Todd Mostak
Dec 23, 2016

The Janus Post

Try HeavyIQ Conversational Analytics on 400 million tweets

Download HEAVY.AI Free, a full-featured version available for use at no cost.


2016 was a pretty amazing year for MapD. Not only did we launch our company with the announcement of our A Round of funding in late March, but we were able to steadily build on that event throughout the year, culminating in the release of our 2.0 version of the product just nine months later.

In the interim, we were fortunate to pick up some prestigious awards including Gartner Cool Vendor, Fast Company Innovation by Design, The Business Intelligence Group’s Startup of the Year, CRN’s 10 Coolest Big Data Startups and Barclays Open Innovation Challenge.

We also saw a series of powerful independent benchmarks published by noted database authority, Mark Litwintschick. To date, we remain the fastest platform (by 75x) that he has ever tested against the 1.2 billion row NYC taxi dataset.

We also revealed an addition to our A round in October - that of In-Q-Tel. The venture arm of the CIA is considered one of the savviest investors in Silicon Valley, executing deep diligence and providing guidance to navigate the complex government and intelligence communities. We were pleased to be in a position to publicly welcome them to the fold.

Finally, this year was marked by ubiquity in the cloud for GPUs. When the year began, only IBM had instances of enterprise compute grade GPUs available. Over the intervening months, IBM upgraded its instances, AWS introduced the state of the art instances and Google Cloud followed introduced GPU instances with specific deep learning attributes. We were fortunate to be included in the private trials and launch messaging of all three of these announcements - something, that like the awards above, tell us that we are on the right path.

While we will fondly remember 2016, we are looking forward to 2017 with incredible anticipation.

As Nvidia’s 3x growth in their datacenter revenues attest, GPUs are no longer exotic compute resources in the enterprise. That growth may very well be trumped by Nvidia’s, super computer in a box, DGX-1 whose demand is through the roof. While deep-learning is a core use case for the DGX-1, most organizations don’t have the internal resources for large scale AI at this point. What they do have is data, too much to handle with legacy CPU compute. As a result, we are already seeing demand for GPU-powered databases and visual analytics for these on-prem instances.

This realization, that their data collection and storage capabilities have far outstripped their insight extraction capabilities, is rooted in their compute platforms.


GPU powered analytics not only offer a way to get in front of this computationally, they also offer far superior economics.

We recently did a webinar with Nvidia where they showed that the time to value for the DGX-1 is a tiny fraction of X86 CPU solutions.

3 year comparison

This point, vastly superior price/performance characteristics, is what will bring GPUs into view for business owners and CFOs in the coming year and will complete the GPUs journey from niche to ubiquitous.

So join us as we raise a glass to 2016 and look forward into 2017 with the knowledge that for us in the GPU space - it will be a very, very good year

Todd Mostak

Todd is the CTO and Co-founder of HEAVY.AI. Todd built the original prototype of HEAVY.AI after tiring of the inability of conventional tools to allow for interactive exploration of big datasets while conducting his Harvard graduate research on the role of Twitter in the Arab Spring. He then joined MIT as a research fellow focusing on GPU databases before turning the HEAVY.AI project into a startup.