Key Highlights

  • Nearly half of delivery operators, 45.8%, say last-mile delivery costs are increasing faster than revenue, according to a survey of more than 500 operators.
  • The median year-over-year increase in cost per delivery held steady at 12% in both 2025 and 2026, with a fifth of operators reporting increases above 20%.
  • Operators with greater control over their delivery network infrastructure and data achieved 95% on-time performance, compared with 65.5% among low-control operators at the same Investment level.
  • A new AI-powered platform developed with Nvidia claims to cut routing computation time by as much as tenfold, compressing a 20-minute problem to under two minutes.

Costs Are Rising as Fast as Revenue, or Faster

Retailers are increasingly seeking to Capitalize on same-day delivery as consumers grow accustomed to faster shipping, but the Economics of building the necessary infrastructure remain a persistent obstacle for most businesses. Cost Inflation is a top concern industry-wide, with 88% of delivery operators reporting that costs are growing at the same pace as Revenue or faster, according to a survey from FarEye, a last-mile delivery management software company. As retailers including Walmart (NYSE:WMT) and Target (NYSE:TGT) continue expanding their delivery networks, the resulting cost pressure is becoming harder for logistics professionals to absorb, with 45.8% of survey respondents saying delivery costs are rising faster than revenue and another 42.2% saying the two are increasing at the same rate.

The survey, which polled more than 500 delivery operators, found the median year-over-year increase in cost per delivery held at 12% in 2026, matching the same 12% increase recorded in 2025. While 60% of respondents reported cost increases above 10%, a fifth reported increases exceeding 20%, underscoring the difficult balancing act facing logistics executives as they try to grow revenue and scale service expectations without allowing operating costs to climb in lockstep.

Orchestration Emerges as the Key Differentiator

Kushal Nahata, chief executive and co-founder of FarEye, said retailers and logistics providers are managing an increasingly complex mix of carriers, owned fleets and regional partners while trying to protect margins, and that the research makes clear investment alone does not create control. He argued the organizations succeeding at this are those able to translate visibility, governance and orchestration, the automated coordination of tasks across multiple systems, into consistent execution across their networks, which he described as ultimately determining whether a delivery promise can be kept at a sustainable cost.

Last-mile delivery software provider OneRail is attempting to address this challenge directly with the launch of OmniStar, an AI-powered delivery decisioning platform designed to help enterprise retailers, wholesalers and distributors rapidly evaluate delivery Options and select the lowest-cost choice that still meets service requirements. The company argues this approach can give retailers a path to offer fast delivery without absorbing the Margin tradeoffs that have historically limited same-day delivery at scale. The FarEye survey found the share of operators implementing or operating AI in their delivery operations rose to 66.3% in 2026 from 46.2% the prior year.

A Computational Approach to an Old Problem

OneRail, which powers two-day and end-of-day delivery services for FedEx (NYSE:FDX), partnered with Nvidia (NASDAQ:NVDA) to develop the OmniStar platform, framing current last-mile challenges as fundamentally computational, since delivery routing and mode selection often involve far more possible combinations than conventional systems can evaluate quickly given constantly changing fuel costs, weather patterns and shipping routes. As a result, the companies say, retailers frequently operate with incomplete information and absorb costs that could otherwise be avoided. According to the partners, Nvidia's AI infrastructure allows OmniStar to process routing and mode selection scenarios up to ten times faster than traditional systems, compressing a 20-minute computation to under two minutes and a weeklong calculation to roughly two days.

Speed Alone Does Not Guarantee Better Service

The FarEye survey found that operators with greater control over their last-mile network's infrastructure and data achieved 95% on-time performance, compared with just 65.5% among low-control operators, despite similar median investment levels. Consumer expectations around speed remain intense: as many as 70% of consumers said they would forgo discounts this holiday season to receive packages on time, according to a separate survey from Radial, and 78% said fastest shipping is the top priority when comparing otherwise similar products.

Yet retailers should be cautious about assuming that maximizing delivery speed alone strengthens the customer proposition. The FarEye survey found that operators prioritizing the fastest possible delivery reported only 76% on-time performance alongside 24% median cost inflation, while those prioritizing predictable delivery instead achieved 88.4% on-time performance with just 10% cost inflation, and operators focused on real-time tracking visibility performed best of all, at 90.3% on-time with only 4.9% cost inflation.

Conclusion

The data suggests that the retailers succeeding in the same-day delivery race are not necessarily those investing the most or moving the fastest, but those building the orchestration and visibility needed to make consistent, cost-effective decisions across an increasingly complex delivery network. Whether AI-driven platforms like OmniStar can meaningfully close the gap between high-control and low-control operators at scale will likely determine how much of the current cost pressure retailers can offset as same-day delivery Demand continues to grow heading into the holiday season.