Klarna's total headcount fell from a peak of approximately 7,000 in early 2022 to 3,422 at end-2024 (per Klarna's NYSE IPO prospectus, which disclosed year-end headcounts of 5,527 in 2022, 4,352 in 2023, and 3,422 in 2024) — a reduction the company's own CEO attributed, with unusual specificity, to AI taking the work of hundreds of customer service agents. By mid-2026, headcount stands at approximately 3,000, per CEO Sebastian Siemiatkowski's public statements in May 2026; net AI-adjacent additions since the 2024 trough are estimated at approximately 700 roles. The new roles are not the old roles. They pay roughly double what was cut, require AI engineering fluency rather than customer service process knowledge, and sit inside a Stockholm team that has tripled in twelve months. The composition of what Klarna is building now — and how it is compensating the people building it — has become the most empirically grounded case study available in European enterprise for what post-AI-transition hiring actually looks like.
What Klarna Actually Cut
The reduction was not ambiguous. Sebastian Siemiatkowski, in multiple public statements throughout 2023 and 2024, made Klarna's AI automation thesis explicit in ways that most European CEOs avoid. His February 2024 announcement — that Klarna's AI assistant, built on OpenAI GPT-4 integration, was performing the equivalent work of 700 customer service agents within one month of launch — was the most public marker of a structural headcount programme that cut the company from its 2022 peak to 3,422 by end-2024, a reduction of more than 35 percent over two years from the formal year-end 2022 figure of 5,527.
The roles eliminated were concentrated in customer service operations: agents handling first-contact resolution, complaint escalation, payment dispute processing, and multilingual support across Klarna's 85 million active consumer base in 45 markets. The GPT-4-based assistant, integrated into Klarna's customer-facing chat interface, handles first-contact resolution in over 35 languages without handoff. Internal operational data cited in Klarna's IPO filings — the company listed on the NYSE in September 2025 under the ticker KLAR — showed resolution times falling and per-contact costs dropping substantially during the 2023-2024 automation period.
Siemiatkowski has been consistent about the underlying thesis. In an April 2026 interview with the Financial Times, he said: "AI costs are falling while human judgment value is rising — that is the core dynamic. We automated the repeatable. We are now hiring for the irreplaceable." The customer service workforce that bore the reduction — roles that paid SEK 280,000–360,000 annually (~€24,000–€31,000, ~$26,000–$34,000 equiv at June 2026 EUR/USD) — was not positioned for that transition. The retraining pipeline that Klarna offered to displaced workers, reviewed in coverage by Dagens Nyheter in Q3 2024, was limited in scope and not designed to move the operational workforce into engineering roles. What the company cut and what it is now adding are categorically different labour market populations.
What Klarna Is Now Hiring
The AI-adjacent hiring since the 2024 trough has been selective. Per ENTRA's tracking of Klarna's publicly posted roles and LinkedIn headcount movement through Q2 2026, the approximately 700 AI-adjacent additions — offset by continued attrition in legacy operations functions — cluster in five categories:
| Role Title | Location | Base Range | EUR/USD Equiv | |---|---|---|---| | AI Model Trainer | Stockholm / Remote-EU | SEK 900K–1.1M | €78K–€95K (~$85K–$104K) | | Prompt Operations Specialist | Stockholm / Berlin | SEK 820K–980K | €71K–€85K (~$78K–$93K) | | AI Customer Experience Engineer | Stockholm | SEK 1.4M–1.8M | €121K–€155K (~$132K–$169K) | | Trust and Safety AI Analyst | Amsterdam / Stockholm | €68K–€82K | (~$74K–$90K) | | AI Conformity Assessment Manager | Amsterdam / Frankfurt | €85K–€105K | (~$93K–$115K) |
Senior AI product engineers — the most experienced tier within the AI Customer Experience and ML platform functions — are clearing €130K–€160K base (~$142K–$175K equiv), against the €75K–€90K ceiling of the customer service manager layer that was eliminated. The 2× comp differential cited in ENTRA's top-line figure is the ratio between the replaced ops management layer and the AI-adjacent roles being added; at the individual contributor level, the gap between displaced service agents and new AI engineering hires is wider.
The Stockholm AI team is the operational core of this expansion. It has grown from approximately 60 people in mid-2025 to approximately 180 as of ENTRA's Q2 2026 tracking — a tripling in twelve months that makes it the fastest-growing AI engineering function in Sweden outside Spotify's Generative Music Lab. The build is anchored in pre-IPO equity: Klarna listed at $40 on NYSE in September 2025 and traded at approximately $17–$18 as of late June 2026 — well below its IPO price — making the equity story more complicated than at grant, but the company continues to issue options as a retention tool for senior technical staff, and the prospect of equity normalisation is structurally embedded in Klarna's H2 2026 technical hiring pitch.
