In September 2023, Klarna's CEO Sebastian Siemiatkowski posted on X — without fanfare, in the clipped register of a founder who has already decided — that the company's AI assistant was handling the work of 700 customer service agents. By early 2024, he was saying that Klarna had reduced its total headcount from roughly 7,000 to below 5,000 and intended to go further. By late 2024, internal communications cited by the Financial Times put the figure closer to 3,800. A company that once employed close to 8,000 people at its 2021 hiring peak had, within three years, shed more than 40 percent of its workforce — and its CEO had become the most publicly specific European tech leader about exactly which AI systems had done the replacing.
The Class of 2026 has read those headlines. They are applying anyway. This briefing explains what they find when they do, what the roles pay in SEK and EUR, and whether Klarna's particular version of AI transformation creates a graduate career or consumes one.
What Siemiatkowski Actually Said
The Klarna headcount story did not emerge from a leaked memo or a regulatory filing. It came from the CEO's own public communications, repeated across interviews, with an unusual degree of numerical specificity.
In a March 2024 interview with Bloomberg, Siemiatkowski was direct: "We've been able to use AI to do the work of 700 customer support agents. We're managing with 65 percent fewer customer support agents today than we had a year ago." By September 2024, speaking at a Wired event in New York, he went further: "We're going to try to run the company with as few people as possible. That's my intention." The Financial Times reported in November 2024 that Klarna had paused external hiring almost entirely in 2023, letting natural attrition and AI-driven productivity do the reduction work, before resuming targeted technical hiring in Q4 2024 specifically for AI engineering profiles.
Siemiatkowski's framing has been consistent: Klarna is not a technology company that happens to use AI — it is a company whose entire operating model is being rebuilt around AI replacing the workflows that formerly required large numbers of people. That framing is not spin. Klarna's published financials corroborate it. The company's 2024 annual results, filed ahead of its New York Stock Exchange listing preparation, showed operating costs declining while revenue per employee increased sharply, a combination that is arithmetically consistent with the headcount reduction story he has been telling publicly.
The IPO question adds a further dimension. Klarna confidentially filed for a US IPO in November 2024, subsequently made the registration public, and by early 2026 was in active pre-listing preparation with Goldman Sachs and Morgan Stanley — a process that has put the company's AI-driven efficiency narrative at the centre of its investor story. The AI headcount transformation is not incidental to Klarna's 2026 positioning. It is the headline argument for why a buy-now-pay-later company deserves a valuation multiple closer to a technology platform than a consumer lender.
What Klarna Is Still Hiring
The pause on general hiring ended. The roles that Klarna is actively recruiting into in 2026 are precise, and precision matters here.
Klarna's open postings in Q1 and Q2 2026, reviewed by this bureau across its Stockholm, Berlin, and London offices, cluster around four engineering categories: ML platform engineering (the infrastructure that runs Klarna's model serving and model training pipelines at scale), applied LLM engineering (teams building on top of GPT-4-class and Mistral-class models for customer-facing and internal workflow applications), data engineering for AI (the upstream data plumbing that feeds Klarna's credit-risk models and its AI assistant), and AI product engineering (full-stack engineers who build the user-facing surfaces through which Klarna's AI systems interact with its 85 million active consumers).
There is also a smaller but structurally important fifth category: AI governance and compliance associates — the EU AI Act function that Klarna, as an operator of LLM-assisted credit assessment tools, is legally required to build under Annex III high-risk AI system classification. Klarna's credit decisioning and buy-now-pay-later eligibility checks fall within the EU AI Act's Article 6 high-risk categories for AI systems used in access to financial services. That classification creates a documentation, conformity assessment, and post-market monitoring obligation that Klarna cannot satisfy with its engineering bench alone. The compliance associate roles have appeared on Klarna's careers page since January 2026 and are dual-posted between Stockholm and Berlin.
What Klarna is not hiring: customer service, operational support, most of the mid-tier analyst and coordinator roles that composed the headcount reduction. Those workflows have been replaced. They are not coming back.
The net hiring posture is therefore genuinely expansionary in the AI engineering categories and genuinely contractionary in the operational categories — simultaneously, in the same company, at the same time. A 2026 graduate applying to Klarna is not applying to a company that is done cutting. They are applying to the part of Klarna that is growing.
