Synthetic Data Companies
Synthetic data companies sell software that learns the shape of real data and generates new records, images or scenes that keep it. This is a reference to who they are, what they publish and how to test what they make.
Get a quote| Field | Real record | Synthetic record |
|---|---|---|
| Customer age | 47 | 52 |
| ZIP prefix | 941 | 940 |
| Plan | Business, annual | Business, annual |
| Monthly spend | $1,284.50 | $1,197.20 |
| Support tickets, last 90 days | 3 | 4 |
Which synthetic data companies are worth a look?
The market splits in two. Tabular vendors generate rows for databases, analytics and test environments. Image and simulation vendors render labeled scenes for computer vision, robotics and driving.
| Vendor | What it makes | Published price | Ownership change |
|---|---|---|---|
| Tonic.ai | Three products: Structural (databases and other structured data), Textual (free text and files) and Fabricate (new data from a schema or prompt) | Not published | Bought Fabricate (built by Mockaroo) on 22 Apr 2025 |
| Syntho (and MOSTLY AI, powered by Syntho) | Synthetic tabular data platform based in Amsterdam; now also runs the MOSTLY AI brand | Not published | Syntho bought the MOSTLY AI trademark and related assets on 9 Jun 2026; the brand continues as "MOSTLY AI, powered by Syntho" |
| Aindo | Synthetic data platform from Trieste, Italy, aimed at regulated sectors | Not published | None found |
| K2view | Data management vendor whose synthetic data tool mixes AI generation, business rules, cloning and masking | No prices on the page; offers a "Start Free" option | None found |
| SAS Data Maker | SAS's synthetic data product, which took in Hazy's software | Not published | SAS bought Hazy's principal software assets on 12 Nov 2024 and folded them into SAS Data Maker |
Some vendors in this table may pay an introduction fee when a team we put in touch signs with them. Whether a vendor pays has no effect on the list, its order or our notes.
| Vendor | What it makes | Data types | Published price |
|---|---|---|---|
| Rendered.ai | Physically accurate, fully labeled synthetic imagery for computer vision, including SAR, infrared, multispectral and X-ray sensors | Images, 3D | Platform as a service from $5,000 a month (Teams) and $15,000 a month (Organizations); other services priced on request |
| Parallel Domain | Synthetic data and simulation for training and testing perception systems in autonomous vehicles and robotics (PD Platform, PD Forma, PD Sim) | Images, video, 3D | Not published |
Some vendors in this table may pay an introduction fee when a team we put in touch signs with them. Whether a vendor pays has no effect on the list, its order or our notes.
Three names you may remember are no longer independent. TechCrunch reported in March 2025 that NVIDIA bought Gretel; Hazy's software went into SAS Data Maker; KPMG agreed to take YData's technology assets. The full synthetic data vendor list has the dates and a running log of changes.
Where these facts come from
- Tonic.ai press release on Fabricate, read 7 Nov 2026
- Syntho announcement on the MOSTLY AI brand, read 7 Nov 2026
- Aindo security and compliance page, read 7 Nov 2026
- K2view synthetic data generation tools page, read 7 Nov 2026
- SAS press release on Hazy, read 7 Nov 2026
- Rendered.ai home page, read 7 Nov 2026
- Parallel Domain "What is Parallel Domain" page, read 7 Nov 2026
- TechCrunch on NVIDIA and Gretel, 19 Mar 2025
- Report of the KPMG agreement for YData assets, 20 Oct 2025
What does synthetic data cost?
Published prices only. A vendor's quote for your project can differ.
Very few synthetic data companies print a price. Of the seven we track, one publishes figures: Rendered.ai lists its platform as a service from $5,000 a month for Teams and $15,000 a month for Organizations, with other services priced on request (Rendered.ai home page, read 7 Nov 2026).
K2view shows no prices on its synthetic data page but offers a free start. Tonic.ai, Syntho, Aindo, SAS Data Maker and Parallel Domain did not publish prices on the pages we read, so expect a sales call and a quote shaped by data volume, seats and where the software runs.
Before you pay anything, decide whether synthetic data is the right tool at all. Our short decision aid on whether to use synthetic data walks through the cases where it is not.
What privacy guarantee does synthetic data actually give?
Two very different promises hide under the same word. De-identification removes or masks direct identifiers, then generates or transforms records. It lowers risk but proves nothing. Differential privacy limits, by a stated amount called epsilon, how much any one person's record can change what the generator learns. Smaller epsilon means stronger protection and usually less faithful data.
