As synthetic intelligence drives unprecedented demand for knowledge processing, Mountain View startup Voltron Information is providing an answer to one among AI’s least mentioned however most important challenges: transferring and remodeling large datasets rapidly sufficient to maintain up.
Voltron Information, which introduced a strategic partnership with Accenture at the moment, has developed a GPU-accelerated analytics engine that might assist enterprises overcome the information preparation bottleneck hampering AI initiatives. The corporate’s core product, Theseus, allows organizations to course of petabyte-scale knowledge utilizing graphics processing models (GPUs) as a substitute of conventional pc processors (CPUs).
“Everybody’s targeted on the flashy new stuff you could contact and really feel, but it surely’s that dataset basis beneath that’s going to be key,” stated Michael Abbott, who leads Accenture’s banking and capital markets observe, in an unique interview with VentureBeat. “To make AI work, you’ve bought to maneuver knowledge round at a pace and tempo you simply by no means needed to earlier than.”
Constructing for the AI tsunami: Why conventional knowledge processing gained’t lower it
The partnership comes as firms speeding to undertake generative AI are discovering their present knowledge infrastructure isn’t outfitted to deal with the quantity and velocity of information required. This problem is anticipated to accentuate as AI brokers change into extra prevalent in enterprise operations.
“Brokers will most likely write extra SQL queries than people did in a really brief time horizon,” stated Rodrigo Aramburu, Voltron Information’s Discipline CTO and cofounder. “If CIOs and CTOs are already saying they spend manner an excessive amount of on knowledge analytics and cloud infrastructure, and the demand is about to step perform increased, then we’d like a step perform down in the price of operating these queries.”
In contrast to conventional database distributors which have retrofitted GPU help onto present programs, Voltron Information constructed its engine from the bottom up for GPU acceleration. “What most firms have achieved after they’ve tried to do GPU acceleration is that they’ll shoehorn GPUs onto an present system,” Aramburu instructed VentureBeat. “By constructing from the bottom up…we’re capable of get 10x, 20x, 100x relying on the efficiency profile of a selected workload.”
From 1,400 servers to 14: Early adopters see dramatic outcomes
The corporate positions Theseus as complementary to established platforms like Snowflake and Databricks, leveraging the Apache Arrow framework for environment friendly knowledge motion. “It’s actually an accelerator to all these databases, slightly than competitors,” Abbott stated. “It’s nonetheless utilizing the identical SQL that was written to get the identical reply, however get there loads quicker and faster in a parallel style.”
Early adoption has targeted on data-intensive industries like monetary providers, the place use instances embrace fraud detection, threat modeling and regulatory compliance. One massive retailer decreased its server depend from 1,400 CPU machines to only 14 GPU servers after implementing Theseus, in accordance with Aramburu.
Since launching at Nvidia’s GTC convention final March, Voltron Information has secured about 14 enterprise clients, together with two massive authorities businesses. The corporate plans to launch a “check drive” model that can enable potential clients to experiment with GPU-accelerated queries on terabyte-scale datasets.
Turning the GPU scarcity into a chance
The present GPU scarcity sparked by AI demand has been each difficult and helpful for Voltron Information. Whereas new deployments face {hardware} constraints, many enterprises possess underutilized GPU infrastructure initially bought for AI workloads, property that may very well be repurposed for knowledge processing throughout idle durations.
“We truly noticed it as a boon in that there’s simply so many GPUs out out there that beforehand weren’t there,” Aramburu famous, including that Theseus can run successfully on older GPU generations which may in any other case be deprecated.
The know-how may very well be notably precious for banks coping with what Abbott calls “trapped knowledge” — info locked in codecs like PDFs and paperwork that may very well be precious for AI coaching however is troublesome to entry and course of at scale. “You’ve seen among the knowledge that Voltron would present you is probably 90% simpler and environment friendly to maneuver knowledge utilizing this know-how than normal CPUs,” Abbott stated. “That’s the ability.”
As enterprises grapple with the information calls for of AI, options that may speed up knowledge processing and cut back infrastructure prices are more likely to change into more and more crucial. The partnership with Accenture might assist Voltron Information attain extra organizations going through these challenges, whereas giving Accenture’s purchasers entry to know-how that might considerably enhance their AI initiatives’ efficiency and effectivity.
