Uber has increased its AI agent workload by 9.4 times since February while keeping infrastructure spending flat. This improvement, achieved through smarter computing and strict budget controls, marks a shift from earlier in the year when spending limits were tested. Investors are now tracking how this operational efficiency helps manage current margin pressures and regulatory risks.
Uber has successfully scaled its reliance on artificial intelligence without adding to its infrastructure bill. Since February 2026, the volume of tasks processed by AI agents has grown 9.4 times, yet the company has kept its total AI-related spending flat since April. This represents a significant operational shift, especially given that many global tech firms are seeing costs rise in lockstep with their AI adoption.
The company achieved these savings by changing how it manages computing power. Instead of using high-performance models for every task, Uber now directs simpler requests to more affordable, specialized models. It also implemented stricter budget management, including a cap of 400,000 tokens per interactive session to prevent runaway usage. Additionally, the company improved its 'prompt caching'—a method of reusing previously processed data—by extending the duration from five minutes to one hour. This reduces the need for the system to re-calculate common requests, driving the cost per session down by 52% since June.
This efficiency drive follows a challenging start to the year. Earlier in 2026, reports indicated that Uber had exhausted its annual AI coding budget in just four months due to rapid, unchecked adoption. To fix this, the company introduced specific caps, limiting spending to $1,500 per employee per month for certain AI tools. The current success in controlling these costs suggests that management has successfully moved from a phase of unrestricted experimentation to a more disciplined, performance-oriented model.
While this operational efficiency is a positive for the company's cost structure, investors are monitoring several broader financial and regulatory risks. Uber’s second-quarter 2026 results showed net profit margins of 17.3%, down from 26.7% in the same period a year earlier. This margin compression, combined with stiff competition in autonomous vehicle technology from rivals like Waymo and Tesla, keeps pressure on the bottom line.
Furthermore, the company faces significant regulatory hurdles. A major concern remains a proposed €825 million (approximately $966 million) fine from the Dutch data-protection authority related to its automated driver suspension systems. As Uber pursues inorganic growth—evidenced by its recent offer document to acquire Delivery Hero SE—the ability to maintain tight control over operating expenses, particularly in AI infrastructure, remains a critical monitorable. The company’s next few quarters will reveal whether these AI cost-saving measures can consistently help improve profitability amidst these sector and regulatory headwinds.