Loading AI Valuations Dashboard...
The AI IPO era arrived in June 2026: Anthropic filed to go public on June 1 at a $965 billion valuation on roughly $47B in annualized revenue, and OpenAI filed confidentially days later targeting $730–850B for a September debut. The defining story of the year: Anthropic passed OpenAI in revenue in April 2026 ($30B vs $25B run-rate) after growing from ~$1B ARR in just fifteen months. Tracking 50+ AI companies valued at $300M+, updated weekly.
| Company | Valuation | ARR (est.) | Rev Multiple | Category |
|---|---|---|---|---|
| Anthropic | $965B (IPO filing) | ~$47B | ~21x | Foundation models / safety |
| OpenAI | $730–850B (IPO target) | ~$25B | ~30x | Foundation models |
| Databricks | $62B | ~$3B | ~21x | Data + AI platform |
| xAI | $50B | ~$1B | ~50x | Foundation models |
| Figure AI | $40B | ~$100M | ~400x | AI robotics |
| CoreWeave | $35B | ~$3B | ~12x | AI cloud infrastructure |
| Scale AI | $14B | ~$1B | ~14x | AI data / RLHF |
| Perplexity | $9B | ~$200M | ~45x | AI search |
| Mistral AI | $6B | ~$100M | ~60x | Open-source models |
| Cohere | $5.5B | ~$200M | ~28x | Enterprise AI |
| Harvey AI | $3B | ~$75M | ~40x | AI for legal |
| Glean | $4.6B | ~$200M | ~23x | Enterprise AI search |
| AI Category | Typical ARR Multiple | Range | Key Driver |
|---|---|---|---|
| Foundation model labs | 15–50x | Compressing as revenue scales | Revenue growth + strategic optionality |
| AI robotics | 100–400x | Extreme (pre-revenue) | Physical AI thesis, demo-driven hype |
| AI infrastructure / cloud | 10–20x | Predictable, narrowing | Contracted GPU revenue, utilization |
| AI data / RLHF | 12–20x | Maturing | Government + enterprise contracts |
| Vertical AI (legal, medical, finance) | 20–45x | High variance | Net retention, domain defensibility |
| AI application layer (copilots, agents) | 15–40x | Bifurcating | Proprietary workflow data, not just LLM wrapper |
| AI DevTools / APIs | 8–18x | Compressing further | Usage growth, switching costs |
The biggest shift in 2026: AI revenue went vertical. Anthropic grew from ~$1B to a $47B annualized run-rate in under eighteen months and passed OpenAI (~$25B) in April 2026 — the fastest revenue scaling in software history. Both are now IPO-bound at 20–30x revenue, multiples that public markets are about to stress-test for the first time.
A clear valuation gap has emerged between foundation model companies (OpenAI, Anthropic, xAI) and AI application companies built on top of them. Foundation labs capture the most value because they control the core technology. Application-layer companies need proprietary data, deep workflow integration, or unique distribution to justify premium multiples — otherwise they're just GPT wrappers.
Proprietary training data and RLHF pipelines remain the most defensible AI asset. Companies like Scale AI are valued as data infrastructure — whoever controls high-quality human feedback data controls model quality. In 2026, companies with proprietary user interaction data (Perplexity, Harvey) command premiums over those relying on public datasets.
AI gross margins have improved meaningfully in 2026 as inference costs dropped 5–10x from 2024 levels. OpenAI and Anthropic are both approaching 50–60% gross margins, up from 30–40%. Companies that demonstrate improving margins with scale are rewarded with higher multiples. The path to profitability is now visible for the first time.
Figure AI's $40B valuation on minimal revenue signals investor appetite for physical AI. Humanoid robots and AI-powered hardware represent the next wave of AI value creation. These companies trade at extreme multiples because the TAM — replacing human labor in physical tasks — is potentially larger than software AI.
Microsoft's $13B+ in OpenAI, Google's $2B+ in Anthropic, Amazon's $4B+ in Anthropic, and NVIDIA's investments across the stack continue to distort private market valuations. Big tech is essentially pre-buying AI infrastructure and optionality. In 2026, these strategic rounds are increasingly being supplemented by sovereign wealth fund capital from the Middle East and Asia.
OpenAI filed confidentially for an IPO in June 2026 with Goldman Sachs and Morgan Stanley, targeting a valuation of $730–850 billion for a September 2026 debut. Its annualized revenue is roughly $25B (up from ~$20B at the end of 2025), driven by ChatGPT subscriptions, Codex, and enterprise API revenue — implying a ~30x revenue multiple at the IPO target. The surprise of 2026: OpenAI is no longer the revenue leader. Anthropic passed it in April 2026 and filed to go public first.
Anthropic filed for its IPO on June 1, 2026 at a $965 billion valuation — the largest of the AI era — on roughly $47B in annualized revenue (~21x). The growth is historic: from about $1B ARR to a $30B run-rate in fifteen months, passing OpenAI in April 2026, then reaching ~$47B by the June filing. Claude's dominance in enterprise coding and agentic work drove the surge. Anthropic's listing in H2 2026 is the first big public-market test of AI-era valuations.
The AI valuation landscape has bifurcated in 2026. Foundation model companies (OpenAI, Anthropic) now have real revenue backing their valuations — multiples have compressed from 60–100x to 15–50x, which is elevated but not unprecedented for high-growth tech. The real bubble risk is in two areas: (1) AI robotics companies like Figure AI trading at 400x revenue on demo hype, and (2) thin AI application wrappers that lack proprietary data or distribution advantages. The healthiest sign is that investors are increasingly demanding revenue metrics, not just model benchmarks, before writing checks.
The gap is narrowing but still significant. Median public SaaS trades at 6–10x NTM revenue; top AI-native SaaS companies trade at 15–40x. Foundation model labs at 15–50x are closer to SaaS multiples than the 100x+ of 2024. The key difference: AI companies are growing revenue 3–5x faster than traditional SaaS did at similar stages. Anthropic went from ~$1B to a $47B annualized run-rate in under 18 months, and OpenAI from ~$6B to $25B over a similar window. Growth rates like that, if sustained, make current multiples look reasonable on a forward basis.
Foundation model companies (OpenAI, Anthropic, xAI, Mistral) build the core AI models and trade at 15–60x revenue. Application layer companies build products on top of these models — think Harvey for legal, Glean for enterprise search, or Perplexity for consumer search. Application companies trade at 20–45x if they have proprietary data and deep workflow integration, but much lower (8–15x) if they're essentially API wrappers. The biggest risk for application layer companies is that foundation model providers expand into their vertical.