Uber president says AI spending is getting ‘harder to justify’
Original reporting by The Verge
Uber, a company often at the forefront of technological adoption, is reportedly confronting a sobering reality: its annual AI budget for 2026 was exhausted a mere four months into the year. This aggressive spending, however, hasn't translated into clear, quantifiable returns, prompting a critical re-evaluation of its artificial intelligence investments. According to Uber president and chief operating officer Andrew Macdonald, the link between soaring token consumption, particularly for generative AI models like Claude Code, and the delivery of more useful features to consumers remains elusive.
The Cost-Benefit Question
Macdonald candidly admitted in an interview that it's challenging to "draw a line between one of those stats and, ‘Okay, now we’re actually producing 25 percent more useful consumer features.’" This sentiment emerges against a backdrop of significant R&D expenditures, which hit $3.4 billion in 2025, a 9 percent increase from the previous year. To manage these escalating costs, CEO Dara Khosrowshahi previously indicated that Uber would compensate by hiring fewer human employees. Yet, as Macdonald underscores, the trade-off between "token consumption and the associated cost versus headcount" becomes increasingly difficult to justify without a direct connection to enhanced user functionality. Uber's predicament highlights a growing challenge for companies pouring resources into AI: demonstrating a tangible return on investment beyond abstract metrics.
Uber's candid introspection regarding its significant AI investments and the elusive link to tangible user-facing features underscores a critical juncture for the entire technology industry. While the company has poured billions into R&D and reduced human hiring in anticipation of AI dividends, the frank admission from President and COO Andrew Macdonald highlights a fundamental challenge: translating astronomical token consumption into demonstrable value. This isn't merely an internal Uber dilemma; it reflects a broader reckoning facing countless organizations that have embraced AI with fervor, often driven by competitive pressure or the promise of unprecedented efficiency, without fully establishing clear, measurable returns.
A Shift in Focus
The "cost versus headcount" equation, as Macdonald articulated, is becoming increasingly difficult to justify without a direct line to enhanced functionality. Moving forward, Uber’s experience could well serve as a bellwether. We are likely to see a marked shift across the corporate landscape from an unbridled "AI at all costs" mentality to a more discerning, ROI-centric approach. Companies will be compelled to develop more robust frameworks for evaluating AI initiatives, demanding transparency on token consumption, and establishing concrete metrics for feature delivery and user adoption. The honeymoon phase of AI investment may be drawing to a close, ushering in an era where strategic deployment and verifiable impact, rather than sheer expenditure, will dictate success. This evolution will ultimately foster more sustainable and impactful AI integration across industries.