UBER, MICROSOFT MOVES SIGNAL NEW PHASE IN ENTERPRISE AI ADOPTION

Expert commentary by Andreas Hassellöf, CEO of Ombori, on how enterprises are turning AI investment into measurable operational value and shifting from experimentation to disciplined adoption centred on workflows, governance, and business outcomes.

Large enterprises are beginning to speak more openly about the growing gap between AI adoption and measurable business outcomes, as companies reassess whether rising AI costs are translating into meaningful productivity gains.

Uber President and COO Andrew Macdonald recently said the company is finding it “harder to justify” increasing AI spending after internal discussions highlighted the difficulty of linking higher usage of AI coding tools such as Claude Code to a proportional increase in useful consumer-facing features. The comments followed reports that Uber had exhausted its 2026 budget for Claude Code within the first four months of the year, while CEO Dara Khosrowshahi confirmed the company is slowing hiring as it increases investment in AI initiatives.

At the same time, Microsoft has reportedly begun reducing internal use of Anthropic’s Claude Code within parts of its business, shifting developers toward GitHub Copilot CLI instead. Reports suggested the move was tied to Microsoft’s broader push toward its own AI ecosystem and internal tooling strategy rather than a retreat from AI adoption itself.

The developments have triggered wider debate around whether enterprises are entering a more measured phase of AI adoption, with greater focus on operational value, integration, and cost management rather than usage alone.

However, Andreas Hassellöf, CEO of Ombori, believes the issue is less about the capability of AI and more about how organisations are adapting to it.

“The real challenge has nothing to do with whether AI can increase productivity. It clearly can,” Hassellöf said. “The harder part is getting people and organisations to adapt how they actually work so the technology delivers results.”

According to Hassellöf, many companies are seeing high adoption rates and surging token consumption but are struggling to convert that activity into measurable business value. “The bottleneck is rarely the technology itself,” he said. “It is how teams change their processes, measure real outcomes, and build new habits around the tools.”

He added that the industry is now entering a more mature phase of enterprise AI adoption, where businesses are beginning to move beyond experimentation and focus instead on operational discipline, governance, and measurable outcomes. Companies that succeed, he said, will be the ones that redesign workflows around AI rather than simply layering tools onto existing processes.

“Just chatting casually with an AI coding tool and expecting it to handle everything is not enough,” Hassellöf said. “It wastes tokens and often creates more problems than it solves.”

Instead, he argues that successful AI implementation requires structured workflows where multiple AI agents handle specialised tasks such as coding, reviewing, testing, and formatting, while humans remain responsible for setting goals, reviewing outputs, and ensuring alignment with business outcomes.

“The technology is powerful, but the human side of adoption will decide whether a company succeeds with AI or whether it becomes just another expensive experiment,” he said.