DVC Launches Live State of AI Report — A Continuously Updated View of the AI Economy

DVC, a San Francisco–based VC focused on AI startups, led by Marina Davidova, Nick Davidov, and the team, has released its State of AI Report — a living, continuously updated analysis of the artificial intelligence landscape. According to the report, the market is mispricing AI.

The State of AI report is based on a framework developed by DVC co-founders Nick Davidov and Marina Davidova over more than a decade of AI investing. It combines meta-analysis of leading industry research with proprietary validation and insights from a community of 200+ operators and investors.

Unlike traditional VC reports, DVC’s State of AI is a live product, continuously updated to reflect changes across the AI ecosystem, from infrastructure and foundation models to application-layer shifts and emerging business models. Updates are generated with AI support and manually reviewed and validated by DVC co-founder and managing partner Nick Davidov. This format aligns with the pace of the AI market, where static snapshots quickly become outdated.

One of the report’s key insights examines how AI companies are evaluated. Traditional SaaS models assume high long-term margins, often up to ~80%, driven by scalable software economics. AI startups operate under a different structure.

A significant portion of revenue is absorbed by an ongoing “inference tax”, typically 40–60%, covering compute, foundation models, cloud, and energy costs. As a result, many AI businesses resemble industrial operations more than classic SaaS, with thinner margins and heavier dependence on scale.

Value is therefore shifting away from pure infrastructure layers and toward distribution, interface ownership, control over user behavior, and speed of market expansion, as upstream components increasingly commoditize.

“NVIDIA’s Jensen Huang has noted that a datacenter should be thought of as a modern-day factory. In that framing, an AI agent for banking AML and compliance is less like SaaS and more like a silverware customization service trying to build a brand. The spoons are mass-produced in factories from sheet metal sourced through industrial supply chains, metal and electricity go in, and utility (and user experience) comes out.

That said, applying SaaS-like valuation multiples to application-layer or agentic AI companies is not necessarily incorrect today. However, the basis is different: it is less about long-term terminal margins and more about significantly higher revenue growth potential.”

At the same time, the report highlights an unprecedented dynamic: small teams scaling to hundreds of millions in revenue and directly challenging incumbents at scale. Companies like Perplexity AI, adding 50% to $300M ARR in a month, or Higgsfield growing from zero to $300M in a year, illustrate this shift.

The full DVC State of AI report is available here: https://state-of-ai.dvc.ai/

Alongside the report, DVC is announcing The AI Rabbit Hole: AI Bubble Tea Party, an invite-only conference on June 5 in San Francisco, bringing together founders, researchers, and investors to explore the key themes shaping the AI landscape in greater depth.