AI

Magic AI secures 320 million dollars in new funding round

Magic AI secures $320 million in a new funding round and announces the construction of supercomputers in collaboration with Google Cloud and Nvidia.

Eulerpool News Sep 2, 2024, 10:56 AM

The AI startup Magic, which develops models for code generation and the automation of various software development tasks, has raised $320 million in a new funding round. Investors include former Google CEO Eric Schmidt, Alphabet's CapitalG, Atlassian, as well as notable names like Elad Gil and Sequoia Capital. This investment propels Magic into the league of well-funded AI coding startups and raises the company's total capital raised to an impressive $465 million.

According to reports from Reuters, Magic aimed to raise over $200 million in July at a valuation of $1.5 billion. The now secured amount significantly exceeds these expectations, although the current valuation of the startup is not yet publicly known.

Magic also announced a partnership with Google Cloud to develop two "supercomputers" on the Google Cloud Platform. These clusters, based on Nvidia H100 GPUs and the upcoming Blackwell chips from Nvidia, are expected to scale to "tens of thousands" of GPUs over time and achieve performance of 160 exaflops – equivalent to one quintillion computing operations per second.

We are excited to work with Google and Nvidia to build our next AI supercomputer on Google Cloud," said Magic co-founder and CEO Eric Steinberger in a statement. These new systems are expected to significantly enhance efficiency in model training and inference.

Magic was founded in 2022 by Eric Steinberger and Sebastian De Ro. Steinberger, who was fascinated by the potential of artificial intelligence since high school, dropped out of his computer science studies in Cambridge after one year to work as an AI researcher at Meta. De Ro, who was previously CTO at the German company FireStart, met Steinberger at the environmental organization ClimateScience.org, which he co-founded.

Magic develops AI-based tools to assist software developers in writing, reviewing, and debugging code. A unique feature of Magic is the architecture of the "Long-term Memory Network" (LTM), which enables extremely long context windows. The latest model, LTM-2-mini, has a context window of 100 million tokens, allowing it to consider about 10 million lines of code or 750 novels simultaneously.

The investors see great potential in the market that Magic aims to tap into. According to estimates from Polaris Research, this market could reach a value of 27.17 billion dollars by 2032.

Access the world's leading financial data and tools

Subscribe for $2

News