Data Startup Labelbox Reaches $ 1 Billion Valuation Thanks to SoftBank Funding


With funding for data startups in a frenzy, data annotation company Labelbox found itself making offers to accept ever larger sums from private market investors at the end of 2021. A Capital Investor -risk to direct the investment on a check for $ 200 million. In the end, he teamed up with a famous big check issuer, SoftBank, but chose to take only about half of the stake.

Labelbox announced Thursday that it has closed a $ 110 million Series D funding round led by SoftBank’s Vision Fund II. It also chose two new strategic investors: Snowpoint Ventures, to help it sell to the government, and Databricks Ventures, to strengthen a partnership with the high-flying data management company which is expected to go public widely this year. Follow-up investments came from Andreessen Horowitz, Group B Capital of Eduardo Saverin and Ark Invest asset manager Cathie Wood. Robert Kaplan of SoftBank, who first invested in Labelbox while working at B Capital, joined as a board observer. Co-founder and CEO Manu Sharma declined to disclose an exact valuation, but said the new funding made Labelbox “essentially a unicorn,” putting it on the brink of the billion dollar precipice.

Launched in 2018 by Sharma, Brian Rieger and Dan Rasmuson (who is no longer with the company), Labelbox tries to differentiate itself from its peers, many of whom were founded years earlier and some of whom have already become unicorns. Sharma and Rieger met while studying for aerospace engineering degrees at Embry-Riddle Aeronautical University, a school an hour from Cape Canaveral known for producing many airplane pilots in the United States. United. The two aviation enthusiasts attempted to start a space hardware business after graduation, but the business failed. In 2017, both started new jobs using artificial intelligence to process data more efficiently. “We were only there for a short time because we saw that there was a whole new paradigm [for AI]. Everything they taught us in school is happening now. says Sharma, who recently moved alongside Rieger to the outskirts of Miami, more for aviation than the Web3 scene, they say.

True to its name, Labelbox is designed for the data labeling portion of the machine learning process, which allows smaller batches of ‘training data’ to be annotated so that AI models can learn how to do it. predictions and to develop information from larger amounts of raw data. Some well-supported startups also providing data tagging solutions include Scale AI, a $ 7.3 billion valuation, which announced last October that it had reached $ 100 million in annual recurring revenue and 1 billion dollars Snorkel. Sharma maintains, however, that over the past 18 months, Labelbox has become the top startup in its field among corporate clients. “When we talk to the business, we don’t see the businesses that you might think of as our competition. On the contrary, he says, Labelbox is primarily in competition with Amazon’s SageMaker and in-house solutions that large companies are trying to develop on their own. Sharma has been quiet about Labelbox’s earnings, but said she has 200 clients and a net retention rate of over 150%.

“We sell to large organizations that have multiple use cases for AI and who fundamentally believe that their differentiation in the future must be with AI or they will become useless,” he says. Auto insurer Allstate, one of the startup’s early clients, used Labelbox to develop models to identify specific features in an insurance claim image, such as license plates and type of vehicle. damage to vehicles. She is now using the software for new purposes, like analyzing aerial images and mobile data, and Sharma says the company was able to go from the idea stage to deploying the AI ​​model in two weeks. NASA’s Jet Propulsion Lab uses Labelbox to tag videos showing the motion of Earth’s molecules to train a variety of algorithms that will search for realistic motions in space.

The new capital injection, well above the total of $ 79 million raised in previous rounds, will serve to accelerate the development and distribution of Labelbox’s products for the corporate segment, Sharma said. The differentiation in its development comes from the company’s emphasis on building software tools to improve data quality, rather than on the code itself. Andrew Ferguson, vice president of corporate and enterprise development for Databricks, said his company chose Labelbox to deepen its partnership because of its technology-driven approach, rather than one where “an army of people ”does part of the task manually:“ It’s a valid approach and some companies are taking it, but it’s less infinitely scalable than a technology-driven approach.

This is a similar reason why Labelbox chose SoftBank as the primary investment in the cycle. Sharma recounts a conversation with Masayoshi Son as “one of the best we’ve ever had” with a VC. “There is a tendency to focus on the here and now with typical venture capital commitments,” says co-founder Rieger, who is the president of Labelbox. Talking with SoftBank turned out to be an ‘aha moment’ due to Son’s focus on the startup’s long-term ambitions. “When you take that framework, the questions significantly turn into big, fundamental questions such as’ How is the market going to evolve? “Or” what are the checkpoints in the market? ”Says Rieger, highlighting future AI developments relevant to Labelbox, such as processing“ an incredible amount of information ”from 4K video data.

Regarding distribution, Labelbox wishes to rely on the contracts it has signed with the Ministry of Defense to develop further in the government sector. “They are focusing on the exact problem that I have identified in government as the most urgent need,” said Snowpoint Ventures co-founder Doug Phillipone, who led Palantir’s strategy to break into government. “In the absence of a Labelbox-like platform, they literally have what they call ‘sweatshop labeling’. . “The government has the most interesting problems in terms of the size and scale of the data,” he says. “This stuff has to work because our nation depends on it.”


Comments are closed.