Independent
Software Developer
HR Manager
Team Leader
Accountant
Copy Writer
Customer Service
Virtual Assistance
Going Private Was the Smart Trade
$1.62B exit for TaskUs. 16-year low for Teleperformance. The market has finished pricing one trade.
TaskUs took a $1.62 billion exit in August 2025. Teleperformance, ten months later, sits at its lowest stock price since 2016 — carrying the highest short interest of any European technology-services name.
The going-private trade and the staying-public trade are now visibly different bets. The market has finished pricing one of them.
Last year’s column on the going-private wave flagged it as a strategic retreat. Ten months on, the bill for not going private has arrived.
TaskUs got out
The TaskUs deal closed at $1.62 billion all-cash with Blackstone. The framing at the time was about reducing public-market noise to focus on AI development.
The framing that’s clearer in retrospect was about escaping the public-market AI narrative before it landed on the stock.
Teleperformance and Concentrix didn’t go private. They have spent 2026 paying the public-market cost of that choice.
TP was removed from the CAC 40 earlier this year. It now trades at a 70 to 80 percent discount to its historical P/E — not because earnings collapsed (management still guides to 15 percent EBITA margins and flat-to-positive top-line in 2026), but because hedge funds have repriced the narrative the stock is attached to.
The going-private move bought TaskUs a closed room to run its AI strategy in. The staying-public stance bought TP a Bloomberg headline calling it “one of AI’s first victims.”
The trade that landed
Hedge fund names you recognize: Marshall Wace. Point72. Citadel Advisors. Squarepoint. Per Hedgeweek and European Business Magazine reporting in June 2026, short interest in TP rose from 3.8 percent in early 2026 to 12.24 percent by May. The average European tech-services short interest is 2.4 percent. TP is now five times that.
This isn’t speculative. It is institutional capital, named and disclosed, betting against a specific labor model — at scale.
The thesis traces back to a single 2024 disclosure: Klarna saying its AI agent handled the workload of 700 full-time customer-service agents. That number is what the entire trade is built on.
It is also a thesis whose evidence has weakened. OA reported in May 2025 that Klarna started rehiring humans — the original catalyst was partly walked back. The trade kept getting added to anyway.
What the bears might miss
The sharpest counter to the short trade comes from a Seeking Alpha piece titled bluntly: Can Teleperformance Survive AI? Valuation Says It Doesn’t Need To.
The math is straightforward. At 5.3 times forward free cash flow and a 15 percent EBITA margin guide, TP is priced as if it’s already a structurally declining business. If AI eats some of the revenue but the margin holds, the equity still works as a cash-return story.
That’s the bears’ second problem. They are right about direction — AI will compress the labor-intensive BPO model. But “right about direction” and “right about the current price” are two different trades. At TP’s current multiple, the bull case isn’t needed. The bears just have to be more right than the price already says.
Where the trade doesn’t apply
There is a clean dividing line between this short trade and the rest of the outsourcing industry.
The trade is targeted at publicly-traded, mature, labor-intensive, US-and-Europe-listed call-center operators. That’s a narrow band. It does not apply to:
— The OA500 universe of mid-market BPOs, most of which are privately held and not subject to public-market AI rerating.
— The captive / GCC build-out, which is being staffed up in Pune, Cairo, and Bengaluru on the same AI shift.
— Specialist providers in healthcare, finance, and engineering services where the work doesn’t compress to a Klarna-style automation case.
— The Philippine and Indian BPO ecosystems, where IBPAP is actively forecasting net job creation from AI, not destruction.
The bears are shorting a narrative about a sub-segment. They are not shorting outsourcing.
The trade landed on the names that stayed public. The names that went private — TaskUs, the captives, the OA500 universe — didn’t pay the rerating cost.
That is the lesson from ten months ago that becomes obvious now: capital structure decided who survived the AI narrative.
The narrative itself may be wrong. The price the public BPOs paid for being attached to it is real.
The question for your business
When capital prices in a narrative, is the narrative what you should believe?
