- Fragmented data undermines AI competitive advantage, IBM storage GM Sam Werner told Fierce
- Telcos face the data fragmentation problem acutely, with long histories and enormous data stores — but the opportunities are great as well, Werner said
- IBM and Nvidia are partnering on storage, analytics, document processing, hybrid cloud and consulting
The tech economy celebrates startups and dismisses old companies. But companies with long histories in business have valuable stockpiles of customer and business practice knowledge to draw on for competitive advantage — if only companies can access that knowledge.
"If you can figure out how to bring data to AI, you create a moat for your company that's very hard to penetrate," said Sam Werner, IBM storage general manager, in an interview with Fierce Network at the Nvidia GTC conference in mid-March.
But that data is inaccessible today, spread throughout the organization in multiple generations of data warehouses, databases and petabytes of unstructured data. "There's no way to get all that data to your AI models without a data management strategy that allows you to bring it forward," Werner said.
Most organizations try to get around the problem by making copies of data and bringing the copies to where they are needed. But the copies grow out-of-date quickly and present security and governance challenges, Werner said.
A big problem — and opportunity — for telcos
The situation is particularly acute for telcos, some of which have histories stretching back literally to the 19th century. Telcos are attempting to leverage data and implement AI to improve internal operations and deliver new services to clients. Much of the telco data resides at the network edge. Telco networks can help enterprises unify data and enable hybrid cloud environments needed for AI, Werner said.
Particularly outside the U.S., telcos are transitioning to AI providers for enterprises, leveraging their networks, technical expertise, data centers and long history with enterprise customers and industry knowledge.
IBM is attempting to help organizations manage data for AI insights, with a partnership it announced at the Nvidia GTC conference on March 16, combining IBM storage, data analytics, intelligent document processing, hybrid cloud and consulting services with Nvidia GPUs. The partnership is designed to help organizations move AI from experimentation to production at scale.
Nestlé gets sweet results
Nestlé worked with IBM and Nvidia to accelerate the performance of the company's Order-to-Cash data mart, which tracks every order, fulfillment, delivery and invoice across 186 countries, processing terabytes of data. Nestlé reduced refresh times from 15 minutes down to three minutes, achieving 83% cost savings and an overall 30x price-performance improvement, IBM said. (Nestlé did not respond to requests from Fierce for confirmation.)
For analytics, IBM's watsonx.data — a data and AI platform — uses Nvidia GPU acceleration to dramatically speed up SQL queries against large enterprise datasets, as demonstrated in the Nestlé work.
IBM and Nvidia are also collaborating on transforming unstructured data, such as PDFs and spreadsheets, into formats AI can use, combining IBM's open-source Docling tool for converting unstructured documents into structured data, with Nvidia Nemotron language models.
Additionally, Nvidia and IBM are working to provide Nvidia Blackwell Ultra GPUs through IBM Cloud. IBM will also provide AI Factory with Nvidia on Red Hat OpenShift.
The IBM-Nvidia alliance comes as Nvidia seeks to pivot from training to inference, transitioning from chip vendor to a full-stack AI provider, including chips, hardware and software. At the GTC conference last month, Nvidia unveiled the Vera Rubin platform — seven new chips in production — with improved inference price-performance over Blackwell, as well as launching as a partnership with T-Mobile to push physical AI to the network edge via AI-RAN. Nvidia also debuted Nemoclaw, an enterprise-ready version of OpenClaw, which Nvidia CEO Jensen Huang said is growing faster than Linux and "essentially the operating system for agentic computers."
Fierce Network Research's take
IBM faces tough competition in data management. Most directly, Dell's AI Factory with Nvidia , which Dell upgraded at GTC, takes a similar integrated approach spanning storage, compute and software. Storage vendors, including NetApp, HPE, Pure Storage and VAST Data, are all positioning their platforms as AI data pipeline foundations, most with their own Nvidia partnerships. And the cloud hyperscalers — AWS, Azure and Google Cloud — offer end-to-end AI data infrastructure for enterprises willing to go all-in on public cloud.
IBM's competitive advantage is that it can manage data across on-premises, edge and cloud environments without forcing migration, backed by consulting depth through IBM Consulting, targeting challenges that the pure-cloud and pure-storage players don't fully solve.
Check out the latest free report from Fierce Network Research: "Risk, reward, and revenue: Defining the telco role in the AI economy" to learn more about this topic.