AI Eats the Call Center
Keith Mander@kmanderuk
+1.4%
since Oct 7
The idea
I believe AI agents are replacing offshore call-center work, and the savings show up in the margins of the companies that used to pay for all that support. So instead of shorting Philippine office landlords, which isn't really possible from here, I own the other side: fintechs, travel and gig platforms, software leaders, telcos, and e-commerce names with heavy support costs and a near-term path to cutting them. I'm keeping a 10% cash reserve to add on dips, with no single name over 8%. If the earnings cuts don't materialize within a year or two, the thesis is wrong.
What it owns
- Customer service is one of the biggest expense lines at consumer fintechs, so AI automation shows up fast in their profits. · Klarna GroupIt replaced a large chunk of its outsourced English-language support with AI agents and talks about the savings publicly., PayPal HoldingsIt runs one of the biggest customer support operations in fintech, much of it the offshore work that AI is replacing., SoFi TechnologiesIt is a bank with no branches, so customer service is almost all phone and chat, exactly the work AI absorbs cheaply., Affirm HoldingsIt faces repetitive payment and dispute questions that AI handles cheaply, similar to the playbook already underway at peers., Block IncIt supports millions of small sellers and Cash App users with a heavy load of routine questions that is textbook call-center work.25%
- These companies run some of the largest offshore support operations anywhere, exactly the work AI is automating. · Uber TechnologiesIt built massive support operations in Manila for rider and driver support, exactly the kind of offshore call-center work AI is replacing., Booking HoldingsIt handles enormous volumes of traveler support calls and has already reported support costs coming down as AI takes over., AirbnbIt supports both hosts and guests across every time zone, a double support load that AI agents are built to handle., DoorDashIt has a three-sided support problem with customers, drivers and restaurants all generating routine tickets handled partly offshore.24%
- Salesforce, Microsoft and IBM have already published hard dollar savings from automating their own support operations. · SalesforceIt has cut its support agent headcount significantly through AI and also sells the AI agents doing the replacing., MicrosoftIt has reported hundreds of millions a year saved from AI in customer support and sells the AI that lets others do the same., IBMIt uses its own AI to cut internal service and support costs and then sells that playbook to companies with offshore operations.18%
- Call centers are a core cost for carriers, and routine billing and service calls are the easiest work to automate. · T-Mobile USIt runs massive call centers handling mostly billing and plan questions that AI agents process well., Verizon CommunicationsIt fields millions of billing and service calls, the plainest version of grunt work that AI is built to replace., ComcastCable customer service is famously expensive, which makes it a prime target for AI automation to shrink costs.12%
- Order status, returns and shipping questions are the most automatable tickets in any industry. · ShopifyIt supports millions of merchants with mostly repeatable questions and has been aggressive about using AI internally instead of hiring., eBayIts support load is dominated by returns, disputes and seller issues, structured problems that AI agents resolve cheaply.11%
- Cash10%