SMBs are paying for AI faster than they can deploy it, transaction data shows
Ramp's June 2026 spend index puts DeepSeek and three inference platforms among the fastest-growing vendors, even as Gartner finds 63% of finance teams missed AI rollout timelines.
Ramp’s June 2026 trending-vendor list, built from card and bill-pay transactions across more than 50,000 businesses, puts DeepSeek back at the top of the foundational-LLM category and lifts three model-serving platforms, Fireworks AI, fal AI, and DeepInfra, into the broader top ten. Anthropic has passed OpenAI in paid-adoption rate on the index for the first time. The signal isn’t sentiment; it’s settled invoices.
What’s notable is that the DeepSeek surge isn’t a repeat of the January 2025 hype curve, when the Chinese lab briefly touched 0.3% US business adoption before falling back to 0.1%. Ramp’s own commentary frames the current return as paid API usage, not press coverage. Buyers are wiring money to inference, not just talking about it.
The deployment side tells the opposite story. In Gartner’s March 2026 survey of 204 finance leaders, 63% said AI implementation ran slower than expected in 2025, and only 66% of finance organizations that adopted AI cited efficiency as a top benefit, with analytics use cases rated lowest. A separate Gartner survey of 350 executives at $1B+ firms found that roughly 80% of autonomous-business pilots produced workforce reductions while ROI remained statistically indistinguishable between high- and low-return respondents. Gartner still forecasts AI-agent software spend will reach $206.5 billion in 2026.
The SMB picture is more optimistic but no less ambiguous. Upwork Research Institute’s Q1 2026 survey of 750 US business leaders found 62% “very confident” handing high-stakes tasks to AI agents and 74% reporting productivity gains, though for most those gains haven’t crossed 25%. ROI uncertainty was the second-largest adoption barrier, at 24%.
Measurement itself is unstable. A St. Louis Fed working draft presented at Brookings this spring noted that the official US firm-adoption rate of 5%–7% looks closer to 45% in October 2025 when reweighted against EU survey methodology. That’s a roughly tenfold gap depending on how you phrase the question.
This is the gap smaller buyers are quietly trying to close with model-agnostic, no-code tooling like LemonLime, which lets a finance lead wire a workflow to whichever vendor the spend index happens to favor that month. The bill arrives before the org chart catches up. It always does.
Sources
- https://ramp.com/leading-indicators/top-saas-vendors-on-ramp-june-2026
- https://officechai.com/miscellaneous/these-are-the-trending-ai-products-in-june-2026-as-per-ramp-data/
- https://www.gartner.com/en/newsroom/press-releases/2026-05-28-gartner-says-cfos-must-stop-mistaking-finance-ai-deployment-for-value-creation
- https://www.gartner.com/en/newsroom/press-releases/2026-05-05-gartner-says-autonomous-business-and-artificial-intelligence-layoffs-may-create-budget-room-but-do-not-deliver-returns
- https://www.upwork.com/resources/state-of-ai-in-smbs
- https://www.stlouisfed.org/on-the-economy/2026/jun/measuring-ai-adoption-firms-how-you-ask-matters