lifting tasks traditionally performed by men. The 17,000-store health and beauty retailer is reportedly considering Hong Kong and London listings at an approximately $30 billion valuation.

Analysis

The relevant read-through is not labor-cost savings but whether AI raises sales per labor hour and customer retention enough to fund a higher-quality store workforce. For specialty retail, frontline service can lift conversion, basket attachment and loyalty, but those gains are difficult to verify and are likely to emerge over 2-4 reporting periods rather than in near-term margins. Any eventual AS Watson listing should be valued against global health-and-beauty peers on same-store sales, digital fulfillment economics and inventory turns—not employee-engagement metrics.

COST is a modest beneficiary of the same operating philosophy: its membership model makes labor retention and service quality economically valuable because renewal rates compound customer lifetime value. The more important near-term risk is that investors may reward AI narratives before evidence of labor-hour productivity or SG&A leverage; Costco's wage investments already leave limited room for a multiple-expansion story based solely on workforce augmentation. Watch quarterly renewal, traffic, wage inflation and SG&A as a percentage of sales.

ORCL's recent workforce reduction should not be read as a direct competitive negative from this retail case; its valuation is driven primarily by OCI capacity deployment and AI cloud contract conversion. The second-order issue is that enterprise customers adopting AI primarily for workflow improvement, rather than headcount removal, may produce slower and less measurable ROI, extending sales cycles and increasing scrutiny of AI infrastructure commitments. That is a risk to aggressive AI-demand assumptions over the next 6-12 months, though one retailer's approach is insufficient to change the thesis.

Contrarian view: “human-centric AI” can be margin-accretive only if automation removes low-value administrative work without creating new systems, training and compliance overhead. Retailers often reinvest productivity into service, leaving EBIT margins flat while customer metrics improve; this is strategically sound but does not justify paying a near-term automation premium. A prospective Watson IPO could therefore be attractive only if disclosed unit economics show both service-led sales uplift and fulfillment-cost leverage.

Key Decisions for Investors

• No standalone trade on the article; treat any AS Watson IPO as a watch item. Require disclosure of same-store sales, online penetration, labor cost/sales, inventory turns and EBITDA margin by geography before underwriting a long position.
• Maintain a modest long bias in COST only on operational confirmation: add after a quarter showing stable-to-higher renewal and traffic with SG&A leverage despite wage pressure. Falsifier: renewal deterioration or SG&A deleverage without compensating comparable-sales acceleration; reassess over 1-3 quarters.
• Do not short ORCL on this signal. Instead, monitor 6-12 month AI-cloud demand quality through OCI remaining performance obligations, capex intensity and management commentary on customer ROI; a widening gap between infrastructure spend and monetized cloud growth would support a tactical underweight versus MSFT.
• For a potential Watson listing, screen a relative-value pair of long Watson / short a regional beauty-retail proxy only if the issuer demonstrates structurally superior digital fulfillment economics and recurring customer engagement; absent those disclosures, IPO scarcity and AI branding are likely to create unfavorable valuation risk.