Definition

Enterprise Data Technical Debt

The accumulated cost of maintaining, working around, and compensating for outdated, disorganized, or incompatible data infrastructure. Analogous to software technical debt but measured at the data layer rather than the code layer. Includes the ongoing cost of maintaining legacy mainframe systems that cannot be retired without major migration, the cost of manual data cleaning and reconciliation work performed by human analysts, and the cost of enterprise AI projects that underperform because data quality cannot support the intended use case. The Wall Street Journal cites this as a $1.52 trillion annual problem in the US alone. David Carmel uses the metaphor of Victorian era plumbing: the pipes are old, they work after a fashion, but they are fundamentally incompatible with the modern infrastructure being built around them.

From the episode “The $1.5 Trillion AI Problem Nobody Talks About Until Now”

Hear it on the podcast1 episode

The $1.5 Trillion AI Problem Nobody Talks About Until Now

Learn the language of AI-native founders

Every week: the new terms, tools, and workflows founders are using, straight from the podcast.

Related terms in Product & Engineering

eBOMEngineering Bill of Materials — the full, structured list of every part in a product and how they relate. The eBOM is the source of truth for what actually gets manufactured.Edge DeviceAny computing hardware that processes data locally without sending it to a remote server or cloud. Examples include smartphones, drones, and the Key Boy. Critical for latency-sensitive and privacy-sensitive applications.ELT PipelineExtract, Load, Transform. The process of moving raw data into a warehouse before transforming it — one of the four highest-value use cases for data agents.Engineer vs. CoderEngineers are system architects — they decide what to build, how components fit together, what problems the architecture needs to solve. Coders implement those specifications. The distinction matters because coders can increasingly be replaced by AI that executes specifications quickly and cheaply; architects cannot. Founders who staff with coders instead of engineers get fast execution without coherent architecture. The correct hire for a technical team is engineering judgment, not implementation capacity.EpicThe highest-level planning unit in the software development lifecycle: a large body of work defining a significant product capability. An epic breaks into multiple requirements; each requirement breaks into atomic tasks. In AI-native development, epics define multi-feature projects (build a multi-tenant Pilates platform with video libraries, subscriber management, and per-studio payment systems) while requirements define individual features within that project. The hierarchy matters: agents given an epic-level prompt will attempt to solve the entire thing in one shot and fail; agents given an atomic task will succeed consistently.Epic / Milestone / TaskDescending units of work — epics are big features, milestones are subfeatures with user stories and success metrics, tasks are small scoped changes.ERPEnterprise resource planning software that runs core operations such as finance and supply chain.ETL PipelineExtract, Transform, Load - a data engineering pattern describing the three-step process of (1) extracting data from source systems, (2) transforming it into a consistent, clean format, and (3) loading it into a destination system for analysis or use. ETL is the foundational workflow of enterprise data infrastructure and has been central to data engineering since the 1970s. Modern AI ETL pipelines extend the traditional pattern by using AI models in the transformation step to process unstructured inputs - documents, images, audio - that traditional ETL could not handle. Mehul Shah's background includes building AWS Glue, Amazon's flagship ETL service, and OpenSearch, a distributed search and analytics engine built on Elasticsearch.

Browse all 1,456 terms