Key takeaways
- DynamoDB shows each table's item count and size in the console and in describe-table, and the summed TableSizeBytes, in GB across every Region, is the number to report.
- DynamoDB updates the counts only about every six hours, and billing adds 100 bytes of overhead per item that describe-table does not include.
- Exports to S3 need point-in-time recovery; a full export is a snapshot at one moment, while incremental exports cover windows of 15 minutes to 24 hours.
- You don't need to export anything to apply: have your size in GB, years of history and account count ready, and the data vendor anonymizes the records before any AI lab sees them.
01
What is Amazon DynamoDB data, and why do AI labs and data vendors buy it?
The short answer
Amazon DynamoDB data is the set of items that a company's application keeps in DynamoDB, Amazon's database service: sessions, orders, device readings, messages and events, each stored under a key such as a customer and a time. AI labs (companies that build and train AI models) and data vendors (companies that anonymize business records and package them for those labs) buy a copy because the items record what an application's users and systems actually did, in great volume.
What a DynamoDB table records
Amazon DynamoDB is a database service run by Amazon Web Services (AWS). Its data lives in tables, and each table holds items, which are individual records made of named attributes, such as an order with its date, its status and its customer. Every item is found through its key, a partition key and optionally a sort key, so a table keyed by customer and time can return everything one customer did, in order.
Many companies use DynamoDB for the busiest paths in their software, where speed matters more than reporting. A single AWS account can hold tables in several Regions, which are the separate geographic locations in which AWS runs its data centers.
Why application records teach an AI model something
An AI model learns by studying a very large number of examples and finding the patterns in them, and those examples are called training data. A table of events keyed by customer and time shows how people moved through a product: what they started, in what order, and how it ended. Public text on the internet has no such records, and they cannot be rebuilt afterward.
AI labs now want to train AI agents, software that carries out a job step by step, and an agent learns a job from records of that job being done. Our guide AI training data: why AI companies want your company's data explains the mechanism.
How DynamoDB fits into a dataset
A DynamoDB export is usually one of several sources in a company's dataset, next to email, documents and tickets. Companies are paid $100K to $2M or more per dataset, according to micro1, the data vendor we typically refer companies to. The price covers the whole dataset, not each source on its own, and what your data is worth explains what moves it.
02
Where does Amazon DynamoDB show how much data you have?
The short answer
DynamoDB shows the item count and the size in bytes of each table, in the console on the table's Overview tab and in the output of the describe-table command. DynamoDB refreshes both numbers only about every six hours. Add the table sizes together across every Region you use and report the total in gigabytes (GB).
How to read your table sizes
A developer or AWS administrator (the person who manages the company's AWS account) does these steps. The fields are defined in the DynamoDB TableDescription reference, and no special plan is needed.
- Make sure the AWS user or role you use may call
dynamodb:ListTablesanddynamodb:DescribeTablein each Region that holds tables. - In the DynamoDB console, go to Tables and open one. Its Overview tab lists the item count and the table size.
- Or use the AWS command-line tool. List the tables in a Region, then ask for the figures of each one:
aws dynamodb list-tables aws dynamodb describe-table --table-name MyTable --query "Table.[ItemCount,TableSizeBytes,CreationDateTime]"
Repeat both commands in every Region you use. - Add up the
TableSizeBytesvalues and divide by 1073741824 to get GB. Recent writes may not show yet, because the numbers update about every six hours.
What each figure counts
| Figure | What it counts |
|---|---|
TableSizeBytes | The total size of the table in bytes. This is the figure to add up and report. |
ItemCount | The number of items in the table. |
IndexSizeBytes | The size of one global secondary index (an extra copy of chosen attributes kept under a different key for faster searching). Reported separately, so leave it out. |
CreationDateTime | When the table was created, which gives its start date. |
Why the totals need a careful reading
DynamoDB adds 100 bytes of overhead per item when it bills for storage, and DescribeTable does not include that overhead, so your bill will show a larger size than the number to report. Replicas in global tables, which keep copies of one table in several Regions, are copies of the same data, so count each global table once.
03
What should you know about your DynamoDB data before you apply?
