Norges Bank Investment Management ( NBIM ), manager of Norway's US$2 trillion sovereign wealth fund, has pursued an ambitious journey to boost productivity by 20% through the use of artificial intelligence ( AI ), a goal its CEO set three years ago.
Oscar Hjelde, a machine learning and artificial intelligence ( AI ) engineer at NBIM, delved into this topic at this year’s Super AI event in Singapore, offering a rare glimpse into how one of the world's largest investors is harnessing AI in the financial space.
The fund, which grew from an initial US$200 million investment of Norwegian oil and gas revenue 30 years ago, now owns an average of 1.5% of all listed companies globally.
For a firm with a portfolio of approximately 7,000 companies across 60 countries and a lean team of 700 employees, efficiency is paramount. "We need to have a huge focus on being more productive," Hjelde stresses. "And we do so through the use of AI."
Support from top leadership
Hjelde outlines a multi-faceted strategy that began with securing strong buy-in from senior leadership. This was followed by a comprehensive, organization-wide push to demystify, understand, and eventually integrate AI. The fund also structured a series of "AI Year" events and mandatory upskilling sessions across its international offices, including Singapore.
A key pillar of this strategy is the "AI Ambassadors" programme. This internal network consists of employees from various departments, such as legal, who receive specialized AI training from partners like Anthropic.
"Their job is, in addition to their day-to-day work, to take the initiative to drive transformation throughout the organization," Hjelde explains. These ambassadors are equipped to identify and implement the most effective AI applications within their specific functions.
Complementing the ambassadors is a dedicated internal AI team of engineers who wear multiple hats: they serve as internal consultants, embed in business units as "forward-deployed engineers" on high-priority projects, and build a common platform that enables other teams to develop their own AI solutions efficiently.
NBIM's AI adoption went through the phases, Hjelde says, using the metaphor of harvesting fruit from a garden. The first phase, "democratization", involved providing tools like GitHub Copilot and Cursor to the entire organization to pick the "low-hanging fruit". The second phase, "identification", focused on prioritizing more complex, high-value projects at the top of the tree, which were tackled by dedicated teams.
Competitive advantage
He highlights two significant projects. The first, nicknamed the "bad apples project", is an agentic AI system designed to screen the fund's 7,000 portfolio companies for qualitative risks like supply-chain disruption or reputational damage.
The system analyzes vast amounts of data, including vendor data, news, government filings, and local media in multiple languages. It flags potential risks for human analysts to then investigate further, enabling "more information, faster analysis, and better decision-making".
The second project, the Company Meeting Assistant and Simulation ( CMAS ), aims to streamline preparation for the roughly 3,200 annual meetings between portfolio managers and company executives.
The tool automatically drafts meeting agendas by pulling information from investment papers, past meeting notes, and recent company news, saving managers a significant portion of the three hours they typically spend on preparation for each meeting.
Hjelde shares key lessons NBIM has learned from the exercise, including a shift to smaller, more agile teams for development and the importance of building for the AI models of the future.
He notes that while the AI models themselves are a commodity, the unique business context is what provides a competitive edge.
For NBIM, the future of AI is not simply about adopting the latest models. It is about creating the organizational knowledge, data foundations, and operational discipline needed to turn rapidly evolving AI technologies into a sustainable advantage.