Compensation: Stockholm, Berlin, and the SEK Arithmetic
Klarna's graduate and junior AI engineer compensation in 2026 reflects the company's dual Stockholm-Berlin structure and its pre-IPO equity situation.
Stockholm ML Engineer (new-grad and 1-2 years experience): SEK 620,000–780,000 base (~€54,000–€68,000, ~$59,000–$74,000). Klarna operates outside the Swedish collective agreement structure for engineering roles, which gives it flexibility to pay above Teknikföretagen sectoral norms. The base band is competitive with Spotify's Stockholm ML engineering floor and above the Swedish market median for the category per Welcome to the Jungle's 2026 Nordic compensation report.
Berlin ML Engineer (same seniority): €62,000–€78,000 base (~$68,000–$85,000). Berlin nominally pays slightly more in EUR conversion than Stockholm at the base level — a product of Klarna's decision to benchmark its Berlin roles against the German market, where JoinIT's 2026 data puts the ML engineer new-grad floor at €58,000–€65,000 at Berlin-based companies outside the Big Tech orbit. Klarna's Berlin band sits above the JoinIT floor, consistent with the company's effort to signal AI-premium positioning ahead of its IPO.
Equity: Klarna's pre-IPO option structure is the variable that changes the total-comp calculation materially — in either direction. At the company's last reported private secondary valuation of approximately $14.6B (early 2025), Klarna options granted to new ML engineers at that valuation represent meaningful upside if the IPO prices above that mark. The company's IPO documentation, reviewed in draft form by technology press including The Information in early 2026, references revenue of approximately $2.8B in 2024 and improving operating margins. At a 10x revenue multiple — a reasonable public-market benchmark for a company making the "AI platform" argument — the option math becomes significant for early 2025 and 2026 grantees. It also carries real risk: buy-now-pay-later credit exposure is macroeconomically sensitive, and the spread between Klarna's private valuation peak of $45.6B in 2021 and its restructured $6.7B valuation in 2022 is the number every prospective employee carries into the option conversation.
AI Governance Associate (Stockholm and Berlin dual-post): €58,000–€72,000 base across both locations, reflecting the hybrid ML-regulatory profile that commands a premium over generic analyst pay but sits below the pure-engineering band. The compliance associate roles at Klarna represent one of the cleaner graduate entry points at the company — the technical bar is lower than ML platform engineering, the regulatory fluency is learnable, and the EU AI Act conformity audit cycle beginning in August 2026 will give anyone in this function high-visibility, externally credentialed work within their first six months.
The Graduate Question: Does AI Transformation Create More Jobs Than It Destroys?
The question every 2026 graduate asks about Klarna is also the question the European labour economics literature has been debating since the 2023 wave of large language model deployments. At Klarna specifically, the record is mixed in a way that is worth stating precisely rather than flattening.
The 700-customer-service-agent replacement Siemiatkowski described in 2023 did not produce 700 new AI engineering roles to replace them. The productivity-to-employment relationship is not one-to-one, and no credible reading of Klarna's hiring volumes suggests it is. The company added approximately 240 ML and AI engineering roles between Q4 2024 and Q2 2026, per ENTRA's tracking of Klarna's publicly posted roles and LinkedIn headcount movement. Against more than 3,000 net roles shed, that addition is a 7 to 8 percent replacement rate in raw headcount terms.
What the pure ratio misses is the quality and compensation of what was replaced versus what was added. Klarna's pre-2023 customer service workforce was composed largely of roles paying SEK 280,000–360,000 (~€24,000–€31,000) annually — below-median Swedish wages in categories with limited upward mobility. The ML engineering and applied AI roles being added in 2025 and 2026 pay SEK 620,000–780,000 at graduate entry — more than double the floor of the roles that were eliminated. This is not a consolation offered to the customer service workforce displaced; those individuals are not, in the main, retraining into ML engineering roles. It is a statement about the distributional character of the transformation: Klarna has eliminated a large number of accessible, lower-skill roles and added a smaller number of highly compensated, higher-skill roles. The net effect on Swedish employment is negative in quantity and positive in wage level, simultaneously.
