NVIDIA has recently invested in a "highly promising AI company for IPO": Databricks.

On Thursday, September 14th, data analytics and artificial intelligence software manufacturer Databricks completed a Series I financing round of over $500 million, valuing the company at $43 billion, firmly establishing itself as one of the hottest players in the field of artificial intelligence.

It is worth noting that while the valuations of many technology startups have been significantly reduced, Databricks not only maintains a stock price of $73.50 per share, roughly equivalent to the pricing of the previous financing round two years ago, but also has a valuation that is $5 billion higher than the $38 billion valuation in August 2021, representing a premium of 13%. The number of employees has doubled to approximately 6,000 since the last financing round.

What is even more eye-catching is that the chip darling of the AI era, NVIDIA, and the venture capital division of Capital One have also made high-profile appearances on the list of co-investors. Capital One is the largest customer of Databricks' competitor, Snowflake, which has led to speculation about whether it will eventually switch sides and support Databricks.

NVIDIA and Capital One becoming strategic investors is particularly noteworthy, as it may have an impact on their competitor, Snowflake.

Tech media outlet TechCrunch commented that Databricks is actually not short of money at all, which makes this financing round look like both a financing round before an IPO and a deliberate introduction of strategic investors such as NVIDIA.

This financing round was led by existing investor T. Rowe Price, with participation from Morgan Stanley's Counterpoint Global, Fidelity, and Franklin Templeton, all of whom are investors who like to participate in financing rounds before IPOs. Other more traditional private market investors also participated in this round, including Andreessen Horowitz, Baillie Gifford, Tiger Global, and the Singapore government investment company GIC.

Ali Ghodsi, CEO of Databricks, expressed his excitement about the strategic partnership with NVIDIA to establish a customized large language model:

"This investment allows us to significantly strengthen our generative AI strategy, and closer collaboration is also meaningful. Essentially, we are in complementary markets with NVIDIA."

According to another tech media outlet, The Information, NVIDIA's external investment activities have "exploded" this year, with investments in at least 16 startups, enabling some of the invested companies to more easily obtain the in-demand GPU chips in the field of AI.

Databricks revealed that it plans to collaborate with NVIDIA to optimize the software for Databricks' sales, helping customers build AI models that run on NVIDIA-powered servers. At the same time, Databricks' software will still be able to run on chips from other providers, including those that primarily use CPUs and servers equipped with Amazon AWS's general-purpose Graviton chips. According to analysts, it is not difficult to understand why NVIDIA invested in Databricks. On one hand, this indicates that Databricks is accelerating the deployment of artificial intelligence capabilities, including building its own large language models, which is built upon its history of selling data analysis and machine learning software. NVIDIA is committed to promoting the application of AI in new markets, investing not only in systems and software, but also in partnerships.

At the same time, Capital One Financial Corporation has become an investor in Databricks, which may have a negative impact on Snowflake, its main competitor that has already gone public and in which Warren Buffett rarely participates in "new listings". Snowflake has repeatedly stated that Capital One is its largest customer, spending over $48 million annually, but Capital One has also been using Databricks' technology for fraud detection in the past two years.

Databricks is the most likely company to go public, but it refuses to disclose the timing and will use the new funds for strategic acquisitions.

Databricks was founded in 2013 and is headquartered in San Francisco, California. According to its second quarter data ending on July 31, this year's total revenue is expected to exceed $1.5 billion, an increase of about 50% compared to the previous fiscal year, which exceeded $1 billion. However, this growth rate may be slowing down, as the revenue growth in the previous fiscal year exceeded 60%, and some speculate that the company has been spending a lot of money recently.

The company claims to have over 10,000 customers worldwide, with more than 300 of them generating annual revenue of at least $1 million from its software and services. The CEO also insists that profitability can be achieved within two quarters if necessary. The second quarter achieved the "strongest quarter-over-quarter incremental revenue growth" in history. The company is currently more focused on organic growth and growth through acquisitions, and refuses to disclose the planning for IPO listing.

The CEO of the company, Ali Ghodsi, revealed that Databricks may use the funds from the latest financing round for more strategic acquisitions. Data companies are potential targets, but he refused to specify which specific areas in the "data industry" have the most promising prospects. He also mentioned that AI startups are now a bit too expensive, and the potential IPO is still on the company's development roadmap, just not at the moment.

Analysts have pointed out that since the end of 2021, there have been no notable venture capital-backed tech IPOs in the United States. Although Databricks hopes to go public and all internal compliance processes are in place, its valuation is 29 times its effective revenue, which seems a bit expensive for the current market. After 2021, Snowflake, its competitor, saw a 45% decline in market value, and other cloud software stocks also plummeted.

In June of this year, Databricks announced its acquisition of MosaicML, a startup company specializing in large language model training and deployment, for $1.3 billion. MosaicML can provide more natural AI speech, and the transaction has been completed today. Databricks announced this year that it will release the open-source large language model Dolly and acquire Okera, a data governance platform focused on artificial intelligence, but did not disclose the acquisition price.

Databricks emphasized that when discussing potential financing rounds with investors a few months ago, the "initial guidance amount did not exceed $100 million," but later, as more investors wanted to join the "AI territory," the final financing amount increased fivefold.

Due to economic uncertainty leading to reduced procurement spending by major customers, many enterprise software manufacturers are facing a slowdown in growth. Databricks emphasized that it has maintained a growth pattern and has not announced any layoffs. Most of the internal cost reductions come from reducing software subscriptions. "We spent $30 million to purchase 300 SaaS software, and now it has been cut in half."

NVIDIA and Snowflake also collaborate, Databricks admits: AI territory is being contested

Analyzing the business lines of Snowflake and Databricks, it can be seen that both companies are betting on AI through acquisitions, collaborations, and product development. Although their positioning is slightly different, they both provide storage, processing, and governance of data for paying customers in the cloud environment. They are also looking for ways to help customers build generative AI and other intelligent applications based on the data stored on their platforms.

In fact, NVIDIA and Snowflake also announced new collaboration projects this year to provide Snowflake customers with methods to build generative AI applications and run them on NVIDIA GPU chips.

Some people believe that in addition to OpenAI, large cloud providers such as AWS are also competitors of companies like Snowflake and Databricks. Databricks also collaborates with Microsoft, the "backer" behind OpenAI.

Databricks CEO Goldstein admitted that many companies in the industry are both customers and competitors, saying, "Now is the time to contest the territory in the field of artificial intelligence."