--- title: "Tripo AI’s fast track to an AI-native 3D future" url: https://www.voxelmatters.com/tripo-ais-fast-track-to-an-ai-native-3d-future/ date: 2025-09-17 modified: 2025-09-25 lang: en author: "Davide Sher" description: "Some might not have realized it yet, but creating 3D models—whether for virtual worlds like gaming or for physical use in 3D printing—has recently become a heck of a lot..." categories: - "3D Modeling" - "AI" - "AM Software" - "Consumer Products" - "Decision Makers" - "Executive Interviews" tags: - "Featured" image: https://www.voxelmatters.com/wp-content/uploads/2025/09/Tripo_a4f736ad-3df1-404e-9984-b0da6ecba171-640x366.jpg word_count: 1511 --- # Tripo AI’s fast track to an AI-native 3D future Some might not have realized it yet, but creating 3D models—whether for virtual worlds like gaming or for physical use in 3D printing—has recently become a heck of a lot more accessible. This shift is driven by a new wave of AI applications that can transform text prompts or 2D images into fully developed 3D models. One of the fastest-growing companies in this space is [VAST, founded by Simon Song, which is making the creation of 3D products easier than ever through Tripo AI.](https://www.voxelmatters.directory/company/tripo3d-tripo-ai-by-vast/) ![](https://www.voxelmatters.com/wp-content/uploads/2025/09/Simon-scaled-e1757949549172.jpg)VAST founder Simon Song. Listening to Simon Song talk about building VAST and Tripo AI, you hear both the researcher and the entrepreneur at work. He studied for more than eight years in the United States, graduating from Johns Hopkins University after attending the Stony Brook School, and even picked up Hebrew and Arabic along the way. His first major stop was SenseTime, one of the biggest AI companies in Asia, a place that scaled at breakneck speed while he was there. He later co-founded MiniMax—now valued in the billions—before moving on to found VAST in 2023 with a sharpened focus: “We want to build a virtual world together. We do a lot of research on AI 3D—use text or image as input to generate 3D.” The early days of AI-generated 3D were anything but smooth. As Simon puts it, the technique born at the end of 2022 “was immature… You cannot get anything, and it always fails and takes hours, maybe days, to generate one 3D model.” He and his team pushed through: “We spent about a year to make it through,” publishing more than fifty research papers accepted by CVPR, SIGGRAPH and ICCV, open-sourcing over twenty projects, [and stacking up roughly 18,000 GitHub stars across releases such as Wonder3D, ThreeStudio and Tripo-SR](https://github.com/VAST-AI-Research). “We love to share our technology and want the industry involved faster… so we can develop a product on it.” Culture is a competitive advantage. Simon describes an engineering org that behaves like a product studio. “Our engineers proactively think, ‘we need this, we need that.’ And then everything goes quickly. If you don’t like it, get rid of it. It’s simple.” He laughs about their cadence: “Human and machine, only one has to sleep.” Researchers queue training jobs at night; results greet them the next morning. That pace explains how Tripo often integrates new ideas, such as recently with Google’s Nano Banana, almost as soon as they appear in the wider AI ecosystem. Profitability is not an afterthought. Simon is blunt: “It’s a profitable business right now.” Inference is efficient; the heavier cost sits in training large 3D foundation models. Yet the commercial traction is clear. Tripo AI already has “close to four million users around the world” and “40,000 business partners” using its API, including giants such as Tencent, ByteDance and HTC, alongside innovators like Stability AI. The business is monetizing at scale—“this year we got up to $12 million in revenues… which means in one month we earn $1 million per month”—and it’s only in its third year. https://youtu.be/cfO9nVaSm5M ## The door to a VAST universe VAST presents itself as a global generative-3D startup building pipeline-ready meshes from text and image prompts. Simon’s algorithm group draws talent from Nvidia, Google, Meta, Tencent and ByteDance, and the company’s ambitions extend far beyond a single app. He characterizes the traditional 3D creation ecosystem—DCC (Digital Content Creation) tools, engines, plugins, and XR pipelines—as “a mess.” The aim is to replace that brittle stack with an AI-native workflow that dramatically lowers barriers for everyone. “We want our developers – or our creators – to create 3D content in our studio,” he says, making the process “much easier than using different DCC tools.” VAST’s commercial ecosystem stretches from web apps and API to plugins and a growing community. The company is also incubating HolyMolly, a “3D TikTok” powered by Tripo’s models. The throughline is accessibility. Simon wants professionals to go faster and newcomers to participate meaningfully. ![VAST and Tripo emerged from an open-source collaboration among AI and 3D engineers. Founder Simon Song tells us how he is making 3D printing truly ubiquitous.](https://www.voxelmatters.com/wp-content/uploads/2025/09/Tripo3D-printed.jpg)Some 2D to 3D printed items after only a few weeks of using Tripo 3D by this writer. That philosophy carries into manufacturing, where VAST is stitching together a flexible supply chain for on-demand production. “We collaborate with nearly all 3D printing companies,” he says, citing Bambu Lab, Anycubic, Creality and Elegoo on the consumer side and Stratasys for industrial printing. Some of the capacity is centralized; much is distributed across “different farms around the world,” and VAST even operates its own machines. Simon believes the time has come for personalization to outgrow standard SKUs. In his view, 3D printing is “a new way of manufacturing in the AI era,” enabling anyone to become an industrial