In a convergence of minds and ideas, Alex Yu and Amit Jain joined hands a few years ago to create a groundbreaking company that would allow individuals to capture 3D objects using just their smartphones – no additional tools required. With Yu as an AI researcher at UC Berkeley and Jain as an Apple employee working on the multimedia features of Vision Pro, the duo embarked on the journey of creating Luma.
Luma’s smartphone app was launched in 2021 and instantly gained traction, attracting millions of users in a short amount of time. But now, as the field of generative AI expands, Yu and Jain have a grander vision for Luma – to go beyond their initial idea and create something even better.
In an exciting announcement, Luma revealed that they will start utilizing a compute cluster of over 3,000 Nvidia A100 GPUs to train new AI models that can, in Yu’s words, “see and understand, show and explain, and eventually interact with the world.”
This is just the beginning of their ambitious plan, which involves developing “next-gen” generative AI models to address the “uncanny valley” problem present in current models. As Yu explains, “AI needs to get a lot smarter to deliver the potential the world sees in it.”
To make this vision a reality, Luma has raised $43 million in a Series B round with the support of investors like Andreessen Horowitz, bringing their total funds to over $70 million. According to sources, this puts Luma’s valuation somewhere between $200 million and $300 million.
Their current focus on launching AI models that can create 3D objects is a highly competitive space, with several object-crafting platforms and startups vying for the top spot. However, Luma believes that their main differentiator is fidelity – the level of accuracy and realism in their models.
Current models are limited by their training on two-dimensional images, resulting in distorted spaces, bodies, and movements when attempting to generate scenes. This hinders their usability and application. With Luma’s advanced generative photorealistic technology, they aim to overcome these limitations and create a seamless experience for users.
Their latest version of Genie, a model for generating 3D objects from text descriptions, is already available on their Discord server. However, Luma has bigger plans on the horizon, with a goal to double their team by the end of next year and create a server cluster with thousands of GPUs.
It’s a bold mission, but one that Luma is committed to achieving. As Yu shares, “With Genie, for the first time, creating 3D objects at scale is possible with AI, and we have already attracted 100,000 users in just four weeks. But our aim is to build even more capable, intelligent, and useful visual models for our users.”
Only time will tell if Luma can fulfill their ambitious goals, but they are wasting no time in making strides towards their vision. With a growing team and cutting-edge AI technology, Luma is poised to make a significant impact in the world of generative AI.