Independent
Software Developer
HR Manager
Team Leader
Accountant
Copy Writer
Customer Service
Virtual Assistance
Half a Million Seats, One Bet
On 19 July, the world’s biggest contact-center operator — the one the market called uninvestible three weeks earlier — told its 500,000 people every last one of them will be AI-enabled by 2027.
Three weeks ago in this column, we walked through what Bloomberg called the market’s “uninvestible” verdict on the pure-play call-center giants — Teleperformance down 80% from its 2021 peak, short interest more than tripled, and analysts declaring the whole sub-sector a question of whether the businesses had a future worth underwriting at all.
On 19 July, TP — that is now the company’s actual name, Teleperformance retired for a reason — sent back the receipt. Every one of its roughly 500,000 employees will be AI-enabled by early 2027. To fund it, the company will accept 0–2% revenue growth this year and swallow €70–90 million in one-time reskilling costs. If you were looking for the moment the incumbents stopped absorbing the funeral and started arguing with it, this is it.
The consultant walks in the door
The person delivering the pledge is not a lifer defending the status quo. Jorge Amar joined as CEO in March 2026 from McKinsey, where for a decade he ran the firm’s Digital Customer Care practice — designing AI-native customer operations for the world’s biggest companies, including many of TP’s own clients. This is not a routine external CEO hire. It is the outside consultant walking through the door and taking the chair, because the board decided the transformation was existential enough to skip the internal apprenticeship.
Amar’s frame is that TP is not shrinking to survive; it is building “a hybrid workforce that is both human and agentic.” Whether you believe him depends on whether you think a McKinsey partner would take the CEO role at a company he’d have privately shorted a year ago.
What “AI-enabled” has to mean
Here is the honest fault line. Roughly 88% of contact centers have deployed AI in some form; only about 25% have operationalized it into their daily workflow. “AI-enabled” can mean 500,000 people with a Copilot tab open in their browser, or it can mean 500,000 conversations running through an orchestration layer that actually measures the delta.
The measurable side is not vaporware. Microsoft’s own copilot data show agents closing 31% more conversations with 12–16% lower average handle time; year-one ROI averages 41%, compounding past 120% by year three. Concentrix’s own tools already cut agent handling times 25–30%. The productivity is real when it lands. The question buyers should be asking is not “did TP make the pledge” — it is what percentage of the 500,000 seats have crossed from the 88% side of the ledger to the 25% side, and how fast.
Which side of the same soil
TP employs roughly 60,000 people in the Philippines — its second-largest hub after India, and one of the country’s largest private-sector employers. That headcount is the piece of the story an OA reader cannot look away from.
If “AI-enabled” turns out to be a Copilot tab, nothing changes on the Manila floor except a new browser tile — and the labor arbitrage that got the seat there in the first place endures. If it is genuine hybrid workflow — outcomes measured, not hours logged — the same 60,000 people deliver higher-value work at higher margin, and Genpact’s pivot from labor to outcomes turns out to have named the destination.
Either way, the bet the Philippines is now taking is not “does outsourcing end.” It is what outsourcing becomes on the same soil.
The empire has one shot
TP is not alone in the counterattack. HCLTech’s AI revenue is up 62%. Salesforce Agentforce is at $800 million and growing 169% year over year. Every incumbent with scale is now shipping the same rebuttal — that the challengers get bundled back into the platform, not around it.
TP is just the largest and most fully committed of them. Half a million seats and a 0–2% growth guidance haircut is not the profile of a company waiting for the storm to pass. If Amar delivers, the market will owe TP an apology and “unbundling the BPO” will read as the year’s most confidently priced mistake. If he does not, the shorts get paid and the funeral resumes.
Either way, the useful question is no longer whether AI eats the call center. It is which operators build the machine that eats it first.
The question for your business
If your outsourcing partner announced 100% AI-enabled by 2027, would you believe it — or would you run the numbers yourself?
