Doing Data Analytics acceleration in hardware with NeuroBlade

Becoming 'the NVIDIA of Data' is NeuroBlade's goal. Refreshing to see companies, especially at a Cloud Field Day, focus on hardware. Let's not forget that all software has to run somewhere, and data analytics is no exception.

Hardware, in this case, is meant to accelerate data analytics, trying to solve the scaling challenges of general-purpose processors and GPUs. They just weren't built for data analytics and SQL in particular, making the economics not work in its favor, including cooling costs, and power usage.

Hence, the need for hardware acceleration for SQL.

This device promises 30x performance at one third of the cost of comparable CPUs at a fraction of the power consumption, and supports many of the popular data analytics stacks through integrations. While not all data stacks are supported, SQL queries do not need to be re-written. Customers can also opt to create their own integration using NeuroBlade's API/SDK.

Consuming the Hardware

The company has OEM integrations with Dell currently, but the card is also available as a stand-alone purchase. As a small hardware startup, go-to-market plays to make the card available through cloud consumption has not materialized. The company's clients are mostly 'hyper-scale' themselves, with large private clouds.

What's next for NeuroBlade?

Looks like this solution is great for those enterprise clients that struggle with the economics of on-prem, bare-metal data analytics performance. They seem to be missing large market segments, including offering their card to data SaaS offerings like DataBricks and Snowflake as well as those end customers running in cloud.

For a young hardware company like NeuroBlade, that doesn't seem out of the order. The question is, will they be able capture the rest of the market through partnerships with cloud hyperscalers and data analytics SaaS providers? Let's revisit in a couple month's time.