Multiverse raises $27M for quantum software targeting LLM leviathans

Multiverse raises $27M for quantum software targeting LLM leviathans

Multiverse Computing, a San Sebastian-based startup, has successfully raised €25 million (approximately $27 million) in an equity funding round led by Columbus Venture Partners. This latest funding values the startup at €100 million (about $108 million). Multiverse Computing, founded in 2017, is positioned at the intersection of quantum computing and high-level computer science. It specializes in applying quantum principles to tackle complex computations in various sectors, such as finance and artificial intelligence.

The company's software platform, Singularity, is versatile, catering to industries like finance, manufacturing, energy, cybersecurity, and defense. Singularity aims to optimize intricate modeling and predictive applications efficiently, offering a solution to the rising costs and complexities associated with advanced computing. Enrique Lizaso Olmos, Multiverse Computing's CEO, emphasizes the central theme of "optimization" in their pitch to clients across different verticals.

In a significant move toward the AI space, Multiverse is unveiling a new product called CompactifAI. This addition focuses specifically on compressing large language models (LLMs), a crucial aspect of AI operations. CompactifAI aims to cut down on noise and enhance the speed and reliability of AI tasks by streamlining the constant calculations involved in building and querying LLMs. The startup claims that, with its software leveraging "quantum-inspired tensor networks," it can achieve an impressive compression rate of over 80%, ensuring accurate results.

Multiverse Computing operates on a service model, providing access to its platform through APIs. The company's collaboration with clients includes developing strategies to enhance efficiency in their computational processes. Notable clients include Moody's Analytics, Bosch, BASF, Iberdrola, Crédit Agricole, and BBVA.

While the company currently focuses on its consumer-facing website, there are plans to evolve into a core video-generation model that could be extended to other users. Although specifics about the underlying model remain undisclosed, the company has initiated closed API testing with developers. This iterative approach, driven by user feedback, is central to Multiverse Computing's strategy for enhancing its quantum-inspired solutions.

Addressing the challenge of the "uncanny valley" problem in AI-generated content, Multiverse Computing emphasizes its dedication to refining the realism of AI-generated figures, especially in scenarios like walking or snowfall. As the company actively recruits for multiple roles across engineering and marketing, its goal is to build a strong and versatile video-generation model that can be widely adopted.

Multiverse Computing competes in a field with established players like OpenAI's Sora, Google, Nvidia-backed Runway, and others. The startup aims to differentiate itself by continually evolving its video-generation model, addressing industry challenges and incorporating real-world physics into its models.

Investors in the AI-powered video-generation space acknowledge the progress made but recognize the ongoing evolution of technology. The broader industry anticipates a transformative "ChatGPT moment" for video applications, indicating significant advancements still lie ahead.

It's noteworthy that Geoffrey Hinton, initially mentioned as an angel investor, is not directly involved with Multiverse as an investor, clarifying his connection with the startup's founders before its inception.