Ramp’s June 2026 software vendor report is out, and the “Trending” list — defined as breakout growth relative to company size — reads like a map of where business AI spending is actually accelerating, not just where the headlines are. Three categories dominate: foundational models, model serving and inference infrastructure, and SEO tooling. Here’s what the list tells us.

DeepSeek Is Still Growing — And That Should Worry Western AI Labs

Leading the trending list is DeepSeek, in the Foundational LLMs category. That’s notable. DeepSeek’s R1 model stunned the AI world when it dropped in early 2025, matching OpenAI’s best models at a fraction of the training cost. Since then, the Chinese AI lab has kept pushing — DeepSeek V4-Pro is now competitive with GPT-5.4 and Claude Opus 4.6 on most benchmarks, and is open-sourced under permissive licensing. The fact that it’s trending on Ramp’s spend data in mid-2026 — not just talked about, but paid for — signals that businesses are actively routing workloads to DeepSeek’s API. That’s a commercial foothold, not just a benchmark trophy.

Model Serving Infrastructure Is Having Its Moment

Three of the ten trending products — Fireworks AI, fal AI, and DeepInfra — sit in the Model Serving & Inference category. This isn’t a coincidence. As AI development matures, the bottleneck has shifted from model access to model deployment. Teams that want to run open-source models (like DeepSeek, Llama, or Mistral variants) at scale need infrastructure layers that handle latency, throughput, and cost optimization. Fireworks AI, fal AI, and DeepInfra all compete in this space, offering managed inference endpoints that let developers skip the complexity of self-hosting. The fact that all three are trending simultaneously points to a broader infrastructure buildout happening inside mid-market and enterprise tech teams right now.

Vast.ai: The GPU Cloud for Builders Who Can’t Afford the Big Guys

Vast.ai trending in the GPU Cloud category is a signal about cost pressure. AWS, GCP, and Azure GPU instances are expensive. Vast.ai operates a peer-to-peer GPU marketplace where individual owners and smaller data centers rent out compute — often at a fraction of hyperscaler prices. The fact that it’s growing fast on Ramp’s business spend data suggests companies are optimizing their AI infrastructure spend, hunting for cheaper compute wherever they can find it.

PheedLoop and Marpipe: AI Eating Niche Verticals

PheedLoop (Event Management) and Marpipe (Ad Creation) represent AI’s continued move into domain-specific workflows. PheedLoop has been applying AI to event operations — scheduling, attendee engagement, and logistics. Marpipe builds tools for automated ad creative testing and generation, a space that was largely manual until recently. Their presence on this list suggests that AI-native tools are winning share from legacy software in verticals that weren’t previously considered “AI-first.”

Paper in Software Design

Paper, listed under Software Design, is a quieter name on this list — but its trending status points to the growing adoption of AI-assisted design workflows at the SMB level, where teams want capabilities without the complexity (or price tag) of larger platforms.

SEO Tooling Gets an AI Upgrade

Both Sitebulb and DataForSEO appear in the trending list under SEO. The timing tracks: as AI-powered search engines like Perplexity reshape how users find information online, businesses are scrambling to understand how their visibility is changing. Sitebulb is a technical SEO crawler; DataForSEO provides raw SEO data via API. Both are infrastructure-layer tools that agencies and in-house teams use to run audits and track rankings at scale. The spike in spend likely reflects a broader anxiety about AI search disruption — and an arms race to stay visible in it.

What This List Tells Us

The Ramp trending data captures something that VC funding rounds and press releases can’t: where actual dollars are moving inside real businesses. The June 2026 picture shows three things clearly. First, open-source and Chinese AI models are becoming a genuine commercial force, not just a research curiosity. Second, the inference infrastructure layer is attracting serious spend as teams move from experimenting with AI to running it in production. And third, AI is permeating niche verticals — events, ad creative, SEO — that aren’t often discussed in the context of the AI race, but where the business impact is already material.