Berkeley-based company Anyscale announced Wednesday that it raised $100 million in fresh financing. The Series C, co-led by Andreessen Horowitz and Addition, cements the company’s status as a unicorn with a valuation of $1 billion.
Anyscale aims to accelerate the development and productization of AI applications. The platform provides developers of all skill levels with a way to build and deploy applications at...well...any scale.
“As we all know now, the pandemic rapidly accelerated the shift to digital for companies across all sectors. Simultaneously, the need to use AI to stay competitive and relevant also became even more urgent during the pandemic,” Robert Nishihara, CEO of Anyscale, said in an email to Built In. “We believe that many organizations came to [us] because [we] eliminated many of the scaling and deployment challenges on the path from idea to production on their AI initiatives.”
Anyscale is the parent company of Ray, an open source project that unifies the end-to-end AI lifecycle with a network of distributed machine learning libraries. With Ray, developers can build apps on their laptops and run them at scale on hundreds of machines, according to the company. Those on the app can scale everything from model training and data ingestion, to model deployment and hyperparameter tuning.
To put it plainly, Ray is the place where the app building happens and Anyscale organizes all the pieces required to get building started. Applications created on Ray can seamlessly migrate to the Anyscale platform and the same goes vice-versa.
Scaling AI workloads typically requires distributed execution across multiple machines, according to the company. That process is not only complex but it also requires special expertise. It’s also expensive. Infrastructure buildout and management costs tend to add up quickly. This is why Anyscale’s solution has proven so valuable.
“No product is quite tackling the challenges of scaling and deploying AI holistically like we are. Yes, there are frameworks that might let you scale a particular piece of the pipeline, say model training, but then you need to get a different tool for scaling your model inference or tuning,” Nishihara continued. “And then suddenly you have six different frameworks stitched together to achieve scale, which is inefficient, expensive and an operational nightmare.”
Ray has been used to deploy apps with a wide array of functionalities. The Anyscale-managed offering has been used to solve problems in supply-chain logistics, virtual environment simulations, retail personalization and more.
More than 18,000 Github users use Ray’s tech in addition to thousands of major corporations like Amazon, Shopify, Uber and Visa.
“We are excited to bring the power of Ray to the broader developer community and organizations worldwide,” Nishihara said.
Following the latest raise, Anyscale plans to expand the size of its team as well as invest in its go-to-market functions. The unicorn is currently on the lookout for 26 fresh faces to join its engineering, product and sales teams, to name a few.
Anyscale has raised $160 million in venture capital financing to date, according to the company.