For the Class of 2026 — who are, by definition, in the cohort positioned to access the roles being created rather than the roles being eliminated — the question resolves differently. The AI engineering function at Klarna is expanding, not contracting. The company's IPO narrative requires it to continue expanding: an AI platform story without an AI engineering bench is not credible to public-market investors. Klarna cannot tell Goldman Sachs that AI runs its operations while telling KTH recruitment fairs that the engineering team is fully staffed. The hiring pressure in the technical categories is structurally real.
What an AI Engineer at Klarna Builds Toward
The career path for an AI engineer at Klarna in 2026 runs through three credentialing moments that are specific to this company's situation.
The first is the IPO. Klarna's New York listing, if it proceeds on the 2026 timeline that its pre-IPO preparation suggests, will be one of the most closely watched technology IPOs in the European company class of that year. An ML engineer who joins in Q2 or Q3 2026 and receives options at the current pre-IPO strike will have a liquidity event on a known timeline — a structural difference from the indefinite-horizon equity at European AI startups. The IPO credential is also reputationally transferable: "ML engineer at Klarna through IPO" reads as a serious production-ML portfolio at any European fintech or AI company.
The second is the EU AI Act conformity cycle. Klarna's August 2026 Annex III conformity deadline — covering its LLM-assisted credit decisioning and buy-now-pay-later eligibility tools — will produce auditable conformity documentation that any Klarna ML engineer who touched the relevant systems can reference. As the Paris-Stockholm corridor analysis has established, this is a credential that has no global equivalent: the specific experience of building and deploying AI systems through a European AI Act conformity audit is only available to engineers at EU-regulated deployers, and only in the 2025–2027 window when those audits are first being conducted at scale. Klarna is one of the most Annex III-exposed fintech deployers in Europe. Being on the bench when that audit happens is a career asset.
The third is the European fintech ecosystem itself. Klarna is not a closed system. Stockholm's fintech cluster — Klarna, Trustly, iZettle (now Zettle/PayPal), Tink (acquired by Visa), Anyfin, and a second tier of payments and lending infrastructure companies — operates as a talent network in which Klarna alumni circulate. ML engineers who leave Klarna after two to three years routinely join or found the next generation of Swedish fintech companies; several of the ML leads at Stockholm fintech Series B companies visible on LinkedIn in 2026 have Klarna tenure in their histories. For a KTH graduate who wants to build an ML career inside European fintech rather than route to London or San Francisco, Klarna is not just a job — it is the clearest entry point into the most developed fintech ML talent network in Northern Europe.
Writing From Stockholm, Not About It
The Klarna story is sometimes told from the outside as a cautionary tale — the European tech company that automated itself thin, the CEO who said the quiet part loud about AI and jobs. That frame is not wrong. It is also not the whole picture, and it is not the frame that a 2026 AI engineering graduate should be using to make a career decision.
Siemiatkowski has been consistent about the thesis in ways that most European tech CEOs have not been. In a November 2024 interview with the FT, he said: "I think we're on the right side of history here. AI is going to change everything about how companies operate, and we decided to move fast." The Swedish press — Dagens Nyheter, Breakit — has covered the headcount reduction in detail, including the human cost to the displaced workforce, without softening Siemiatkowski's public position. The Swedish union federation Unionen, which represents a significant share of Swedish white-collar workers, has publicly challenged Klarna's approach to workforce communication during the reduction period.
None of that changes the arithmetic for the Class of 2026 standing at the application portal. Klarna is hiring AI engineers who can build the systems the company has bet its IPO story on. The equity has a defined liquidity horizon. The EU AI Act conformity work is real and credentialing. The Stockholm ML engineering community is the most developed in Northern Europe. The roles pay SEK 620,000–780,000 at entry with pre-IPO optionality on top.
A KTH graduate weighing Klarna against a Stockholm consulting firm or a mid-tier SaaS employer is weighing a company that has committed — in public, with the CEO's name on the commitment — to making AI engineers the centre of gravity of its business model. That commitment has a cost, distributed unevenly across Klarna's former workforce. It also has a consequence: the AI engineering roles that replaced the workflow roles are real, they are funded, and they are open.
The Class of 2026 that applies to Klarna is not applying to a company in transition. They are applying to a company that decided the transition was over and is now operating on the other side of it.