AI21 Funding Rounds, Valuation, and the NVIDIA Endgame
| Round (date, lead) | Amount Raised | Post-Money Valuation | Notable Investors |
|---|---|---|---|
| Seed / Early (2017-2018) | ~$9.5M | n/a | Pitango, Ben Horowitz |
| Series A (2019) | $25M | n/a | Pitango, Walden Catalyst, Ahren Innovation Capital |
| Series B (2022-07, Walden Catalyst) | $64M | $664M | Walden Catalyst, Pitango, TPY Capital, SCB10X, Comcast Ventures |
| Series C + extension (2023-08/11) | $208M total ($155M first tranche Aug 2023, $53M extension Nov 2023) | $1.4B (last confirmed) | NVIDIA, Google, Walden Catalyst, Pitango, SCB10X, Comcast Ventures, Intel Capital, Ahren Innovation Capital |
| Series D (2025-05, NVIDIA + Google) RUMORED, never closed | $300M reported (not closed per Calcalist) | ~$1.4B reported (no step-up) | NVIDIA, Alphabet/Google, existing investors (round not reflected in cap table) |
| NVIDIA acquisition talks (2025-12) RUMORED | $2B-$3B reported deal value (acqui-hire) | n/a (acquisition, not a priced round) | NVIDIA (acquirer); deal targets ~200 employees, unconfirmed by either party |
Updated at 2026-06
AI21 Model-Family Release Timeline (Jurassic to Jamba)
| Date | Model / Product | Architecture | Significance |
|---|---|---|---|
| Aug 2021 | Jurassic-1 (Jumbo 178B) + AI21 Studio | Autoregressive Transformer | One of the earliest commercial non-OpenAI LLM APIs; J1 Jumbo at 178B params rivaled GPT-3 on size |
| Oct 2020 | Wordtune | App layer (consumer) | AI-writing assistant; consumer brand that later passed 10M users, AI21's distinctive consumer asset |
| Mar 2023 | Jurassic-2 (J2) + task-specific APIs | Autoregressive Transformer | Customizable LLM with grammatical-error and text-improvement APIs; enterprise pivot begins |
| Mar 2024 | Jamba | Mamba-Transformer hybrid (SSM-Transformer) | First commercial Mamba-Transformer hybrid from any lab; 256K context, open weights; AI21's defining technical bet |
| Aug 2024 | Jamba 1.5 Large + Mini | Mamba-Transformer hybrid | 256K context; the only open model to back the long-context claim on the RULER benchmark; on Vertex AI and Bedrock |
| Mar 2025 | Jamba 1.6 (Large + Mini) | Mamba-Transformer hybrid | Accuracy and speed refresh; AI21 claims it outperforms open competitors from Mistral, Meta, and Cohere on enterprise tasks |
| Jul 2025 | Jamba 1.7 + AI21 Maestro (GA) | Mamba-Transformer hybrid + orchestration | Jamba 1.7 via Studio and Hugging Face; Maestro multi-agent orchestrator GA, AI21's pivot to agentic enterprise workflows |
| Oct 2025 | Jamba Reasoning 3B | Mamba-Transformer hybrid (3B, on-device) | Tiny open reasoning model; first to run 250K-token context on laptops/phones; 2-5x efficiency vs DeepSeek/Google/Meta/Microsoft small models |
Updated at 2026-06
AI21 Revenue and Valuation Plateau (2022 to 2025)
2022-2025
Est. Annual Revenue (USD M)
Updated at 2026-06
AI21 Strategic Backers and Model Distribution Channels
| Partner | Relationship | Distribution / Channel role | Since |
|---|---|---|---|
| NVIDIA | Strategic investor + reported acquirer | Jamba via NVIDIA NIM microservices; co-marketing | Series C 2023; acquisition talks Dec 2025 |
| Google / Alphabet | Strategic investor | Jamba 1.5 family on Vertex AI Model Garden | Series C 2023 onward |
| Amazon Web Services | Distribution | Jamba models served via Amazon Bedrock | 2024 |
| Walden Catalyst | Lead investor | Series B and C lead; deep-tech thesis (no channel role) | Series B 2022 |
| SCB10X (Siam Commercial Bank, Thailand) | Strategic investor | Asia-Pacific enterprise distribution signal | Series B 2022 |
| Comcast Ventures / NBCUniversal | Strategic investor | Media-industry enterprise customer signal | Series B 2022 |
| Hugging Face | Open-weights distribution | Jamba 1.7 and Jamba Reasoning 3B open weights hosted | 2024 onward |
Updated at 2026-06
AI21 Labs in AI Software
OpenAI
$PRIVATE
Largest AI lab by revenue. Approximately $13B annualized run-rate revenue (mid-2025) from a mix of ChatGPT consumer subscriptions, ChatGPT Enterprise / Team, the OpenAI API, and Microsoft Azure OpenAI revenue share. ChatGPT reaches roughly 800M weekly active users (late 2025). Builder of the GPT family (GPT-4o, GPT-4.5, GPT-5, o-series reasoning models) and Sora video. Microsoft is the dominant compute and distribution partner; SoftBank led the $40B March 2025 primary round at $300B.