None of the vendor pages we read on 7 Nov 2026 stated an epsilon value. Aindo lists a Europrivacy certification, a GDPR certification scheme, on its security page. Ask every vendor for its privacy claim in writing.
Four questions for the vendor's privacy team
- Is differential privacy applied during training, and with which epsilon and delta?
- Which privacy tests run on every output (nearest-record distance, membership inference)?
- Does any of our data leave our own cloud account or servers?
- Who owns the generated data and the trained generator after the contract ends?
When should you not use synthetic data?
A generator can only reproduce patterns it has seen. That rules it out more often than vendor pages suggest.
- Rare events are the point. Fraud, defects and outliers are exactly what generators smooth away or copy too closely.
- The source table is small. With a few thousand rows the generator has little to learn from and a higher chance of memorizing.
- You need an audit trail to real people. Regulated reporting needs the real records, not lookalikes.
- Masking would do. For a test database, masked or subsetted production data is often cheaper and good enough.
- The real world has moved. Synthetic data inherits the period and the biases of its source.
The decision aid turns these into five questions with a plain verdict.
How do you test synthetic data quality?
Three tests, run on every dataset a vendor delivers. A vendor's own report is a start; repeat at least the utility test yourself.
| Test | The question it answers | Typical checks |
|---|---|---|
| Fidelity | Does the synthetic data look like the real data, column by column and in combination? | Compare distributions per column, correlations between columns, category frequencies, and joint patterns between related tables |
| Utility | Does a model or analysis built on it give the same answers? | Train on synthetic, test on real held-out data (TSTR) and compare with train on real, test on real; rerun the key queries |
| Privacy | Can anyone learn about a real person from it? | Distance from each synthetic row to its closest real row, exact-copy counts, membership inference attacks, and the stated epsilon |
For a worked plan with pass marks to agree before a trial, read how to test synthetic data quality.
How does a synthetic data vendor trial usually run?
- Write the brief. Data type, row or image counts, where the software must run, the privacy promise you need and the model or test the data will feed.
- Shortlist two or three vendors that support your data type and deployment. Tabular and image vendors rarely overlap.
- Trial on a slice you would be comfortable sharing, inside your own environment where the vendor allows it.
- Score fidelity, utility and privacy against pass marks agreed before the trial started.
- Read the contract for data ownership, retention of your source data and the trained generator, and how pricing scales.
Image and simulation projects add one step: measure the gap between synthetic and real scenes on a real validation set. The page on synthetic images and simulation covers it, and tabular synthetic data covers databases.
Short answers about synthetic data companies
What do synthetic data companies sell?
Mostly software. Tabular vendors such as Tonic.ai, Syntho and SAS Data Maker sell platforms that learn from your tables and generate new rows. Image and simulation vendors such as Rendered.ai and Parallel Domain generate labeled scenes for computer vision. A few also sell services around the software.
Is synthetic data anonymous?
Not automatically. A generator can memorize and repeat rare records. Only a formal method such as differential privacy, with a stated privacy budget, gives a guarantee you can put a number on. Ask each vendor what it promises in writing. See privacy guarantees.
How much does a synthetic data platform cost?
Most vendors do not publish prices. Of the vendors we track, Rendered.ai publishes platform plans from $5,000 a month (Teams) and $15,000 a month (Organizations), read 7 Nov 2026. K2view shows no prices but offers a free start.
Who bought Gretel, Hazy and MOSTLY AI?
TechCrunch reported in March 2025 that NVIDIA acquired Gretel. SAS bought Hazy's principal software assets in November 2024 for SAS Data Maker. Syntho bought the MOSTLY AI trademark and related assets in June 2026.
Can synthetic data replace real data for training a model?
Sometimes for part of it. It works best for filling rare classes, testing pipelines and sharing data that cannot leave a team. Always measure a model trained on synthetic data against real held-out data before you rely on it.
What should I ask a vendor before a trial?
Which data types it supports, where the software runs, whether it offers differential privacy and with what parameters, which quality reports come with each dataset, and how pricing scales with rows, seats or compute.
Get a quote
Three parts: the company location, a contact for replies, then what the generated data has to do. The final part is optional.
Disclosure: this is an introduction service, and any vendor it introduces you to can pay the site for that introduction. Each brief is read by a person and matched against vendors' published terms; you get the names first, and nothing goes to any vendor without your go-ahead.
Please do not send data, samples, schemas with real values or credentials through this form. A written description is all we need.
Your brief is in.
- We read it and compare it with the published terms of the vendors we track.
- We email you the names of the ones that fit, and why.
- A vendor hears about your project only if you tell us to go ahead.