In addition to Avenir, Linse Capital, Hearst Ventures and Lightbank also participated in the round, betting on the deterministic opportunity for machine vision to replace human eyes in supply chain bottlenecks. Compared to just 3 customers during its Series A in 2022, Kargo now serves 45 Fortune 500 companies, with annual revenue tripling, validating the replicability of its 'hardware-software integrated + per-pallet billing' business model.
The company calls its hardware 'AI camera towers' – essentially a data collection hub that packs an Nvidia Jetson edge GPU, self-developed deep learning models, and a 5G backhaul module into a waterproof tower body. The tower performs damage detection, label reading, and dwell timing in 12 milliseconds, writing structured logs directly into the customer’s ERP system. Each tower saves the equivalent of 4.6 full-time inspectors, with the ROI cycle compressed to 9 months for partners like Walmart.
Scale effects are driving chip selection upgrades: Kargo’s new-generation tower has adopted Ambarella’s CV3-AD SoC, which natively supports Transformer detection heads, delivering 5x edge compute performance over the previous generation while reducing power consumption by 18%, reserving computing redundancy for expansion into high-frame-rate cold chain scenarios in 2026.
The company revealed that 70% of the Series B funds will go toward the 'Kargo Intelligence' agentic AI platform, aiming to automatically map physical evidence streams collected by vision towers into financial flows – by comparing site photos, carrier EDI, and invoice PDFs, the system completes damage assessment, claim amount calculation, and GL entry within 30 seconds, estimated to save $25 million in administrative costs annually for a single distribution center.
From an industry perspective, warehousing accounts for nearly 30% of global logistics costs, yet visual automation penetration is still below 10%. Kargo’s 'hardware first, SaaS later' approach uses edge computing for front-end data cleaning, then feeds high-value structured data to cloud large models, reducing network bandwidth costs and alleviating queue delays caused by cloud GPU shortages, offering a 'light-heavy combination' paradigm for similar AI companies.