AI glossary Data & Retrieval
What is a data lake?
← All glossary termsA central repository that stores structured and unstructured data in its raw form, as a source for analytics, machine learning and AI.
Explained
A data lake is a centralised repository that stores large amounts of structured, semi-structured and unstructured data in its original form. Unlike a data warehouse, it does not require a schema to be defined before data is loaded; structure is applied when the data is read. This makes it a flexible source for analytics, machine learning and AI applications that need varied data, such as documents, logs, images or event streams. Without cataloguing, access control and quality rules, a data lake quickly becomes hard to use.
Data lake, data warehouse or lakehouse?
- A data warehouse stores cleaned, structured data with a schema designed in advance. It is optimised for reporting and business intelligence.
- A data lake stores raw data of any type at low cost and applies structure when reading it. It suits exploration, data science and machine learning.
- A lakehouse combines both: open file formats on cheap storage, with the transactions, schemas and query performance of a warehouse on top.
- Many organisations use more than one, with the lake as the landing zone and curated layers or a warehouse for reporting.
Why it matters for AI
AI projects often need data that never fitted a warehouse: contracts, emails, call transcripts, images, sensor readings. A data lake keeps that material in one place, in its original detail, so it can be used to train or fine-tune models, build retrieval indexes or create features for machine learning.
The raw form is also its weakness. Models trained or grounded on poorly understood data inherit its errors and gaps, so lineage, quality checks and permissions matter as much as storage.
Avoiding the data swamp
- Catalogue what enters the lake: source, owner, meaning and sensitivity.
- Organise it in layers, from raw to cleaned to curated, with clear rules for each.
- Apply access control at the right level, especially for personal and confidential data.
- Track lineage, so you know where each dataset came from and what uses it.
- Remove what nobody uses or is no longer allowed to keep.
Frequently asked questions
What is the difference between a data lake and a data warehouse?
A data warehouse holds structured, processed data with a schema defined before loading, and is built for reporting. A data lake holds raw data of any type and applies structure when the data is read, which makes it more flexible for data science and AI but harder to govern.
What is a data swamp?
A data swamp is a data lake that has lost its usefulness because nobody knows what is in it, where it came from or whether it can be trusted. It is usually the result of loading data without a catalogue, ownership or quality rules.
What is a data lakehouse?
A lakehouse is an architecture that adds warehouse capabilities, such as transactions, schemas and fast SQL queries, on top of data lake storage. The aim is to serve both analytics and machine learning from one copy of the data.
Where is a data lake usually built?
Most data lakes today are built on cloud object storage, with separate services for cataloguing, processing and querying the data. They can also be built on premises with distributed file systems.
Do you need a data lake to do AI?
Not always. A focused use case can start with the data it needs from existing systems. A data lake becomes valuable when several use cases need varied, large or unstructured data and you want one governed place to find it.
Decision guides that use this term
How to put it to work
Where it shows up at Thinkia
Further reading
Related terms
Sources
- AWS · What is a data lake?
- Gartner · Definition of data lake (IT glossary)
- Armbrust et al. · Lakehouse: a new generation of open platforms that unify data warehousing and advanced analytics (CIDR 2021)
Updated: 2026-10-21