NBIM, which manages Norway’s sovereign wealth fund, has reported a productivity gain of about 20% from using Claude, Anthropic’s AI assistant. That equals an estimated 213,000 hours saved a year, and nobody got fired. The fund hit a record 20 trillion Norwegian crowns, and at that size small gains in how work gets done add up to a real advantage.
The hours came from removing friction between a question and a usable answer, without squeezing people harder. The same team produces more, and the business case rests on keeping that team.
Research beyond this case finds the same kind of gain. In one study, access to an AI assistant cut the time tasks took and improved the quality of the output. A working paper found measurable productivity gains after an AI assistant was introduced in customer support. In both, the gains came from people doing more with the tool.
Leaders who treat AI as a reason for layoffs do the opposite. They cut headcount first and hope output holds. That usually destroys the very capability needed to turn saved time into better decisions, so the organization gets leaner, then slower. It also has less slack to absorb shocks or pursue new opportunities.
The alternative is to expand what people can do, and the logic is arithmetic. Cuts stop at zero, and every cut shrinks the room to invest, because the savings arrive with lost capacity and lost learning. Expansion has no such wall, so capacity can keep compounding through better tools and better people.
That is why keeping good talent matters in an AI rollout. The upside lies in what the team does with the time it gets back, such as covering more ground, analyzing faster and clearing bottlenecks in approvals and follow-through. Counting the gain in hours, as NBIM’s figure does, lets the business track where that time goes, instead of leaving it as a vague claim of efficiency. An intact team can move those hours to new priorities as they shift, and it gets faster as it learns the tool. It also keeps the people who can spot the tool’s mistakes.
