Databricks CEO Ali Ghodsi has denied paying over the odds to acquire large language model (LLM) startup MosaicML for $1.3 billion, rejecting any notion that such deals are fuelling an industry bubble.
Interest in generative AI has surged in 2023, and it’s underscored a flurry of venture capital activity alongside an explosion in new AI startups. But this has also sparked fears that generative AI is the latest in a long series of tech bubbles to have emerged and burst in recent memory, including crypto and metaverse technology.
Most investors responding to a TechCrunch survey, for example, cited AI when asked to predict the next major bubble. Stocks in both new AI companies and existing companies expanding into AI have increased dramatically in 2023.
Databricks has also made a string of recent AI acquisitions, with the MosaicML purchase following deals to acquire Okera in May and Rubicon in June.
Founded in 2020, MosaicML offers the tools for enterprises and startups to train their own LLMs and generative AI systems using data collected first-hand. Crucially, it lets customers own the IP for these systems.
Ghodsi cites MosaicML’s massive surge in revenue over the course of 2023 as among the main factors behind Databricks’ $1.3 billion valuation, which is six times greater than its latest investor-round valuation of $222 million.
The startup has grown from roughly $1 million in revenue to $20 million over six months, Ghodsi said, with the Databricks CEO predicting revenue to climb to $130 million within a year once the firm is granted access to Databricks’ platform and sellers.
Because many startups are being valued at roughly 20 times their revenue rates, Ghodsi said his firm’s valuation is relatively conservative.
“Just to absolutely push back on anyone who says this is a crazy number, just do the math,” he said.
“It’s a very sound, rational financial investment with a good ROI in the very short term, and I think it’s a very strategic long-term investment that’s going to pay off massively in the long run for the ecosystem and for Databricks as well.”
The Databricks CEO added the firm performed more complicated discounted cash flow (DCF) modeling and consulted with lots of customers too.
“Let’s make it clear, we’re not buying the company because we want to get 10x — that’s absolutely not the case,” he continued. “We are super excited about democratizing AI to all startups. We think this is going to be massive over the next five years.”
