The 'Steam Engine Moment' for 3D Content Production: VAST Raised 1 Billion RMB

1 billion RMB.

That's the size of VAST's (Sanqi Wanwu) newly completed Series A3 funding round.

The lineup of investors is interesting - Geely Capital, Giant Network, 4399, Tanwan Games, plus a bunch of top-tier investment firms and local industrial funds. Existing shareholders Ce Capital and Origin Capital also increased their investment.

Friends, this isn't just another AI startup raising money. I think there's a signal worth pondering behind this.

What exactly is this company doing?

In one sentence: VAST enables AI to understand and generate 3D worlds.

In the past six months, they've done three things:

1. 3D Foundation Models Tripo H3.1 and Tripo P1.0 - can rapidly generate high-quality 3D models
2. 8K textures and intelligent segmentation algorithms - making 3D content more refined and controllable
3. World Model Project Eden - proposing a technical route of "decoupling underlying state inference from native visual rendering"

I need to say more about that last item.

What is a "world model"? Simply put, AI can not only "see" 3D scenes but also "understand" how things happen in that scene and what will happen next.

Give it a room, and it not only knows what the room looks like, but can also reason that "if you throw a ball here, it will roll there" - that's a true "world model."

VAST's proposed technical approach separates "understanding how the world works" from "rendering the world" - I think that's a smart approach.

Why are these giants betting?

I noticed a detail: there are particularly many industrial investors in this round.

Geely Capital (cars), 4399 and Tanwan Games (gaming), Giant Network (also gaming) - why are these industrial parties investing in VAST?

Because they all lack one thing: the capability to efficiently produce 3D content.

Game companies need massive amounts of 3D assets (characters, scenes, props); automakers need 3D visualization to showcase products and train autonomous driving; the metaverse, VR/AR fields can't function without 3D content.

How slow is traditional 3D content production? A skilled modeler might take days or even weeks to create a high-quality character.

But what can VAST's Tripo achieve? Single-image 3D model generation in 0.5 seconds.

This isn't "efficiency improvement," this is a paradigm-level shift.

So when giants like Geely and Giant Network all bet on VAST, the signal I see is: the UGC infrastructure era for 3D content might really be coming.

What is a World Model? Why is it important?

Let me explain this concept properly to friends.

Most current AI models are essentially "image generators" or "text understanders" - they can generate nice images and understand what you said.

But world models are different - they answer the question: how does this world work?

A simple example:

• Image generator: generates a picture of "an apple falling from a tree"
• World model: understands "why the apple falls," "what trajectory it will take," "what happens when it hits the ground"

The first is "drawing," the second is "understanding."

What VAST's Project Eden aims to do is enable AI to truly "understand" 3D worlds - not just render nice-looking 3D scenes, but reason about object relationships, causality of events, and future developments in that scene.

What's this good for?

I can think of several scenarios:

Games: NPCs are no longer script-following wooden puppets, but intelligent agents that can truly "understand" the environment and make reasonable decisions Simulation: testing new products and solutions in virtual worlds, with results directly informing real-world decisions Creation: creators describe a scene, and AI automatically generates an entire world - including visuals, logic, and playability

All of these require "world models."

How do I see this?

In this 1 billion RMB funding round, I see a few key points:

1. Industrial clustering - shows this isn't just financial investment, but giants really need this technology
2. Existing shareholders continuing to invest - Ce Capital and Origin Capital both participated again, showing they've seen actual progress
3. Clear technical roadmap - Tripo handles "3D generation," Project Eden handles "world understanding," division of labor is clear

But I also need to throw some cold water:

VAST isn't the only player in the world model race. Genie and SimMC abroad, quite a few teams domestically. How high are the technical barriers, how fast can it land in industrial scenarios, can it monetize - these are all open questions.

But one thing I'm quite sure of: the cost of 3D content production is dropping exponentially.

Making a AAA game used to burn hundreds of millions just for art assets; in the future, a small team with AI tools might create works rivaling major studios.

This change's impact on the entire gaming, metaverse, and VR/AR industry might be bigger than we imagine.

Final Thoughts

Friends, I often say: every industry has its "steam engine moment" - when production efficiency increases by tens or hundreds of times, the entire industry's logic gets rewritten.

Steam engines for manufacturing, containers for shipping, the internet for information dissemination.

Now, 3D content production might also be facing its "steam engine moment."

VAST raised 1 billion RMB, showing that capital and industry both believe this moment is coming.

But can the technology truly land, can it change the industry, can it benefit creators and users - these remain to be seen in the next two years.

Leaving you with a question: Do you think the "steam engine moment" for 3D content production has really arrived? I'd love to discuss in the comments.