- Uber’s COO Andrew MacDonald warns that current AI spending is unsustainable due to escalating token costs.
- Generative AI tools have improved productivity in engineering, customer service, and logistics, but costs are rising faster than benefits.
- Uber’s AI token consumption surged 37% between 2024 and 2026, driven by engineering and customer support teams.
- The company spent $89 million on third-party AI APIs in 2025, with most expenses tied to OpenAI and Anthropic models.
- Uber is reevaluating its AI investment framework due to rising costs and limited headcount reduction gains.
Uber’s Chief Operating Officer Andrew MacDonald has raised concerns about the sustainability of current AI spending, stating it is becoming increasingly difficult to justify the cost of AI token usage across the company’s operations. In comments reported by Business Insider in May 2026, MacDonald highlighted that while generative AI tools have enhanced productivity in engineering, customer service, and logistics planning, the exponential rise in token consumption—particularly from large language models (LLMs)—is outpacing measurable returns. With token costs rising nearly 40% year-over-year and limited headcount reduction gains, Uber is reevaluating its AI investment framework. This development matters because it signals a maturing phase in enterprise AI adoption, where early enthusiasm is giving way to rigorous cost accountability—a trend likely to influence tech-sector spending across Silicon Valley and beyond.
AI Token Usage and Cost Escalation at Uber
According to internal Uber benchmarks cited in the report, AI token consumption across departments surged by 37% between Q2 2024 and Q1 2026, driven primarily by engineering teams using language models for code generation and customer support automating ticket responses. In 2025, Uber spent an estimated $89 million on third-party AI APIs, up from $65 million the prior year, with most expenses tied to OpenAI and Anthropic models priced per token. While some efficiency gains were documented—including a 22% reduction in average ticket resolution time and a 15% acceleration in software deployment cycles—these have not translated into proportional cost savings or headcount optimization. The company found that many AI-assisted workflows still require extensive human oversight, diminishing net productivity. As Reuters reported in early 2025, several major firms began tracking ‘AI yield per dollar’ as a new KPI, reflecting growing scrutiny over whether generative AI delivers real economic value beyond pilot projects.
Key Players and Strategic Shifts in AI Deployment
Andrew MacDonald, who joined Uber in 2021 and oversees global operations, product, and engineering, has emerged as a leading voice questioning unchecked AI spending. Working closely with CTO Srikanth Thirumalai, MacDonald has initiated a company-wide audit of AI tooling to identify redundant or low-impact use cases. Meanwhile, AI startup partners like Anthropic and Mistral are under pressure to offer more cost-efficient models tailored for enterprise workflows. Internal teams at Uber are now required to submit ‘AI justification memos’ before deploying new LLM integrations, detailing expected ROI, token budget, and fallback protocols. This shift mirrors broader moves at firms like Salesforce and Meta, where AI experimentation is transitioning from innovation labs to centralized governance models. MacDonald’s comments follow similar cautionary notes from Microsoft CFO Amy Hood, who in Q4 2025 warned investors that ‘AI spend is not automatically margin-enhancing.’
Trade-Offs Between Innovation and Financial Discipline
The dilemma Uber faces reflects a fundamental trade-off in today’s AI adoption curve: balancing innovation speed against financial sustainability. On one hand, delaying or restricting AI use risks falling behind competitors in automation, personalization, and operational agility. On the other, unchecked spending could erode margins without delivering scalable efficiencies. For Uber, the stakes are particularly high given its narrow profitability in core ride-sharing and delivery units. While AI has improved routing algorithms and fraud detection, many frontline applications remain supplementary rather than transformative. Moreover, the environmental cost of AI—measured in energy and carbon footprint from data centers—is gaining attention from ESG investors. Uber has not yet disclosed its AI-related emissions but may face regulatory pressure as the EU’s AI Act and U.S. executive orders on AI transparency take effect in 2026.
Why the Timing of This Shift Matters
The timing of MacDonald’s warning is significant: it arrives amid a broader market reassessment of AI’s economic value. After the initial post-2023 LLM boom, public and private tech companies are entering a ‘sobering phase’ where investors demand clearer metrics on AI-driven productivity. Cloud providers like AWS and Google Cloud have reported rising customer inquiries about AI cost optimization tools, while startups offering model compression and caching solutions—such as OctoAI and Together AI—are seeing increased enterprise traction. At Uber, leadership now emphasizes ‘precision AI’—targeted, high-impact deployments over broad experimentation. This marks a departure from 2024, when executives encouraged widespread AI tool adoption with minimal oversight. The shift also aligns with Wall Street’s growing skepticism toward AI spending without clear earnings impact, as reflected in P/E ratio adjustments for AI-heavy tech firms in early 2026.
Where We Go From Here
In the next 6 to 12 months, three scenarios could unfold. First, Uber may consolidate its AI vendors, negotiate bulk pricing, and shift toward fine-tuned smaller models to reduce token dependency. Second, the company could spin off its internal AI efficiency task force into a formal Center of Excellence, setting new standards for AI governance across the industry. Third, if cost pressures intensify, Uber might scale back non-core AI initiatives—such as AI-generated marketing content or experimental chatbots—redirecting funds to core logistics AI where ROI is clearer. Each path will hinge on whether Uber can demonstrate that AI is not just a productivity aid but a true profit lever. As other firms watch Uber’s approach, its next moves could set a precedent for responsible enterprise AI adoption.
Bottom line — Uber’s COO signaling difficulty in justifying AI token costs marks a pivotal moment in enterprise AI, where early adoption gives way to disciplined investment, forcing companies to prove AI’s real financial and operational value beyond hype.
Source: Businessinsider