Dive Brief:
- Legacy architecture is failing to keep up with AI's demands, according to a report from cloud data management provider Cloudera. The company surveyed 1,500 global enterprise architects, data architects and cloud infrastructure leads for the Tuesday report.
- Nearly three-quarters of enterprises said that their infrastructure needs to be revamped to support AI. The majority of businesses have been forced to delay or cancel AI initiatives in the past year due to data governance, compliance and regulatory challenges posed by legacy systems, the report said.
- “Many enterprises are discovering that the architectures built for traditional analytics weren’t designed for the scale, governance, and flexibility AI demands today,” Cloudera CTO Sergio Gago said in a press release accompanying the report.
Dive Insight:
Flexible, hybrid infrastructure is emerging as the critical standard underpinning enterprise success with AI.
Data and infrastructure flexibility allows IT leaders to experiment with new models, prevent vendor lock-in and address evolving regulatory risks, according to a July report from consulting firm West Monroe. Building flexible data foundations and infrastructure to support AI capabilities also enables enterprises to advance innovative projects, Erik Brown, senior partner of technology and experience at West Monroe, told CIO Dive in an August interview.
Over the next two years, 1 in 4 enterprises will prioritize hybrid-first infrastructure, “reinforcing that the future of enterprise AI will be defined, in part, by flexibility rather than a single infrastructure strategy,” the Cloudera report said.
Two-thirds of businesses said they have already transitioned AI workloads from public cloud environments to on-premises or private cloud offerings, running workloads in the location they perform best, according to Cloudera.
Cost is another driving force behind the transition to hybrid architecture, as 84% of respondents said AI workloads have caused higher infrastructure costs.
“Organizations will prioritize moving data into multiple environments to get the most out of AI models,” the Cloudera report said. “Ultimately, it’s a shift that requires building flexibility to run the right workload in the right place while maintaining visibility, governance, performance, and cost control across the entire data landscape.”
Liberty Mutual Insurance is one example of a company prioritizing flexibility. The insurance giant is taking a hybrid approach to AI model providers, building a model-agnostic abstraction layer on top of its mainframe data, Andrew Palmer, EVP and CIO of global retail markets at Liberty Mutual, told CIO Dive in a June interview.
