TL;DR:
- Luma, co-founded by Alex Yu and Amit Jain, originally focused on 3D object capture via smartphones.
- The company has pivoted to develop AI models capable of “seeing, understanding, showing, explaining, and interacting with the world.”
- Luma raised $43 million in a Series B funding round, valuing the company at $200 million to $300 million.
- Their goal is to create advanced generative AI models to address challenges in 3D object generation.
- Luma aims to differentiate itself through improved fidelity in 3D model creation.
- The company plans to expand its workforce and build a robust model-running server cluster with thousands of GPUs.
Main AI News:
In a dynamic partnership, Alex Yu and Amit Jain embarked on a journey to revolutionize 3D object capture using nothing but the humble smartphone. Yu, an AI researcher at UC Berkeley, joined forces with Jain, an Apple visionary shaping the Vision Pro’s multimedia realm. The outcome? Luma is a groundbreaking smartphone application that soared to prominence in 2021, boasting over two million users at the time of this publication.
But in a landscape now teeming with generative AI innovation, Yu and Jain are steering Luma towards a grander vision, one that promises to surpass its initial conception. Luma’s recent announcement signifies a pivotal shift as the company prepares to harness the power of approximately 3,000 Nvidia A100 GPUs to train next-generation AI models with a profound ability to “see, understand, show, explain, and ultimately interact with the world,” in Yu’s own words.
The initial phase of this audacious plan revolves around crafting models capable of translating textual descriptions into intricate 3D objects. Luma made its debut in this arena earlier this year with the launch of Genie on its Discord server. Their subsequent endeavor will involve the development of “next-gen” generative AI models designed to tackle the pervasive “uncanny valley” challenge currently faced by existing GenAI.
Alex Yu articulates Luma’s strategic focus, stating, “We believe that multimodality is critical for intelligence. To go beyond language models, the next breakthrough will come from vision.” He adds, “AI needs to become significantly smarter to unlock its full potential.“
To bring this visionary ambition to fruition, Luma has successfully secured $43 million in a Series B funding round, with participation from prominent backers, including Andreessen Horowitz, among others. Insider sources indicate that this funding round values Luma at an impressive range of $200 million to $300 million, bolstering their war chest to over $70 million.
However, the realm of 3D model creation through AI is fiercely competitive, featuring established players like 3DFY and Scenario, along with emerging startups like Hypothetic, Kaedim, Auctoria, and Mirage. Even industry giants such as Autodesk and Nvidia are venturing into this domain with applications like Get3D and ClipForge.
So, what sets Luma apart from the competition? According to Yu, it’s all about fidelity. He explains, “Current models are primarily trained on two-dimensional images, and when tasked with generating complex scenes, they often distort spaces, bodies, and movements. Achieving coherent and usable output is a significant challenge in initial attempts, limiting their practicality. We aim to introduce the most advanced generative photorealistic technologies through an intuitive app.”
While these aspirations appear ambitious, particularly given the early stage of Luma’s ambitious roadmap, the company shows no signs of slowing down. Luma intends to double its current 24-person team by the end of the coming year. Additionally, they are diligently assembling a high-capacity server cluster, equipped with “thousands” of GPUs dedicated to model execution.
The release of an enhanced Genie today marks just the beginning of Luma’s journey toward more sophisticated generative AI models. As Amit Jain asserts, “With Genie, we’ve made it possible for AI to create 3D objects at scale, attracting 100,000 users in just four weeks. However, our ultimate goal is to construct far more capable, intelligent, and practical visual models for our users.“
Conclusion:
Luma’s significant funding and strategic shift towards AI-driven 3D modeling positions them to compete in a burgeoning market. Their focus on fidelity and the development of advanced generative AI models sets them apart, signaling their intent to lead in this transformative field. The influx of capital and expansion plans underscore their commitment to achieving this vision.
