Klarna CEO: AI to Collapse Software Valuations
Klarna CEO Sebastian Siemiatkowski argued that generative AI is driving the cost of creating software toward zero, which could cause traditional SaaS company valuation multiples to fall from 20-30x to as low as 1-2x revenue. He also predicts that AI agents will make it trivial to migrate data between platforms, eroding the data lock-in that protects many incumbents. Siemiatkowski noted that AI has already enabled Klarna to significantly reduce its workforce while increasing efficiency.
- Klarna's AI-powered customer service assistant, developed in partnership with OpenAI, handled 2.3 million conversations in its first month, which is two-thirds of all customer service chats. This AI is doing the equivalent work of 700 full-time agents and is projected to increase Klarna's profits by $40 million in 2024. - The company has nearly halved its workforce since 2022, from 5,527 employees to 2,907, largely by not replacing departing staff and instead integrating AI. CEO Sebastian Siemiatkowski anticipates the employee count will drop below 2,000 by 2030 as AI capabilities expand. Despite the staff reduction, Klarna reported a 108% increase in revenue while keeping operating costs flat. - While traditional SaaS companies trade at valuation multiples of 2.5-7x revenue, AI-native SaaS companies are commanding significantly higher multiples, with a median of 25.8x. This premium is attributed to factors like defensible data moats, winner-take-all market dynamics, and superior unit economics. - The demand for AI-related skills is reshaping the software engineering job market, with AI/ML engineers earning salaries between $130,000 and $200,000+. Engineers with AI skills can command a salary premium of nearly 18% compared to their peers. Conversely, traditional software engineering roles are projected to decline by as much as 20-24%. - In the AI chip market for data centers, Nvidia's market share has surged from 25% in 2021 to 86% in late 2025. In contrast, Intel's share has fallen from 68% to 6% during the same period. The total market for data center AI chips is forecasted to exceed $400 billion by 2030. - Venture capital investment in AI is robust, with AI startups attracting $89.4 billion in 2025, accounting for 34% of all global VC funding. Investors are shifting focus from traditional SaaS metrics to a startup's data moats, model performance, and the technical depth of the team. - Hyperscalers like AWS, Google Cloud, and Microsoft Azure are engaged in a build-versus-buy dilemma for AI infrastructure, facing a shortage of high-density computing capacity. To bridge this gap, they are partnering with specialized "neo-cloud" providers like CoreWeave and Lambda Labs to lease large blocks of GPU capacity. - The intense energy consumption of AI data centers is becoming a critical bottleneck, shifting the competitive focus from chip speed to electrical power capacity. Training a model like GPT-4 consumes over 391 GWh annually, and global data center electricity consumption is projected to reach up to 1,800 TWh by 2040.