designer. He imagines people generating original characters, printing them as gifts, and even selling to fans. The company is preparing two mobile apps: one for interactive, real-time 3D experiences—“you punch it, it will cry,” he jokes—and another for creator-led design and 3D printing. “Everyone could generate their own character or their own piece of love as a gift or sell to other people,” he says, envisioning a community where users remix famous IPs, swap techniques, and produce short runs through VAST’s supply network. [![](https://www.voxelmatters.com/wp-content/uploads/2025/09/Tripo_IMG_9341-copia-640x480.jpg)](https://www.voxelmatters.com/wp-content/uploads/2025/09/Tripo_IMG_9341-copia.jpg) Monsters from the Monster Hunter videogame series [![](https://www.voxelmatters.com/wp-content/uploads/2025/09/Tripo_IMG_9340-copia-640x480.jpg)](https://www.voxelmatters.com/wp-content/uploads/2025/09/Tripo_IMG_9340-copia.jpg) Aniaml (or animal statues) photos made into 3D models and printed. [![](https://www.voxelmatters.com/wp-content/uploads/2025/09/Tripo_IMG_9343-copia-640x480.jpg)](https://www.voxelmatters.com/wp-content/uploads/2025/09/Tripo_IMG_9343-copia.jpg) Some extremely detailed flowers. [![](https://www.voxelmatters.com/wp-content/uploads/2025/09/Tripo_IMG_9339-copia-640x480.jpg)](https://www.voxelmatters.com/wp-content/uploads/2025/09/Tripo_IMG_9339-copia.jpg) Replicas of statues from a visit to the Dalì house museum in Figueres ## From Tripo to everything Tripo AI is VAST’s flagship: a global 3D foundation model launched in 2024 that turns prompts, sketches and images into high-quality meshes in minutes. The latest release, Tripo 3.0, marks a leap in precision and production-readiness. Simon explains that the team “introduced a new way of representing 3D… called Sparse Flex.” This representation, open-sourced on GitHub as tripo-SF, makes models “more detailed… more accurate and also faster,” while lowering data needs and training cost. In his words, it’s “cheaper, faster, bigger, more accurate.” On top of the new geometry-and-texture pipeline, Tripo 3.0 adds two output modes. Standard Mode delivers optimized topology for industrial use—ideal for product design, games and real-time engines—while Ultra Mode pushes geometric clarity to the edge for scenes where fidelity is king. The studio’s experience has also been enhanced, with improvements in privacy controls, smoother navigation, and faster generation. ![VAST and Tripo emerged from an open-source collaboration among AI and 3D engineers. Founder Simon Song tells us how he is making 3D printing truly ubiquitous.](https://www.voxelmatters.com/wp-content/uploads/2025/09/Tripo_29d1293b-66d9-46f6-a8c8-0acc3e60aa47.jpg) Workflow features round out the release. There’s first-class integration with cutting-edge image models (Flux and GPT-4o), a sketch-to-3D pipeline that lifts rough drawings into printable assets, and character exports in T-pose with auto-rigging for rapid animation. Magic Brush 2.0 offers finer control when painting and blending, and Smart Part Segmentation breaks complex models into logical pieces for easier editing, retexturing and, crucially, for printing as separate components. Simon highlights the cross-over value: “Segmentation is useful in 3D printing… and the same for gaming as well, because they want to edit single parts.” Auto-rigging helps both worlds, too. While full pose editing still routes through separate CAD like Blender today, he promises it will arrive natively: “Maybe in one or two months we can move this function into our studio.” Tripo’s 3D results increasingly “look good up close” and drop into pipelines with minimal cleanup—precisely what teams need when deadlines loom. Simon is confident about the competitive landscape, inviting practitioners to try direct comparisons: “Try our 3D generation technology. And compare the results. Ours will be better.” ![](https://www.voxelmatters.com/wp-content/uploads/2025/09/robot-scaled.jpg) ![](https://www.voxelmatters.com/wp-content/uploads/2025/09/Tripo_IMG_9342-copia.jpg)A robot character created using Google's Nano Banana and then made into 3D. The version on the left is printed in multiple colors on a Bambu Lab; the version on the right was divided into sections using Tripo, then printed on a Form 4 and reassembled. This is just a small glimpse of what can be achieved with Tripo Studio. In the photo above, the system is generating another version of the robot. ## A VAST 3D future Asked where VAST is heading, Simon returns to the community. He foresees a wave of AI-native micro-studios—teams of one to five people who thrive on taste and ideas rather than deep technical training. “What they need are ideas and a sense of beauty… AI will do the rest for them.” His goal is to give these creators the scaffolding to flourish: education, tools, opportunities, even sales channels. The same platform that powers major gaming and film companies should empower a teenager to sketch a character on a phone and see it come alive in minutes, ready for animation or ready for print. The path from research to real-world use is what energizes him. He remembers when a single model might take days to produce and often failed outright. Now, with Tripo 3.0 and Sparse Flex, generation lands in minutes and pipelines are cleaner. He’s quick to acknowledge that the work is unfinished—“It’s not perfect now”—but the trajectory is unmistakable. As he puts it, “That is the beginning of our journey,” and the destination is an AI-native creative economy where the hard parts of 3D production melt into the background. For readers who want to explore the technology or try it in their own pipeline, Tripo AI’s studio and documentation are publicly available at the company’s site, with developer access via API and plugins across popular tools. Listen to the full interview in the VoxelMatters Podcast: