The AI boom hasn’t stopped U.S. companies from hiring cheap offshore labor, and overseas call center employment is still skyrocketing

Apollo chief economist Torsten Slok argues that AI is not reducing offshore customer service employment; call center jobs in the Philippines have nearly doubled to 2 million from 2016 through 2025, while unemployment fell from 9% to about 4% from 2021 to March 2026. He frames this as Jevons paradox: lower per-interaction costs from AI are expanding, not shrinking, demand for customer support. The article also cites a 14% hourly productivity lift from an AI assistant in a 2023 study of more than 5,000 agents.

Analysis

The market is still framing agentic AI as a pure labor substitution story, but the more important second-order effect is demand expansion. If AI reduces the marginal cost of a support interaction, companies will not only preserve service levels with fewer humans; they will also open more channels, take more low-value customers, and serve more geographies that were previously uneconomic. That is a structural tailwind for offshore labor hubs and for software vendors that sit in the middle of the workflow, because usage can scale faster than headcount savings.

CRM is the cleanest beneficiary on the enterprise software side, but the earnings upside is not from near-term seat growth; it is from higher attach, more usage-based consumption, and better retention as AI becomes embedded in customer operations. The risk is that investors are over-discounting labor displacement while underpricing the monetization layer: AI can compress costs, but the budget often gets reallocated to more interactions, not less. That dynamic tends to favor platforms with distribution into service, sales, and support over pure point solutions.

EBAY is a useful contrarian read-through because earlier AI-driven translation/productivity gains did not kill the cross-border marketplace story; they expanded it. The same mechanism can apply to service-heavy commerce: easier language handling and lower support costs increase conversion in smaller markets and for lower-AOV transactions that were previously uneconomic to service. The main risk to this thesis is a rapid jump in model capability that materially improves end-to-end automation of complex cases, but that looks more like a multi-quarter to multi-year path than a near-term earnings shock.

The consensus-miss is that AI can be deflationary per interaction while being inflationary for total interaction volume. That favors companies with large installed bases, international exposure, and monetization tied to activity rather than headcount. If the data continue to show stable or rising offshore employment over the next 6-12 months, the market will likely rotate from "AI replaces labor" to "AI expands labor-intensive service demand," which is bullish for platforms and bearish for linear labor-cost compression narratives.

Key Decisions for Investors

• Long CRM vs short a basket of legacy BPO/service outsourcers for 3-6 months: benefit from AI-driven workflow expansion and software monetization while outsourcing names face margin pressure if clients use AI to renegotiate pricing.
• Buy CRM Jan-2027 calls on weakness; target a 2-3x payoff if AI attach rates accelerate and the market re-rates the company as an agentic workflow platform rather than a CRM seat story.
• Long EBAY 6-12 months: the market still underestimates how AI-enabled translation and support reduce friction in cross-border GMV, supporting conversion and take-rate stability; use a tight stop if management commentary shows slower international mix.
• Pair long CRM / short WDAY over 2 quarters if you want to express the view that customer-facing AI monetization will outpace back-office HR/workflow AI; CRM should see more direct usage upside from higher interaction volumes.
• If offshore labor data remain firm into the next print cycle, add to positions in AI infrastructure beneficiaries rather than labor-displacement names; use the data as confirmation that the real winner is usage growth, not headcount reduction.