Last month, investors ramped up their funding for artificial intelligence infrastructure in Sweden. This shift followed a warning from the government’s AI Commission, which highlighted that insufficient private investment was slowing down the country’s scientific and economic progress.
In just two weeks at the end of May, Sweden, lagging behind in AI and supercomputing compared to other developed European nations, secured a €10 billion investment for an AI data center. They also launched a £23 million procurement for one of 13 public AI factories in Europe aiming to compete with U.S. corporate AI. Plus, there’s a homegrown initiative to develop an AI supercomputer powered by a consortium of major corporations.
AI Sweden, a state-backed organization, expressed relief at the increased private investment. The AI Commission stated that Sweden must act fast to build a competitive AI sector. Martin Svensson, managing director of AI Sweden, emphasized the need for public funding to create AI compute facilities accessible to small businesses, not just large corporations.
The commission also urged a reevaluation of Europe’s GDPR data protection rules, as innovators struggle with data sharing regulations. Many Swedish AI developers have turned to foreign generative AI models due to restrictive copyright laws.
“Sweden is caught in a dilemma—dependent on foreign tech while barred from developing its own models,” the commission noted.
AI Sweden is working on OpenEuroLLM, a multilingual open-source generative AI, stressing the importance of establishing native capabilities to participate in the global AI race.
Meanwhile, the influential Wallenberg family announced plans for Sweden’s largest AI factory, with contributions from major companies like AstraZeneca, Saab, Ericsson, and SEB bank. Nvidia’s CEO Jenson Huang flew out to unveil the project and received an honorary doctorate from Linköping, a university involved in EU-funded supercomputing initiatives.
“No company should outsource their intelligence. No country should,” Huang said during his visit, emphasizing the importance of data ownership.
However, neither the Wallenberg family nor the corporations would disclose specifics about their investment, including access for smaller firms.
In a related development, the Canadian investment fund Brookfield announced it too would build Sweden’s largest AI factory, a €10 billion 750 MW data center, reportedly negotiating with U.S. hyperscalers.
In February, Brookfield allocated €20 billion for AI investments in France, aiming to establish Europe’s largest factory. Nvidia claimed that around 100 AI factories powered by its specialized chips are being constructed globally.
In another effort, the EU has earmarked €30 million for Sweden’s Mimer supercomputer project, which will join 13 others across Europe to bolster the continent’s AI capabilities against U.S. firms.
Despite being among the last to invest in EuroHPC supercomputers, Sweden recently committed to funding its first with the EU, although it still lacks a top-tier supercomputer like Finland’s world-ranking Lumi.
Karim Nouira, CEO of Superintelligence Computing Systems, urged Europe to invest in groundbreaking projects rather than following U.S. trends. His company secured part of a €387 million funding pool from the European Innovation Council designed to encourage investments in high-risk tech ventures.
“Europe is mostly producing more of the same AI models that already exist,” he said, noting the challenges faced by startups in Sweden. Despite reluctance from local investors to support deep-tech ventures, he recognized contributions from Sweden’s innovation agency that jumpstarted his company.
Jim Dowling, CEO of Hopsworks, echoed similar sentiments about the need for more substantial venture support, stating that most funding in Sweden is directed toward companies building on existing models rather than pioneering new technology.
He highlighted the vibrant startup scene in Stockholm, which ranks second in the world for producing unicorns per capita. The Wallenberg AI, Autonomous Systems and Software Program has also fostered local talent since 2017, addressing the knowledge gap in AI research.
“WASP has been crucial in building a talent pool,” he said. “Sweden has significantly narrowed its gap with other nations in AI research.”
Marcus Wallenberg remarked that Sweden’s reliance on innovation for exports emphasizes the necessity for sovereign computing, especially as data is poised to become a valuable commodity.