The Prompt Operations Specialist title is worth separating from the engineering roles. It is a role that Klarna appears to have created internally — it does not appear in European tech job boards at scale outside Klarna's own listings — and it reflects the operational reality of running a GPT-4-based customer system at 85 million user scale. The role manages the prompt architecture, output quality monitoring, and human escalation thresholds for Klarna's AI assistant: a workflow function that did not exist before LLM deployment and is not straightforwardly an engineering or operations role. Compensation for this title at Klarna sits below the engineering band but above the customer operations floor — approximately SEK 820K–980K (~€71K–€85K), a deliberate mid-tier designed to attract former operations leads who have acquired prompt literacy.
The EU AI Act Overlay
Klarna's credit-assessment AI — the LLM-assisted system that evaluates buy-now-pay-later eligibility across 45 markets — falls within Annex III, section 5(b) of the EU AI Act (Regulation (EU) 2024/1689): high-risk AI systems used to evaluate creditworthiness or establish credit scores. That classification carries mandatory conformity assessment, Article 11-compliant technical documentation, and post-market monitoring obligations. Klarna's August 2026 compliance deadline — the point at which Annex III enforcement becomes fully active under the Act's phased timeline — is generating a distinct category of hiring that runs parallel to the AI engineering build.
Amsterdam and Frankfurt function as Klarna's EU compliance nodes, separate from the Stockholm engineering centre. Amsterdam houses Klarna's EU regulated entity and is where the AI Governance Associate and Trust and Safety AI Analyst functions are posted; Frankfurt is the secondary compliance hub for the German market, where Klarna operates under BaFin supervision. The Conformity Assessment Manager role — budgeted at €85K–€105K base (~$93K–$115K equiv), per ENTRA's job board monitoring for Q2 2026 — owns the Article 43 conformity process for Klarna's credit AI system and interfaces with any notified body involvement. This is a role that combines knowledge of the Annex III classification framework with practical experience of model documentation at production scale: a profile that has no direct precedent in the European compliance market before 2025 and is currently being built, by necessity, from adjacent populations — fintech compliance officers who have acquired AI Act literacy, and AI governance researchers who have acquired financial services regulatory knowledge.
The August 2026 deadline is a forcing function rather than a theoretical milestone. The AI Office, operational since March 2025 under DG CONNECT, has been explicit in its Q3 2025 enforcement guidance that Annex III deployers are expected to present conformity documentation on demand. For Klarna, which operates a GPT-4-based system in direct credit assessment scope, the documentation burden is material. The compliance hiring is not optional overhead — it is the legal prerequisite for continuing to run the AI assistant that generated the efficiency gains that funded the company's IPO narrative.
European Implications
What SAP's leadership in Walldorf is watching in Stockholm, what Publicis Groupe's transformation office is tracking from Paris, what Deutsche Telekom's AI division in Bonn is mapping — is whether the Klarna model holds across enterprise types larger and more structurally complex than a fintech.
Publicis Groupe executed a comparable but less numerically transparent workforce reset between 2023 and 2025, reducing its global headcount from approximately 103,000 to 97,000 while expanding its AI and data engineering bench — the French holding company's equivalent of Klarna's AI assistant build is Marcel, its proprietary AI creative and workflow platform. The Publicis model shares the Klarna logic: automate the repeatable process layer, retain and expand the judgment and engineering layer, accept that the distributional consequence falls unevenly across the workforce. The difference is that Publicis has not put numbers on it the way Siemiatkowski has.
For the Stockholm labour market, the Klarna AI team's growth to 180 people coincides with Spotify's 240-researcher AI lab expansion and Northvolt's AI-manufacturing hiring programme — creating unusual, concentrated demand for ML engineers in a city that was, until 2024, consistently below the European senior-IC compensation ceiling. The combined effect has repriced senior Stockholm AI roles to €220K–€310K total comp at the principal level, per ENTRA's H1 2026 Nordic Talent Index. Klarna's tripling of its Stockholm AI team is a direct contributor to that repricing: the competition for ML platform engineers between Klarna, Spotify, and Ericsson Kista is a Stockholm-specific talent market dynamic that did not exist at this intensity two years ago.
Whether the Klarna model represents a net gain for EU tech employment depends on the frame. In raw headcount, the estimated 700 AI-adjacent roles added against roughly 2,100 cut is a replacement rate below 35 percent. In wage-weighted terms, the roles created carry aggregate compensation roughly 2.2 times the aggregate compensation of the roles eliminated — a net increase in the wage bill for the retained workforce even as total headcount continues to fall. The EU employment debate will not resolve on Klarna's numbers alone. But Klarna's numbers are the clearest empirical input that debate has had in European enterprise in H1 2026.
By August 2026, when Klarna's Annex III conformity deadline arrives and the first round of AI Office enforcement reviews begins, the company will have completed the first full cycle of post-AI-transition workforce restructuring in European fintech — cut, automate, rehire selectively, comply. Every European enterprise CFO running a similar analysis is reading the same Klarna filings.
Global Career PlatformFind AI talent. Find your next role.
Booking is hotels. · Airbnb is apartments. · ENTRA is global careers.