The short answer
Have three facts ready: the summed table size in GB, how many years of items your tables hold and how many people or accounts the items describe. They go into the data manifest, the written summary of which sources hold your records, how many years each covers and roughly how much. You don't need to run an export to find them.
The size in GB
Report the summed TableSizeBytes in GB. Leave out the global secondary indexes, which report their own IndexSizeBytes, and count a table once even when global tables replicate it into other Regions.
The years of history
CreationDateTime gives each table's start date, and the oldest one is a first estimate. If the tables hold a date attribute, the earliest value in it is more accurate, because a table can be rebuilt long after the activity it describes began.
The number of seats
For a database, the seat count is the number of users or accounts whose items the tables hold. The AWS users who log in to the console are not seats.
What you don't need to do first
Applying shares no data with anyone, and you don't need to export or clean anything first. The data vendor anonymizes records before any AI lab sees them, and nothing moves until a lawyer-reviewed license is signed. The application and each later step are laid out in how to sell your company's data.
04
How do you export your full Amazon DynamoDB history?
The short answer
A developer turns on point-in-time recovery for the table, then requests an export to an Amazon S3 bucket from the console (Exports to S3 > Export to S3) or with aws dynamodb export-table-to-point-in-time. A full export writes the table as gzip-compressed files in DynamoDB JSON or Amazon Ion. It is a snapshot at one moment, not a history of changes, and it does not use the table's read capacity.
Step 1: turn on point-in-time recovery
Point-in-time recovery (PITR) is a DynamoDB feature that keeps continuous backups of a table, so that it can be restored or exported as it stood at any moment in a recovery period of 1 to 35 days. DynamoDB's export guide says an export request for a table without PITR fails, and PITR is billed while it is on.
- In the console, open the table, and on its Overview tab, in Table details, choose Enable next to Point-in-time recovery. Some console versions also show this setting on the table's Backups tab, under Point in Time Recovery, with an Edit button.
- Or turn it on from the command line:
aws dynamodb update-continuous-backups --table-name MyTable --point-in-time-recovery-specification PointInTimeRecoveryEnabled=true
Step 2: request the export
The person requesting the export needs permission for the DynamoDB action dynamodb:ExportTableToPointInTime and for writing to the destination bucket (s3:PutObject, s3:PutObjectAcl and s3:AbortMultipartUpload). The request must come from the same AWS account that owns the table, although the bucket may belong to another account or Region.
- In the DynamoDB console, choose Exports to S3 in the left navigation, then Export to S3.
- Choose the source table and the destination bucket. A bucket outside your account is entered as
s3://bucketname/prefix, where the prefix is an optional folder name. - Choose Full export and select the point in time, which can be any moment inside the PITR window, or Current time for the latest snapshot.
- Under Exported file format, choose DynamoDB JSON or Amazon Ion, then choose Export.
The same request from the command line uses the table's ARN, its unique Amazon identifier, and the bucket name:
aws dynamodb export-table-to-point-in-time --table-arn arn:aws:dynamodb:REGION:ACCOUNT:table/MyTable --s3-bucket my-bucket --export-format DYNAMODB_JSON
Step 3: wait for the export and find the files
An export shows the status PENDING until it succeeds, when it becomes COMPLETED, or fails, when it becomes FAILED with a reason. The files appear under AWSDynamoDB/ExportId/ in your bucket, where the data files are gzip-compressed files in a data folder. The file manifest-summary.json records the item count, and manifest-files.json lists the data files. The console keeps the list of past exports for 90 days, though the files in your bucket stay.
Full and incremental exports
A full export captures the table exactly as it stood at the chosen moment. An incremental export instead lists the items that were inserted, updated or deleted over a window of 15 minutes to 24 hours, and its summary can show both the old and the new version of each changed item.
Limits, cost and what an export leaves out
An export can only use a point in time inside the table's recovery period. Exports run in the background, take no read capacity from the table and leave its speed and availability untouched. A full export is charged by the size of the table at the chosen moment, plus Amazon S3 storage and requests for the files it writes.
