Will Decart's Real-Time AI Become A Trillion-Dollar Business?

Key Concepts

• AI-generated digital worlds
• Real-time video models
• Diffusion-based models
• Latency reduction
• Infinite clip generation
• Kilo-corn (trillion-dollar company)
• Consumer experiences
• Enterprise licensing
• Ethical considerations of AI video manipulation

1. Introduction to Decart and AI-Generated Digital Worlds

• Dean Leersdorf, co-founder and CEO of Decart, discusses building AI models that generate digital worlds.
• Decart aims to create a new creative experience using AI video and audio, moving beyond endless scrolling on existing platforms.
• The goal is to provide a fun, creative outlet, similar to how chatbots revolutionized information access.

2. Mirage: A Concrete Example of Decart's Technology

• Decart launched Mirage, which allows users to transform any video stream (camera, screen share) into different virtual worlds.
• Example: A Zoom call could be transformed into a scene from Frozen or Star Wars.
• Users have been using Mirage to create real-world scenarios with AI filters, such as turning broomstick fights into lightsaber battles or transforming their homes into Minecraft environments.
• Mirage enables real-time video model application, unlike traditional AI video generation, which takes time.
• Future developments include generating videos from scratch based on user descriptions, allowing for interactive video creation. Example: Summoning a sword or a monkey into the video feed.

3. New Use Cases and Applications

• While similar to Snapchat filters, Decart's technology opens up new use cases.
• VR integration: Users can walk around cities in VR headsets and transform the environment into imagined scenarios, like a Godzilla fight.
• Decart aims to create real-time video interaction, similar to how chatbots enabled real-time text interaction.
• Potential applications range from FaceTime calls with AI characters to transforming living rooms into different environments.

4. The Internet and the Role of AI

• The internet is categorized into knowledge, creativity, communication, and shopping.
• Chatbots have revolutionized the knowledge category.
• Decart aims to revolutionize the creativity category by enabling on-the-fly video generation.
• Potential use cases include infinitely generating personalized Tik Tok streams based on user emotions.

5. Technological Breakthroughs in Training Real-Time Video Models

• Two main categories of AI models: text-based models (LLMs) and diffusion-based models (video/audio generation).
• Mirage is the first real-time video model, generating video in approximately 40-50 milliseconds.
• Key breakthroughs:
  • Latency Reduction: Reducing video generation time from tens of seconds to milliseconds.
  • Infinite Clip Generation: Generating continuous, long-form videos instead of short clips.

6. Competition and the Vision of Building a Kilo-Corn

• Decart aims to build a "kilo-corn" (trillion-dollar company) by creating new consumer experiences based on deep technological breakthroughs.
• The company acknowledges the intense competition in the AI field.

7. Business Model and Revenue Streams

• Decart employs a dual business model: selling to consumers and licensing technology to enterprises.
• The company has generated revenue from licensing its real-time video models to enterprise customers.
• Examples of enterprise interest: hotels using Mirage in lobbies, hospitals using it to distract children during surgeries.
• Decart aims to create a new experience that enhances creativity and leisure time for users.
• The company's goal is to answer the question: "What AI do people use when they're not doing their homework?"

8. Funding and Future Plans

• Decart has raised over $150 million and burned less than $10 million due to revenue from licensing.
• Future plans include releasing new models with enhanced capabilities and launching consumer products like VR, FaceTime-like, and Snap-like apps with AI twists.
• The funding will be used to scale infrastructure (GPUs) to serve millions of users.

9. Revenue and Customer Base

• Decart reported $20 million in revenue in 2024.
• Customers include companies in gaming, real estate, and entertainment.
• In the past year, Decart licensed tech that enabled running models efficiently.
• Starting next week, Decart will provide public access to its models via an API, creating a new revenue stream.

10. Dean Leersdorf's Background

• Dean Leersdorf's background includes experiences in Israel, Switzerland, and Palo Alto.
• He realized the high level of competition in Palo Alto, which motivated him to excel.
• He completed his undergrad, master's, and PhD in distributed computing in five and a half years.
• His PhD focused on distributed computing, which is now foundational for AI.

11. Ethical Considerations and Safety Measures

• The potential for manipulating video content raises ethical concerns.
• Setting expectations that content is imaginative, not necessarily factual, is crucial.
• The industry needs to address how to adapt to a world with AI-generated content.
• Decart is not currently manipulating audio due to safety concerns.

12. Moonshot Vision for Decart

• Decart aims to create a fun, relaxing, and creative experience with AI, moving away from doom scrolling.
• The goal is to impact hundreds of millions or billions of people by providing technology that enhances their lives.

Synthesis/Conclusion

Decart is developing real-time AI video generation technology with the goal of revolutionizing creative expression and entertainment. Their product, Mirage, allows users to transform real-world video streams into imagined scenarios. The company is focused on technological breakthroughs in latency reduction and infinite clip generation. Decart aims to build a trillion-dollar company by providing both consumer-facing applications and enterprise licensing of their technology, while also addressing the ethical considerations of AI-generated content.