Investors Bullish on GPU-Based Database Startup
MapD Technologies, the big data analytics platform startup developing a parallel SQL database that runs on GPUs, has more than doubled its venture-funding total with the close of its latest investment round led by New Enterprise Associates (NEA).
San Francisco-based MapD, which leverages GPU technology to speed its SQL query engine, said Wednesday (March 29) it latest funding round garnered $25 million from NEA and existing investors Nvidia (NASDAQ: NVDA), Vanedge Capital and Verizon Ventures. The Series B round brings MapD’s total venture funding to just over $37 million.
The analytics startup said it would use the funding to accelerate analytics platform development as it seeks to make inroads in the enterprise big data market. The company uses Nvidia’s GPUs to run enterprise applications that include machine learning and numerical computations. An upgraded platform will expand data analytics capabilities.
The startup’s GPU-based SQL query engine platform combines with data visualization designed to allow analysts and data scientists to crunch multi-billion-row data sets. “GPU-powered analytics is going to radically change the data analytics market,” predicted Greg Papadopoulos, a venture partner at NEA.
Since announcing a Series A funding round in March 2016, MapD said it has unveiled new query engine features, including the second version of its Core database and Immerse visual analytics platforms. Meanwhile, Amazon Web Services, Google, Microsoft Azure and IBM SoftLayer have all launched GPU cloud instances that have increased enterprise access to the MapD platform.
Amazon (NASDAQ: AMZN) Web Services announced last fall it would offer public cloud services based on the Tesla K80 GPUs from MapD investor Nvidia.
Founded in 2013, MapD Technologies originated from research at the MIT Computer Science and Artificial Intelligence Laboratory. Other seed investors include In-Q-Tel, the CIA’s venture capital arm.
The startup places heavy emphasis on leveraging fast hardware to maximize speed. “The first thing is we try to cache the hot data across multiple GPUs,” MapD CEO Todd Mostak stressed in an interview last fall. “We’re a column store. We’re compressing the data, so you can have many, many billions of rows in that GPU memory.”
MapD investor and GPU leader Nvidia has been pitching its latest GPU technology as a way to speed SQL workloads. In one example, MapD’s platform fuses its GPU-based database with a collection of visualization tools to enable users to work with huge geospatial data sets.
Another area ripe for GPU-backed databases is Internet of Things deployments that continue to generate huge troves of data. Mostak has argued that current CPU-based approaches won’t scale as new requirements emerge for merging streamed data and historical analysis extending out to 90 days.
Mostak predicted recently that infrastructure-heavy industries such as telecommunications would be among the early adopters of the GPU-based analytics. That prediction is supported by early investments by Verizon Ventures.