Deep Render Earns $9M for AI-Powered Video Compression Technology

Image credit: John Land/Getty Images

Deep Render, a startup developing AI-powered technology to compress video on the web, announced today that it has raised $9 million in a Series A funding round led by IP Group and Pentech Ventures. A source familiar with the matter told TechCrunch that the round valued him at $30 million. That’s a respectable number for a five-year-old startup given the recent market turmoil.

Co-founder and CEO Chri Besenbruch told TechCrunch that the new funding, which does not include a new $2.7 million grant from the European Innovation Council, will move product research and development and Deep Render’s customer acquisition efforts to the United States. Said it would be filled.

“Fundraising will be driven by Deep Render achieving internal milestones and triggering inflection points,” Besenbruch said in an email. “As a very deep ‘hard tech’ company, we have successfully completed R&D and commercialization. ”

Deep Render was founded in 2018 by Besenbruch and Arsalan Zafar. The two met while studying Computer Science, Machine Learning and AI at Imperial College London. Besenbruch and Zafar were inspired by a jointly contributed research project involving transmitting terabytes of video data over a network to explore AI-powered video compression techniques. The project frequently ran into technical roadblocks related to network congestion, so Besenbruch and Zafar explored another (and better, luckily) alternative.

“We decided to combine machine learning, AI, and compression technology to develop a fundamentally new way of compressing data that would significantly improve image and video compression rates,” says Besenbruch. “We started Deep Render to free the world from all bandwidth limitations by pioneering AI-based compression. Instead of trying to “manipulate,” replace it completely. ”

Take those remarks with a grain of salt. Deep Render isn’t the only venture applying AI to the video compression problem, and its AI isn’t necessarily a silver bullet.

Alphabet’s DeepMind adapted an AI algorithm originally trained to play board games to compress YouTube videos. Elsewhere, Nvidia, Disney Research, and the University of California, Irvine, conducted independent experiments using AI-driven compression techniques for streaming video.

Besenbruch believes Deep Render is differentiated by its AI compression algorithm trained on a dataset of over 10 million video sequences. The company used a mix of on-premises and cloud hardware for training, the former of which he consisted of over 100 GPUs.

Besenbruch claims the algorithm is up to five times better than industry-standard codec HVEC (he didn’t disclose criteria) and can run in real time on the latest chipsets from Qualcomm, Apple and Nvidia. doing. (For starters, a “codec” is software, sometimes hardware, that can encode and decode data, usually audio or video data.)

“There is a big challenge in the compression industry to find new methods and look for new innovations,” says Besenbruch. “Traditional compression isn’t moving fast enough. After decades of repeated progress, we’ve reached a peak…Deep Render is resetting the industry and benefiting our customers. ”

Whether it’s true, the last few years offal Cisco predicts that video will account for 82% of all IP traffic by 2022, and sees streaming video traffic grow exponentially. The ability to stream without sacrificing other bandwidth-related areas makes it an attractive investment.

“If it’s hard to grow the internet pipe, then the only option is to make the data flowing through the pipe smaller,” says Besenbruch.

Luckily for Deep Render, there is little startup competition. His Shanghai-based rival Tucodec appears to have gone bankrupt. I can no longer access the website. Already one of his competitors, WaveOne, disappeared almost without a trace after raising his $9 million during its nine-year existence.

Deep Render has raised a total of $15 million and will soon assume debt from the venture. The startup is unprofitable, but Besenbruch claims he has proof-of-concepts and pilots at three big tech companies with over $300 billion in market capitalization (he declined to name them). Did).

But a proof of concept is one thing. Only time will tell if Deep Render can convert them into recurring revenue. After all, companies with in-house talent don’t see much reason to rely on a third-party vendor like Deep Render. Netflix has reduced bandwidth needs by as much as 20% without it A.I. So the challenge for Deep Render is to convince potential customers that the product is really good in some way.

“We expect to have 40 million daily users by 2023 and 300 million users by 2024. We expect annual recurring revenue to exceed $40 million in 2024.” , says Besenbruch optimistically. “Integrating our software into the platform systems of our Big Tech customers will enable this growth.”

Assuming growth happens, London-based Deep Render plans to grow its workforce from 25 to around 45 by the end of the year.