The data is not transactionally consistent, which means that one multi-item transaction can be split across two exports. A full export holds one snapshot and no earlier values, so history comes only from a series of incremental exports. Requester-pays S3 buckets are not supported as a destination.
| Export | Covers | Window |
|---|---|---|
| Full | A snapshot of the whole table at one moment | Any point in the PITR period (1 to 35 days) |
| Incremental | Items changed, updated or deleted in the period | 15 minutes to 24 hours, within the PITR period |
How long an export takes
AWS gives no guarantee of completion time. It depends on the table size, how the data is spread across the table and the other exports running, so don't build a schedule that assumes a fixed duration.
Sources: Amazon DynamoDB, API Reference: TableDescription, Data export to Amazon S3: how it works, Requesting a table export and Table export output format. Steps checked against these pages in November 2026.
05
What stays private when DynamoDB data is licensed?
The short answer
The company decides which tables and attributes leave DynamoDB, and the export lands in a bucket the company controls, so sensitive attributes can be removed before any file is shared. The data vendor then replaces names and other personal details with consistent stand-ins before an AI lab sees anything. The company keeps ownership of its data, and the license covers defined records for a defined use.
You choose what is in scope
Exports go to an Amazon S3 bucket that you choose, so tokens, passwords and personal attributes can be stripped from the files before anything is shared. DynamoDB can also write a filtered export, which uses a filter expression to export only some items, and a projection expression to choose which attributes appear in the output. The data vendor deletes records too sensitive to clean safely instead of anonymizing them.
Names are replaced before a lab sees them
Anonymization means replacing every detail that identifies a person or a company, such as a name or an email address, with a stand-in that is the same every time it appears. The items still link to each other and their timestamps still run in order, so a lab can learn from them, but they no longer say who anyone is. The method, including what is deleted outright and how you review samples, is in how company data is anonymized for AI training.
A short example shows the idea on a DynamoDB item. An order item that carries a customer's name and email keeps its order number, its dates and its status, while the name and email become labels such as "Contact 203" and "contact203@example.test".
Ownership and payment
Licensing gives the buyer limited rights to a copy, while the company keeps ownership of the originals. The terms that matter, such as exclusivity, permitted use and deletion, are covered in data licensing, and a lawyer's review of what may be shared is covered in whether it is legal to sell company data. micro1 pays in full within 30 days of accepting the data, usually as a single payment.
Questions
Frequently asked questions
Why hasn't the DynamoDB item count changed after I added data?
DynamoDB refreshes ItemCount and TableSizeBytes only about every six hours, so recent changes may not show yet. Check again later in the day.
Is point-in-time recovery required before a DynamoDB export?
Yes. DynamoDB rejects an export of any table that lacks point-in-time recovery, and PITR carries its own charge while it is on. Switch it on before you ask for the export.
Will a DynamoDB export slow down my table?
No. Exports run in the background, draw nothing from the table's read capacity and leave its speed and availability untouched. You pay for the export by the table's size, plus the S3 storage and requests for the files.
Does a DynamoDB export include the history of changes to each item?
No. A full export captures the table at a single moment. Incremental exports record changes over windows of 15 minutes to 24 hours, with old and new versions of each item, but they only reach back as far as the PITR period of 1 to 35 days.
Do I need to export DynamoDB before I apply?
No. The application shares no data. You only need the summed table size in GB, your years of history and your number of accounts, and describe-table gives you the first. The export is needed later, after you sign the license.
Can I export a table that belongs to a different AWS account?
The export request has to be made from the account that owns the table, so a person in another account first assumes a role in the table's account. The destination bucket may be in any account, as long as its policy allows the write.
What is DynamoDB data worth to AI labs and data vendors?
A database is priced as part of a whole dataset rather than on its own. Companies are paid $100K to $2M or more per dataset, according to micro1, and the price depends on the years of history and how completely the records show each job. What your data is worth explains the factors.
Bottom line
Know what is in Amazon DynamoDB, then see what it is worth
You don't need to export anything to apply. Apply in two minutes, describe your records on the intro call, and micro1, the data vendor we typically refer companies to, prices the whole dataset and sends an offer 1–2 days later.
Not legal, tax or financial advice. Figures come from micro1's published pages and public reporting as of November 2026.