The data warehouse market is shaping up to become a battle of the elements as Firebolt, a cloud-native data warehouse startup, emerged from stealth today with $37 million in financing and an offering that claims to boast speeds hot enough to melt Snowflake’s hype into a springtime puddle. 

Snowflake’s record-setting initial public offering (IPO) as the first cloud data warehouse to decouple storage and compute not only blew the roof off the New York Stock Exchange earlier this year, but pioneered what is emerging as the dominant architecture for the modern cloud data warehouse. 

The Israel-based startup built its cloud-native data warehouse on a decoupled storage and compute architecture, and that is where – evolutionarily speaking – Firebolt starts and Snowflake ends. 

Firebolt took the best of Snowflake, improving its limitations in storage, indexing, and query optimization, and built a cloud-native data warehouse that, according to Firebolt CEO Eldad Farkash, is more cost effective for ad hoc interactive, high performance, semi-structured data, and operational or customer-facing analytics that require continuous ingestion.

In terms of performance, the newcomer claims to be 18,200% faster than any alternatives at the multi-terabyte and petabyte scale that delivers a new level of elasticity to data warehouses for greater control and choice over resources. 

The cloud-native data warehouse is also built on a consumption-based model. According to Farkash, a cloud data warehouse can only be truly elastic if customers can add and remove resources as needed. 

Firebolt, he claims, enables more users to extract more value at a fraction of the cost of the alternatives through greater efficiency, complete choice of resources, and 100% cost transparency.

Farkash contends that Firebolt is not just another data warehouse, particularly in how it tackles semi-structured data.

As an alternative to extract, transform, load (ETL-) like transformations that flatten data into columns or unstructured data, Firebolt’s offering features what the company calls rapid warehousing, which is really just an SQL query engine for semi-structured data with native array manipulation functions.

Farkash said he and Firebolt COO Saar Bitner, the former co-founders and executives of business analytics software company Sisense, co-founded the Israel-based startup in 2018 after deciding “it was obvious that there is no difference between a data lake and a data warehouse."

The lines between a data warehouse and a data lake have been blurred, and “today the discussion about those two models is mainly driven by where the companies who offer the product are coming from,” Farkash said in explaining how a data lake mindset is to Databricks as a data warehouse mindset is to Snowflake — both vendors tying category to product.

Instead, Firebolt’s cloud-native data warehouse as a service unites the two categories into one offering because its target base is “always starting with massive amounts of data,” according to Farkash. 

“We did not set out to solve yesterday's problems that existing data warehouses try to solve,” Farkash added. “Nor did we need or try to solve problems of scale because that is already behind us. What we did want to solve is what we believe will be the next biggest driver and motivator for using compute or data warehousing, which is efficiency and speed.”

Today, more data crosses the internet every second than was stored in the entire internet just 20 years ago, and how that data is collected, accessed, and analyzed can determine whether a business sinks or swims. 

“If you look at companies who switch from understanding their business to driving their business with data, using the traditional data warehousing or the traditional data lake approach just doesn't cut it anymore,” said Farkash.

Turning big data into insights under the traditional model, Farkash explained, is either painfully slow, expensive, or labor intensive, if not all of the above. Those companies, he added, are spending a fortune on building data lakes and pipelines, and cleansing the data only to realize that storing big data is different from analyzing big data. 

To that end, Firebolt is a complete redesign of the data warehouse for the era of the cloud and data lakes, Farkash said. “Our aim is to enable organizations to deliver an incredible data analytics experience regardless of the size and usage patterns of a company’s data without having to constantly be worried about performance and costs.”