- AI ROI is lagging for most enterprises piloting the technology
- A handful of companies - those who have prioritized an open, cloud-native and portable architecture - are ahead of the pack
- Vendor lock-in could end up being a key limiting factor for many companies chasing their AI dreams
We’ve all seen the dire statistics: depending on which study you read, anywhere from 80% to 95% of enterprise GenAI pilots are failing to generate a significant financial impact for the companies deploying them. But why?
Opinions on the reason behind this flop are mixed, ranging from too heavy an emphasis on horizontal use cases (like chatbots) to tools that are poorly integrated and don’t match workflows. But EnterpriseDB has a different answer: vendor lock-in.
“Lock-in is the silent killer of AI ROI,” EDB Chief Product Officer Nancy Hensley told Fierce. “If all your most precious workloads live in one hyperscaler’s walled garden or a closed-source stack, you’re stuck with their prices, their pace of innovation and their compliance limitations. That means slower deployment, higher switching costs and less leverage when you need to move fast.”
According to a new report from EDB, those achieving the most ROI from their AI deployments are so-called “Deeply Committed” enterprises that have focused on establishing a cloud-native, open, portable data foundation on which AI applications can be built and managed. Only 13% of the 2,000-plus companies surveyed for the report are in this category, but these are getting five times the AI ROI as companies in other categories.
The report uses the word sovereignty a lot to describe this open, independent architecture, but Deeply Committed companies aren’t limited to those in countries with strong digital sovereignty laws.
Admittedly, this take is a bit self-serving coming from EDB – the company is a key contributor to the open source Postgres community and offers Postgres-based software and services to enterprises. But that doesn’t mean EDB is wrong about lock-in being a bad thing.
DE-CIX called out the same issue in an interview with Fierce in 2023. We’ve also highlighted the importance of sovereign cloud and AI and open source models in other recent stories. SAP, Backblaze and others have likewise made the point that vendor lock-in can kill AI innovation.
The pressure is on to avoid the trap.
“Enterprises that remain locked in aren’t just a little behind, they’re structurally limited,” Hensley warned. “They will hit a ceiling on ROI because they can’t move data and workloads freely, can’t respond to regulatory shifts quickly, and can’t scale AI where it’s most valuable.”