Weights & Biases raises $ 135 million for Series C to advance MLOps software – TechCrunch
To update: The round in question was $ 135 million, not $ 100 million as originally stated. I am sorry for that mistake!
What do you call AI today? ML in a suit.
ML or machine learning is a big market today. This is thanks to modern companies that are collecting data like hoarders and maturing data science as a work category. The former can be seen in the growth of Databricks over the past few quarters, and the latter in how much money big tech companies are willing to spend on ML-focused roles.
The market in which Weights & Biases plays is therefore active today. So it’s no big surprise the startup just raised more than $ 135 million in an oversized Series C. The company is now worth around $ 1 billion, according to a press release. Felicis, Insight Partners, Bond and Coatue contributed to the deal.
According to Carta data, data- and analytics-focused Series C rounds have median values of $ 43.75 million and resulting median valuations (after receipts) of around $ 416 million since early 2020. That effectively makes this round a double of what we might expect based on historical data.
In terms of the product, Weights & Biases plays in the area of ”MLOps” or the market for machine learning processes. MLOps is, of course, analogous to DevOps, although it’s a newer category.
According to Weights & Biases co-founder Lukas Biewald, the software world has a number of tools that developers can use to write and deploy code well. This could include a Git-style service (GitLab, GitHub, etc.), monitoring (Atlassian, Datadog, etc.), and the like.
The goal of his company is to build a similar service stack for the ML world. And today, he explained, many ML teams are using ad hoc tools or simply without software assistance.
The need for such a stack could be strong. One difference between development and the ML world, according to Biewald, is that while code crashes when it fails, ML work can “behave badly” in more subtle ways.
Enter weights and distortions, of course. The startup’s product life began with experiment tracking, which Biewald compared with code versioning in the DevOps stack. Git, he said, is great for versioning code that people write, but it is a little bad at handling different versions of computer-generated code, such as those found in machine learning. That’s the kind of problem Weights & Biases wants to address.
The effort is sure to attract the attention of investors. Felicis investor Aydin Senkut told TechCrunch that he had been keeping an eye on Weights & Biases for a while, but that other investors led the last two rounds. This time, Senkut got into the cap table by pre-empting the company. Per Biewald, Weights & Biases would have raised a similar lap, albeit later, if Felicis hadn’t taken the lead.
TechCrunch delved into the startup’s pricing scheme before chatting with the company. The price list looked cheap compared to the productivity Weights & Biases seems to deliver in expanding its service. Note that this is not a compliment in and of itself; Underpricing is a way to quickly transfer the value of the company – and the investors – to customers.
Biewald said Weights & Biases prices their service so that it is easily accessible to everyone. Senkut added that during Felicis’ customer reviews for the investment, customers said the startup undervalued its service by a factor of three.
The investor added that he was excited about the prospect as other companies like Shopify had a similar long-term greed for short-term income.
To be honest, I’m fascinated by what Weights & Biases seem to want to build. Let’s see how far an early nine-digit check can get the company. During the next conversation, it will be time to dig deeper into the growth metrics and review the margin impact of the free service (unless, of course, that particular line is included in their sales and marketing spend line).