The artificial intelligence software startup Modular has raised $100 million in a fresh funding round led by the investment firm General Catalyst. The round also saw participation from Google Ventures, SV Angel, and several others. With this latest investment, the company’s valuation has climbed to $600 million. Modular is setting out to solve a stubborn problem in the AI industry: the overwhelming reliance on one company, Nvidia, for both hardware and software.
Key Takeaways
- Significant Funding: Modular has raised $100 million to speed up product development and strengthen its team.
- Challenging Nvidia: The startup wants to create software that works seamlessly across hardware from multiple manufacturers, not just Nvidia.
- Core Technology: Modular’s foundation is its new Mojo programming language, paired with a unified inference system called the Modular AI Engine.
- Experienced Founders: The company is led by Chris Lattner, the creator of Apple’s Swift language, and Tim Davis, a former leader at Google.
- Breaking the Lock-In: The project directly confronts the vendor lock-in tied to Nvidia’s CUDA software, which has long been the industry’s default.
For years, artificial intelligence has leaned heavily on Nvidia’s graphics processing units, largely because of CUDA. CUDA isn’t just software; it’s a toolkit that makes developing and running AI models far simpler on Nvidia chips. The catch is that once software is written for CUDA, it won’t run elsewhere. That means companies looking to scale AI systems often have no real alternative but to keep buying Nvidia hardware, which has given the company near-monopoly-like control.
Modular wants to break this cycle. The company was founded by two well-known figures in tech: Chris Lattner, best recognized for creating Swift and leading major engineering projects at Apple and Google, and Tim Davis, who also comes from a leadership role at Google. Their idea is fairly ambitious, but also straightforward at its core: give developers the freedom to run AI software on any hardware they choose.
The platform has two central pillars. The first is Mojo, a new programming language designed to combine the ease of Python, which nearly every AI developer already uses, with the performance of lower-level languages such as C++. In practice, that means developers could write code more quickly while still getting the kind of speed and efficiency needed for advanced AI.
The second piece is the Modular AI Engine. This technology is built to run models from popular frameworks like PyTorch and TensorFlow across a wide array of chips. In theory, you could develop an AI application once and then deploy it wherever it makes the most sense, whether on an Nvidia GPU, an Intel CPU, or even a custom-built accelerator.
The newly secured funding will help Modular bring in more engineers and expand the engine’s capabilities. If it works as promised, this approach could give developers and businesses far more flexibility. Instead of being forced into Nvidia’s ecosystem, they could weigh performance and cost across different options and make choices that better fit their needs.
Frequently Asked Questions (FAQs)
Q. What is Modular?
A. Modular is an AI software company that is developing a unified platform for AI development that can work with hardware from different manufacturers.
Q. What is the Mojo programming language?
A. Mojo is a new programming language from Modular. It combines the easy-to-use nature of Python with the high performance of languages like C++, making it suitable for AI development.
Q. Why is Nvidia so dominant in AI?
A. Nvidia’s market dominance is largely due to its CUDA software platform. CUDA makes it simple for developers to use Nvidia’s GPUs for AI tasks, creating a strong ecosystem that developers find difficult to move away from.
Q. What is CUDA?
A. CUDA is a software platform and programming model created by Nvidia. It allows software developers to use the power of Nvidia GPUs for general-purpose computing, which is critical for training and operating AI models.
Q. How does Modular plan to compete with Nvidia?
A. Modular is creating a hardware-independent platform. This will allow AI software written once to run on chips from any company, reducing the industry’s reliance on Nvidia’s closed CUDA system.
Q. Who founded Modular?
A. Modular was co-founded by Chris Lattner, who is best known for creating the Swift programming language, and Tim Davis, who has held leadership roles at Google.
Add Comment