The Exponential Scale of Enterprise Data: Why DVC Invested in Ocient

Every enterprise we speak with wants to put AI to work on its own data, and while the models are ready, the systems holding that data mostly are not. That is why Dallas Venture Capital has invested in Ocient, the company building the data platform for AI on petabyte-scale enterprise data.

Data has outgrown the systems built to query it

Enterprise data is now doubling roughly every three years. IDC projected the global datasphere would reach 175 zettabytes by 2025, and 97% of 500 data leaders in a 2022 Ocient survey expected their data volumes to grow fast or very fast. For the largest enterprises, that means petabytes, with hundreds of millions of new events landing every hour.

At that scale, even a simple question is an engineering problem. Scanning a single petabyte at 10 gigabytes per second takes more than 27 hours, and at 100 gigabytes per second it still takes close to three. An analyst who needs an answer before the meeting ends cannot wait that long.

Most enterprises compromise. To control costs, they sample, pre-aggregate or delete history, and lose the detail that made the data worth keeping. Others stitch together a warehouse, a search engine, a Spark cluster and a geospatial store. That setup is slow, expensive and error-prone.

Agentic AI makes this worse: agents query far more often than people, and decade-old architectures cannot serve them affordably.

How Ocient stands out

Chris Gladwin co-founded Ocient in 2016 to build analytics for this scale from the ground up. Three design choices stand out.

Compute next to storage. Ocient puts processing directly next to high-speed NVMe drives, which gives it up to 420x the data bandwidth of remote-storage warehouse architectures, by Ocient's measure, using a tenth of the drives.

Indexing. Most systems rely on a single clustering key. Ocient builds a full set of secondary indexes at load time, so every query can take the fastest path to the data.

One platform. Ocient brings SQL, machine learning, geospatial and real-time analytics together. In one example, a question that took more than 786,000 tokens across three separate systems was answered with a single SQL query and 141 tokens. In another, a 5-petabyte workload that needed five racks and 115 kW ran on half a rack drawing 12 kW, roughly a tenth of the energy and cost.

Ocient also deploys where sensitive data must stay: on premises, in private cloud or in air-gapped environments, with fixed-capacity pricing that lets teams query continuously instead of rationing expensive queries. Ventana Research analysts have noted that it can run ad hoc analytics on hyperscale workloads.

Why DVC invested

Team. Chris Gladwin previously founded Cleversafe, the storage company IBM acquired for $1.4 billion. Many of Ocient's top 20 people built Cleversafe with him, and the wider team of about 150 comes from IBM, Teradata, AWS, HPE, Vertica and Google. CEO John Morris now leads the commercial push.

IP. Ocient has filed more than 500 patent applications, with 169 granted, the product of over 600 person-years of development. That is a deep moat.

Wins and traction. With one customer, Ocient cut query times by 96% and costs by 80%, paying for itself in four months. A Tier 1 US telco chose Ocient after other platforms failed its acceptance tests.

Another consolidated ten workloads onto one cluster, cutting query times from 24 hours to minutes. Several customers are on their seventh to tenth purchase order.

Market. Ocient competes in hyperscale analytics, a specialized layer of the data stack with very few purpose-built commercial systems. The market is still early, but we expect it to grow with every AI agent an enterprise puts into production, which is why Ocient fits our AI picks-and-shovels thesis. Ocient is focused on communications service providers, AdTech and national security, and is backed by Valor, OCA Ventures, Greycroft and In-Q-Tel.

Looking ahead

We think the next decade of enterprise AI will belong to the companies that can keep all of their data, query it in seconds and afford to do both. That is what Ocient is built for, and we are proud to welcome the team to the DVC family.

‍

Author
Dayakar Puskoor
Founder & Managing Director
Dallas Venture Capital