Uber president and chief operating officer Andrew Macdonald has raised concerns that the company's massive investments in artificial intelligence are not yet yielding measurable returns, casting doubt on the prevailing narrative that AI spending automatically translates into business value. In an interview with Rapid Response, Macdonald said that despite Uber exhausting its annual AI budget just four months into 2026, there is a disconnect between rising token consumption for tools like Claude Code and the delivery of meaningful features to consumers.
"That link is not there yet, right? I think maybe implicitly there is more that is getting shipped, but it's very hard to draw a line between one of those stats and, 'Okay, now we're actually producing 25 percent more useful consumer features,'" Macdonald stated. He acknowledged that while underlying metrics such as token usage are "trending in a really astronomical direction," connecting those numbers to tangible productivity gains remains elusive. This skepticism comes at a time when tech companies globally are pouring billions into AI infrastructure, often with little evidence of a direct return on investment.
Uber spent $3.4 billion on research and development efforts in 2025, a 9 percent increase from the previous year. The company's R&D budget has grown steadily as it integrates AI into its ride-hailing, food delivery, and freight logistics platforms. Earlier this month, Uber CEO Dara Khosrowshahi indicated that the company is offsetting rising AI costs by hiring fewer human employees, a trend that has sparked debate about the trade-offs between automation and employment. "We're going to have to start talking about token consumption and the associated cost versus headcount," Macdonald said. "So if you're not actually able to draw a direct line to how much useful features and functionality you're shipping to your users, that trade becomes harder to justify."
The comments reflect a broader unease in the tech industry about the efficacy of AI spending. While companies like Microsoft, Google, and Amazon have committed hundreds of billions of dollars to AI data centers and model training, a growing number of executives are questioning whether these investments are translating into genuine competitive advantages. For Uber, the challenge is particularly acute because its core business relies on operational efficiency—matching drivers with riders, optimizing delivery routes, and predicting demand—rather than flashy AI features that can be monetized directly.
Uber's use of AI extends across multiple domains. The company employs machine learning models to set dynamic pricing, forecast rider demand, and improve driver incentives. In its Uber Eats division, AI helps personalize restaurant recommendations and streamline delivery logistics. The company has also invested heavily in autonomous vehicle technology, though its plans for self-driving cars have evolved over the years. Macdonald's remarks, however, focus on the less visible but rapidly growing cost of AI development: the consumption of tokens from large language models (LLMs) like Anthropic's Claude Code, which engineers use to assist with coding and debugging.
Token consumption refers to the usage of text-generating AI models, which charge based on the number of tokens processed. As Uber's engineering team embraces AI-assisted coding tools, the cost of these tokens has ballooned. Macdonald noted that the company's annual AI budget was exhausted by April, forcing a reassessment of priorities. This echoes a trend seen across the tech sector, where companies are struggling to balance the allure of AI productivity gains with the reality of escalating operational expenses.
The productivity paradox, first observed in the 1970s with the advent of computers, suggests that new technologies often fail to deliver immediate efficiency gains. AI appears to be following a similar pattern. While individual developers may feel more productive using AI coding assistants, measuring the overall impact on output is notoriously difficult. "The link between AI usage and productivity is not there yet," Macdonald admitted, echoing the sentiments of other tech leaders who have recently expressed caution.
For context, Uber's R&D spending has increased significantly in recent years. In 2023, the company spent approximately $3.1 billion on R&D, which rose to $3.4 billion in 2025. The growth is partly attributable to AI investments, but also to broader technological upgrades and expansion into new markets. However, the pace of spending has accelerated so quickly that it has outstripped the company's ability to demonstrate value. Macdonald's comments suggest that Uber may be reaching an inflection point where further AI investment will require stronger justification.
Industry analysts have noted that the situation at Uber is a microcosm of a larger issue. According to a recent report from the McKinsey Global Institute, only about 10 percent of companies that have deployed AI at scale report substantial revenue growth from the technology. Many firms are still in the experimental phase, investing heavily in AI infrastructure without a clear path to monetization. For Uber, which operates on thin margins in a highly competitive industry, the pressure to show returns is particularly intense.
Macdonald's interview also touched on the broader shift in how tech companies evaluate AI spending. Historically, R&D investments were seen as long-term bets, with success measured over years rather than quarters. But the current AI gold rush has created an expectation of rapid returns, leading to what some observers call "AI fatigue." As the hype cycle matures, executives are demanding more rigorous accounting of costs and benefits. "We need to start having those conversations internally," Macdonald said, referring to the need to link token consumption to feature delivery.
The implications extend beyond Uber. If one of the world's largest ride-hailing companies is questioning the value of AI spending, it may signal a broader recalibration across Silicon Valley. Venture capital funding for AI startups has already slowed from peak levels in 2024, and public market investors are increasingly scrutinizing AI-related expenditures. Companies like Microsoft and Alphabet have also faced pressure to justify their AI spending, particularly as their cloud businesses face margin compression.
Despite these concerns, Uber remains committed to AI as a long-term strategy. Macdonald emphasized that the company is not abandoning its AI efforts, but rather seeking a more disciplined approach. "We want to make sure that every dollar spent on AI is driving real value for our customers and our business," he said. This sentiment is likely to resonate with other tech leaders who are grappling with similar challenges.
In the meantime, Uber is experimenting with ways to better track the impact of AI. The company is developing internal dashboards that correlate token usage with specific feature releases and customer satisfaction metrics. It is also exploring partnerships with AI vendors to negotiate more favorable token pricing. Additionally, Uber is investing in fine-tuning smaller, specialized models that may be more cost-effective than relying on large, general-purpose LLMs.
The broader context of Macdonald's comments includes the ongoing debate about AI and employment. By hiring fewer humans to offset AI costs, Uber is participating in a trend that has raised alarm among labor advocates. A 2025 study by the Oxford Internet Institute found that companies adopting AI at scale have reduced their workforce growth by an average of 3 percent annually. For Uber, which employs over 30,000 full-time workers and millions of gig workers, these trade-offs are politically and operationally sensitive.
As the industry moves forward, the question of AI's return on investment will only become more pressing. Uber's experience suggests that even companies with deep pockets and extensive AI expertise are struggling to connect the dots between spending and outcomes. The next few quarters will be critical as Uber and its peers attempt to bridge this gap. If they succeed, the AI boom may justify its enormous costs; if not, the industry may face a reckoning.
Source: The Verge News