AI Labs Ranked by Valuation (Mid-2026)
A snapshot of the foundation model and AI research lab landscape, sorted by reported valuation. Includes active labs, stealth labs, and recent exits.
Tier 1: $100B+ Valuations
| Rank | Lab | Valuation | Status | Key Backers |
|---|---|---|---|---|
| 1 | OpenAI | ~$500B (some reports $850B) | Active, market leader | Microsoft, SoftBank, NVIDIA |
| 2 | Anthropic | ~$350-400B | Active, enterprise leader | Amazon, Google |
| 3 | xAI | ~$200B (pre-merger) → absorbed into SpaceX at $1.25T combined | Merged with SpaceX Feb 2026 | Musk, sovereign funds |
Tier 2: $10B-$100B Valuations
| Rank | Lab | Valuation | Status | Notes |
|---|---|---|---|---|
| 4 | Thinking Machines Lab | ~$50B (late 2025/early 2026 reported raise) | Semi-stealth, early products | Mira Murati, ex-OpenAI talent |
| 5 | Safe Superintelligence (SSI) | ~$32B (April 2025), reportedly raising at $50B+ | Stealth, no product | Sutskever, Daniel Gross |
| 6 | Mistral AI | ~$11.7B (mid-2025) | Active, European leader | French/EU government, Microsoft |
Tier 3: $1B-$10B Valuations
| Rank | Lab | Valuation | Status | Notes |
|---|---|---|---|---|
| 7 | Reflection AI | ~$8B (late 2025) | Active, pivoted to open-weights | NVIDIA, Sequoia |
| 8 | Cohere | ~$5.5B (mid-2024) | Active, enterprise focus | NVIDIA, Cisco |
| 9 | World Labs (Fei-Fei Li) | ~$4-5B (estimated, late 2025) | Active, spatial AI / Marble | a16z, Radical, NEA |
| 10 | DeepSeek (High-Flyer subsidiary) | Unclear/private — parent ~$8B+ | Active, frontier open-weights | Chinese hedge fund parent |
| 11 | Moonshot AI (Kimi) | ~$3.3B+ (2024) | Active, China frontier | Alibaba, HongShan |
| 12 | Zhipu AI | ~$3B (2024) | Active, China frontier | Alibaba, Tencent |
| 13 | Baichuan AI | ~$2.7B (2024) | Active, China | Tencent, Alibaba |
| 14 | 01.AI (Kai-Fu Lee) | ~$2.5B (2024) | Active | Alibaba, Sinovation |
| 15 | MiniMax | ~$2.5B (2024) | Active, China consumer | Alibaba, Tencent |
| 16 | Stepfun | ~$2B (2024) | Active, China | Tencent, others |
| 17 | Liquid AI | ~$2B (late 2024) | Active, alternative architecture | AMD, MIT-linked |
| 18 | Sakana AI | ~$1.5B (Japan) | Active, regional | NEA, Lux, Khosla |
| 19 | Magic | ~$1.5B (mid-2024) | Status unclear, gone quiet | Eric Schmidt, Friedman, Gross |
| 20 | Periodic Labs | ~$1.5B (late 2025 seed) | Stealth, scientific AI | a16z, Felicis, NVIDIA |
| 21 | AI21 Labs | ~$1.4B (2023, likely lower now) | Active but struggling | NVIDIA, Google |
| 22 | Black Forest Labs (FLUX) | ~$1B (mid-2025) | Active, image gen | a16z, ex-Stability team |
Tier 4: Sub-$1B / Undisclosed
| Rank | Lab | Valuation | Status | Notes |
|---|---|---|---|---|
| 23 | Imbue | ~$1B (2023, valuation likely down) | Status unclear | Astera Institute, founders fund |
| 24 | Inception Labs (Mercury) | Undisclosed, est. $200-500M | Active, diffusion LLM | Multiple seed investors |
Recent Exits (Acquired / Absorbed)
| Lab | Outcome | Acquirer |
|---|---|---|
| Inflection AI | Absorbed (2024) | Microsoft (founders + IP) |
| Adept | Absorbed (2024) | Amazon |
| Character.AI | Absorbed (2024) | |
| Stability AI | Bankruptcy/restructured (2024) | — |
| Reka | Reportedly struggling | — |
Hyperscaler Internal AI Labs
These don't trade as standalone companies but matter for competitive landscape. If valued separately, several would rank in tier 1.
| Lab | Parent | Implied Standalone Value |
|---|---|---|
| Google DeepMind | Alphabet ($2T+) | $500B-$1T (estimate) |
| Meta AI / FAIR | Meta ($1.7T+) | $300-500B (estimate) |
| Amazon AGI / Nova | Amazon ($2T+) | $100-200B (estimate) |
| Microsoft AI | Microsoft ($3T+) | Embedded in OpenAI partnership |
| Apple AI | Apple ($3T+) | Undisclosed |
Observations on the Distribution
The top three are worth more than the next 30 combined. OpenAI, Anthropic, and xAI/SpaceX collectively command ~$2T+ in implied AI lab value. Every other AI lab on this list adds up to less than half that. This concentration suggests either winner-take-most expectations are fully priced in, or the frontier requires capital scale most second-tier labs can't match.
Stealth labs are pricing thesis, not revenue. SSI ($32B+) and Thinking Machines ($50B) have minimal or no products. Investors are pricing the founding teams' expected future value. These are the purest binary bets on the list — either dramatically right or dramatically wrong.
Chinese cohort is structurally undervalued. DeepSeek, Alibaba's Qwen team, Moonshot, Zhipu, Baichuan, MiniMax operate at frontier or near-frontier capability but at fractional Western valuations. Whether this gap closes depends on geopolitical fragmentation vs. integration of AI markets.
The specialized lane bet. World Labs (spatial), Periodic (scientific), Sakana (Japan-sovereign), Liquid (alternative architecture), Black Forest (image) are explicitly not competing at the language model frontier. They're betting on vertical depth in categories frontier labs may not fully capture.
Important Caveats
Valuations move fast. SSI went from $5B (June 2024) → $32B (April 2025) → reportedly $50B+ (late 2025) in 18 months. Any list older than 3-6 months is structurally wrong.
Private market valuations vary by source. Last primary round vs. secondary trades vs. reported-but-not-closed rounds give different numbers. Where possible, the table uses last completed primary round.
Chinese lab valuations are particularly opaque. Most numbers come from Chinese press conversions and may differ from real picture.
Stealth labs by definition aren't on this list. SSI and Periodic make it because they've had public funding rounds. There may be 5-10 well-funded stealth labs operating below public visibility.
"AI lab" definition excludes: pure application companies (Cursor, Lovable, Replit), AI infrastructure (CoreWeave, Lambda), vertical AI (medical/legal specialists), and hyperscaler internal divisions (DeepMind, FAIR).
What to Watch Next 12-24 Months
- Top 3 stay or grow: OpenAI, Anthropic, xAI/SpaceX likely consolidate position
- Tier 2 consolidation: 2-3 of (SSI, Thinking Machines, Reflection, Cohere, Mistral) reach tier 1 scale; others stagnate or get acquired
- Chinese consolidation: From ~6 frontier labs down to 2-3 national champions
- Tier 3-4 attrition: Most get acquired or fold; a few specialized winners emerge
- SpaceX IPO repricing: Expected June-July 2026 IPO at $1.5-1.75T will re-rate the entire private AI comp set
No redactions applied.
Published via wiis. Originals are encrypted at rest; this page reflects the redacted version. report abuse.
Discussion
No comments yet.