Fragmented data sinks 70–80% of enterprise AI projects
Michael Cronin, Couchbase’s Vice President for Asia-Pacific, Japan, and India, told TechSparks 2026 that most enterprise AI projects fail before they ever reach production. The culprit? Fragmented data infrastructure. He shared how consolidating data layers—rather than piling on more APIs—can slash failure rates and why India is becoming a powerhouse in global AI innovation.
Editor, Lazyfounder
30 SEC SUMMARY
- Fragmented data infrastructure causes 70–80% of enterprise AI projects to stall before reaching production.
- Consolidating data layers, rather than adding more APIs, reduces failure rates and latency for AI systems.
- India is emerging as a global leader in AI innovation, with enterprises like Air India and Tesco using local teams to drive projects.
- An MIT study found 95% of AI projects never turn a profit, often due to siloed or unreliable data systems.
TABLE OF CONTENTS
KEY HIGHLIGHTS
- 70–80% of AI development projects never reach production due to fragmented data infrastructure.
- An MIT study found 95% of AI projects fail to become profitable.
- Tesco’s consolidated data layer handles millions of transactions per second with sub-millisecond latency.
- 80% of Couchbase’s top 50 global accounts now run significant operations in India.
- Air India’s next-gen entertainment platform uses Couchbase to deliver offline personalization.
- Machine traffic on the internet surpassed human traffic 18 months ahead of schedule.
The hidden cost of fragmented data
Speaking at TechSparks 2026, Michael Cronin, Couchbase’s Vice President for Asia-Pacific, Japan, and India, didn’t mince words. Most enterprise AI projects never leave the pilot phase, he said, because their data infrastructure is too fragmented to scale. Without a unified system, reliability and performance suffer.
Cronin pointed to industry data showing 70–80% of AI development projects never reach production. An MIT study reinforced the point, finding 95% fail to turn a profit. The root cause? Siloed data systems that can’t handle real-world demands.
Why APIs can’t save broken systems
Machine traffic has already surpassed human traffic on the internet—18 months earlier than predicted. Cronin argued this shift demands data layers that can keep pace with "machine-speed" volumes, governance, and latency. Siloed systems simply weren’t built for this scale.
He warned against the common fix: adding more API connectors to patch gaps. Each new connector, Cronin said, becomes another potential failure point. Instead, enterprises should consolidate their data layers, starting with their most critical use case.
Tesco’s playbook: start small, scale fast
UK retailer Tesco took this advice. It began with a single instance and use case in Bengaluru, consolidating its data layer before expanding. The result? Systems that now handle millions of transactions per second with sub-millisecond latency.
Cronin’s recommendation was clear: shorten the data path and bring intelligence closer to customers. A unified operational data layer, he said, does exactly that—eliminating the need to rebuild governance and performance frameworks for every new AI application.
India leads, not follows, in AI innovation
Couchbase’s customer roster in India includes Air India, Airtel, BigBasket, and Mahindra Finance. Of its top 50 global accounts, 80% now run significant operations in the country. These aren’t just execution hubs—India-based teams are leading AI innovation and business decisions.
Air India’s next-generation in-flight entertainment platform, built on Couchbase, exemplifies this shift. The system delivers personalized shopping and entertainment even when an aircraft is offline. Cronin called it proof that India’s role has evolved from cost-saving execution to driving global innovation.
What this means
Lazyfounder analysis — our interpretation, not reported fact.
Fragmented data doesn’t just slow AI down—it sinks most projects before they ever deliver value. Companies that consolidate their data layers early will outpace competitors stuck patching together brittle systems.
Key takeaways
- 70–80% of enterprise AI projects fail to reach production due to fragmented data infrastructure.
- Consolidating data layers improves scalability and reduces latency, as seen with Tesco’s sub-millisecond transaction speeds.
- India is now a hub for AI decision-making, not just cost-saving execution.
- Operational data platforms can eliminate the need to rebuild governance and performance layers for every AI use case.
Related on Lazyfounder
Sources
- YourStory · 2026-11-05
Why AI’s biggest challenge lies beneath the model
This story is an original summary drafted with AI by Lazyfounder from the reporting listed above and checked by automated validation. Facts are attributed to their original publishers; sections marked as analysis are Lazyfounder's. Where a source is in another language, facts were machine-translated and quotations are reported, not reproduced. Read the original coverage via the links, and see our AI policy and corrections policy.
About the author
Editor, Lazyfounder
Tarun Mottlia edits LazyFounders, covering Indian startups, funding rounds, AI and product launches. Every story on the site is AI-assisted and checked against its cited sources before publication.
More stories by Tarun MottliaGet the LazyFounder Brief
Startup, funding and AI news in a five-minute read. Join the early-access list.
