This blog was first published on Silicon UK.
Klarna’s bold decision to abandon SaaS giants like Salesforce and Workday signals a major shift in tech strategy, embracing DIY, AI-powered solutions over costly, generic platforms. With advances in large language models (LLMs) and GenAI, companies now have the power to build tailored, efficient systems that align perfectly with their unique goals.
News that Klarna is ditching SaaS providers Salesforce and Workday may have been greeted with scepticism by some in the tech sphere – but it also represents a significant paradigm shift.
Klarna’s CEO Sebastian Siemiatkowski recently announced that the BNPL pioneer—buoyed by a 27% YoY revenue jump—would “shut down” its reliance on two of the world’s best-known SaaS platforms. Instead, the Swedish company plans to build internally developed AI systems crafted on “more lightweight and effective” large language models (LLMs).
It may be bold, but Klarna’s move proves that SaaS solutions no longer hold the edge in a climate where proactive leaders seek to craft their own highly tailored AI-powered solutions.
Of course, the all-mighty SaaS realm won’t change overnight. The companies choosing “large internal initiatives” over outbound support may be outliers for now, but this remains a watershed moment for SaaS platforms.
Here’s why:
SaaS is too expensive: CEOs are growing wise to the huge cost of their SaaS platforms, including the initial investment and ongoing maintenance. According to recent reports, spending on SaaS tools now costs an average of $3,500 per employee. It is forecast to reach $197 billion this year, making it the company’s third biggest expense after the cost of staff and premises. While platforms like Salesforce are moving to a per-conversation rather than a per-seat basis, businesses will increasingly be weighing up the cost of SaaS versus a bespoke solution built on advanced AI foundational models and deciding that SaaS is no longer commercially viable.
Foundation models are eroding the power of SaaS: Before ChatGPT arrived two years ago, building bespoke AI tools and platforms was out of reach for many companies. But GenAI’s “secret sauce” is that it has made foundation models into accessible commodities open to whoever chooses to use them. Businesses can now deploy an LLM through their cloud provider and integrate that into their infrastructure to create bespoke customer journeys and employee experiences.
By contrast, the SaaS world is built on solutions catering to the most general use case that can be sold to many different companies. But this broad-stroke approach jars with today’s capacity to build high-level tailored and personalised AI solutions, which is now available on tap with GenAI.
Advancement in AI is moving faster than SaaS platforms can: As we have seen with the volume of new features unveiled by Google, Anthropic, Meta, Microsoft and OpenAI, the foundational models move much faster than the monolith SaaS solutions trying to capitalise on them. The SaaS platforms can’t adapt quickly enough to the new AI paradigm, a move which involves rethinking their entire proposition with an AI lens and redesigning it from the ground up in a way relevant to a whole market of players.
What’s more, with the developer role becoming more productive, engineers can use AI to write code internally and develop more software for their organisations, so there is less demand for off-the-shelf and ultimately rigid SaaS solutions.
SaaS naysayers provide even more reason to act: With Salesforce shares plunging 18% earlier this year – in a drop blamed partly on fierce competition around AI – it makes sense that CEO Marc Benioff is questioning Klarna’s move to go solo. In particular, Benioff has cast doubt on Klarna’s handling of data, compliance and governance issues. However, there is a certain irony to his stance.
When Benioff co-founded Salesforce in 1999, he was the rebel persuading others to take on a perceived data risk by migrating to the cloud. Now, the industry he gave rise to is facing its own dramatic decline in “megadeals” worth $100 million or more over the past year. In response, the sector is rushing to put a new fleet of AI agents (including Salesforce’s own “groundbreaking” suite, Agentforce) at the centre of future growth plans. But in this AI-enabled era, businesses are still far better off customising existing foundational models – including many open source options, than forming an unhealthy reliance on off-the-shelf solutions.
DIY AI has momentum. Once a few companies across different market sectors decide to reject third-party SaaS, it will trigger a domino effect. Shareholder pressure around cost savings and increased efficiencies will force businesses to question why they are paying out on expensive contracts and to review the partnerships they once assumed to be staples. Just two years on from the launch of ChatGPT, GenAI is firmly on the agenda of company boards and CFOs (note the number of GenAI mention])in company reports. This is in stark contrast to previous tech revolutions such as the Cloud where board level awareness took much longer to build.
The DIY approach to AI isn’t just about cutting costs, it’s about crafting solutions that align perfectly with business objectives and customer needs, setting the foundation for long-term innovation and growth. The more companies are willing to break free from the established status quo, the bigger the rewards will be.
