Insight VC describes Databricks’ wild $10B deal and the bad advice the CEO ignored

It’s been a wild week for investors clawing their way into Databricks’ record-breaking $10 billion fundraising, one of the VCs leading the deal told TechCrunch.

“There were calls that went well late into the night, and that’s okay, that’s how good opportunities emerge,” George Mathew, managing director at Insight Partners, described with a grin. Along with new investor Thrive, Joshua Kushner’s firm, Insight was one of the six firms that led the deal. All but Thrive were existing investors.

“We worked to make sure that we could be a co-lead, despite being already an investor on the cap table,” Mathew said. Insight first invested in Databricks in 2021. But to get into this enormous deal, Insight had to tap into the Insight Partners Public Equities fund, which was set up to buy public stocks, under managing director John Wolff.

There was so much rabid interest that the allocation — and valuation — rose fast. In mid-November, the deal was on track to be around $8 billion, Reuters reported at the time. A few days later, it was $9.5 billion at a $60 billion valuation, and by Tuesday, it had closed at $10 billion with a $62 billion valuation.

For perspective, this is bigger than OpenAI’s $6.6 billion raise in October , the largest venture round of all time.

“There was so much institutional demand and interest for a generational company,” Mathew said. “I’ve been an investor at Insight for the last four years on all things related to data, AI, ML. This is the thing I live for.”

The investment involved a large secondary tender offer, where Databricks employees or other existing investors can sell shares. New preferred shares were issued to the new investor. Databricks didn’t specify how much of the raise was secondary, except to call the $10 billion “nondilutive,” which implies a good chunk.

Interestingly, Databricks, founded in 2013, could have been a tragic tale. A decade ago its founders created a technology, Spark, that was key to yesteryear’s “big data” trend. Spark helped enterprises analyze their in-house big data super fast.

With the rise of data hosted in the cloud, the company was processing data, then handing it over to other players. It could have found itself slowly relegated to an irrelevant big data feature.

Databricks co-founder and CEO Ali Ghodsi (pictured) sought out advice from Mathew, who had run big data company Alteryx as COO before becoming a VC. The two had been friends since Databricks’ early days.

“Ali called me a few years ago and said, ‘Hey, I’m thinking about going into the data warehousing market.’ And I just said, ‘That’s the stupidest idea I’ve ever heard.’ And I could not have been more wrong,” Mathew laughs, adding he’s glad Ghodsi didn’t listen to him, nor hold his bad advice against him.

At the time, traditional data warehouse vendors — which store vast amounts of enterprise data used for analytics — were also struggling against the likes of rising cloud stars like Snowflake and products owned by the cloud vendors, like AWS’ Redshift.

But in late 2020, Databricks launched its data warehouse product anyway — Databricks SQL — and quickly became a big Snowflake competitor.

Then came large language models (LLMs), which are continuously thirsty for high-quality enterprise data. “Where is this high-quality data coming from? For the enterprise, it’s going to come from a place like Databricks,” Mathew said.

Flash forward to the end of 2024, with an IPO market still locked and investors dying to get a piece of AI infrastructure products, like data warehouses that can serve LLMs.

Databricks says that by the end of its fiscal fourth quarter, it will be on a $3 billion revenue run rate, with a $600 million revenue run rate for Databricks SQL, up 150% for the year.