Highlights –

  • Stable Diffusion is an open-source neural network created by Stability AI that can produce images in response to user text input.
  • The business claimed that its new USD 101 million fundraising round will be used by its AI to develop more AI models. The AI models will focus on consumer and business industries.

Stability AI Ltd., an Artificial Intelligence (AI) startup, has received USD 101 million in funding from investors led by Coatue, Lightspeed Venture Partners, and O’Shaughnessy Ventures.

The London-based startup made the revelation about the funding today. As per reports, its total valuation stands at one-billion-dollar today.

Stable Diffusion, an open-source neural network, was created by Stability AI. It can produce images in response to user text input. In August, the startup released the Stable Diffusion source code. Since then, according to Stability AI, more than 200,000 developers have downloaded the neural network.

Another significant startup offering is DreamStudio, a tool that allows consumers to use Stable Diffusion without having to write any code. Stability AI claims that DreamStudio has an installed base of more than one million customers across more than 50 nations.

The firm developed Stable Diffusion via a machine learning development technique known as Diffusion. The approach has also been used in internal AI research by a few other businesses, including Google LLC.

In an AI project that involves Diffusion, engineers provide a neural network with images that contain Gaussian noise, a type of error. Thereon, the neural network is responsible for erasing the noise to restore the original image. The neural network learns to create new images by repeatedly performing the process.

Stability Diffusion is one of several AI systems that can produce visuals in response to text instructions. DALL-E 2, another such system, was unveiled in April by machine learning research organization OpenAI Inc. Additionally, Meta Platforms Inc. has created a neural network that can create graphics based on input from the user.

OpenAI’s DALL-E system consists of 3.5 billion parameters, the configuration settings that ascertain how a neural network processes data. In contrast, Stability Diffusion has 890 million parameters. According to Stability AI, its neural network’s effective design enables it to function on consumer-grade graphics cards, as opposed to the need for servers for similar AI systems.

The business claimed that its new USD 101 million fundraising round will be used by its AI to develop more AI models. The AI models will focus on consumer and business industries.

According to reports, the researchers at Stability AI created neural networks in a customized development environment hosted on Amazon Web Services. The environment is said to contain 4,000 A100 chips, which are top-of-the-line data center graphics cards built by Nvidia Corp. and are tailored for AI applications. There are more than 50 billion transistors on each A100 chip.

Experts’ Take

Emad Mostaque, the founder and Chief Executive Officer of Stability AI, said, “AI promises to solve some of humanity’s biggest challenges. But we will only realize this potential if the technology is open and accessible to all. Stability AI puts the power back into the hands of developer communities and opens the door for groundbreaking new applications. An independent entity in this space supporting these communities can create real value and change.”

Lightspeed partner Gaurav Gupta stated, “At Lightspeed, we believe that an independent company like Stability AI is best positioned to democratize generative AI. The company’s open-source model and its collaboration with AI developer communities have allowed it to innovate rapidly and make the technology freely available to people everywhere.