Moonvalley

Product & Competitive Intelligence

Licensed-data foundation video model for professional film and enterprise production.

Media

|

Generative Video

|

Series A

|

Valuation:

~$800M

Company Overview

Moonvalley builds Marey, a text-to-video and image-to-video foundation model trained exclusively on licensed footage, with frame-level camera, motion, and pose controls. Serving customers across film studios, creative agencies, and enterprise brand teams (via CAA, ComfyUI, fal.ai, Adobe).

Latest Intel

Zeitgeist tracks private signals to determine where the company is investing and what it means competitively.

May 6, 2026
|
Burn Rate and Runway
|
Confidence:
Medium

New Intel: Moonvalley raised $154M across four rounds in nine months. Licensed video data plus HD model training is a cash-heavy bet, and the next round will need traction.

The raise cadence points to heavy compute and licensing burn.
May 6, 2026
|
Product Direction
|
Confidence:
High

New Intel: Moonvalley is trying to become the video model layer inside Adobe, fal.ai, ComfyUI, and API workflows, not another Runway-style front end.

API distribution could make Moonvalley the model layer inside other tools.
May 6, 2026
|
Competitive Positioning
|
Confidence:
High

New Intel: Moonvalley is selling legal safety more than raw video quality. The real buyer is the studio legal team, and CAA plus Comcast Ventures give it routes into that room.

Studios buy legal safety before they buy model quality.

What They're Building | Public Roadmap

Products the company has committed to building publicly.

Marey Realism v1.5:

Higher-fidelity HD model with improved physics and prompt adherence, available via web, Discord, ComfyUI, and fal.ai.

Public API:

Developer and enterprise access to Marey, currently in early-access signup.

Adobe Integration:

Marey inside Adobe's creative ecosystem, announced as a distribution partnership.

Asteria Collaboration:

Co-developed workflows with the AI-native studio Asteria for feature film production.

Competitive Landscape & Moat

The training dataset is the moat. Marey is trained only on licensed footage, which gives Moonvalley legal indemnity to sell into studios and brand-safe enterprise buyers who cannot touch Runway or Sora output without risk. Replicating that data position requires years of licensing deals, not a bigger GPU cluster.

Direct Competitors

Runway ML:

Larger install base and stronger creative-user adoption, but trains on scraped web video, which limits enterprise and studio sales.

Google Veo 3:

Better raw model quality and Google distribution, but Moonvalley has cleaner IP provenance and an indie-studio go-to-market Google will not prioritize.

Luma AI (Dream Machine):

Focused on consumer and prosumer creators with faster iteration cycles, while Moonvalley targets controllable, studio-grade output.

Add To Your Portfolio

Founding Team

Naeem Talukdar

Founder and CEO (ex-Zapier)

Mateusz Malinowski

Co-Founder and Chief Scientist (ex-Google DeepMind)

Mikołaj Bińkowski

Co-Founder and VP Research (ex-DeepMind)

John Thomas

Co-Founder and COO

Founder Force Multiplier

Malinowski and Bińkowski are both ex-DeepMind researchers with deep video generation and generative modeling pedigrees, which is the technical half of the problem. Talukdar brings consumer growth and product instincts from Zapier and a YC company, which is exactly the distribution muscle most research-first video labs lack.

Funding History

2024 | $70M Seed led by General Catalyst

2025 | $43M venture round (April)

2025 | $10M extension (May)

2025 | $84M Series A with General Catalyst (July)

Back To All Companies >