US Justice Department reportedly probes Nvidia's Hugging Face deal

The US Justice Department is reportedly asking whether Nvidia's planned $12.93 billion purchase of Hugging Face could harm competition in artificial intelligence. Reuters reported the inquiry on 28 October 2026, citing people familiar with the matter. Officials are reportedly examining whether the deal would give Nvidia control of a key distribution point for AI models. Neither the Justice Department nor Nvidia has commented on the report. For a business that runs automation on open AI models, nothing changes today. The deal has not closed, and Nvidia has said it will keep the Hugging Face platform open. The practical step is to know which models your workflows depend on and to keep the ability to switch them.
In short
- The US Justice Department is reportedly examining whether Nvidia's $12.93 billion Hugging Face deal could harm AI competition.
- Neither the Justice Department nor Nvidia has commented on the report, and the deal has not closed.
- Nvidia says it will keep Hugging Face open and expects closing in the first half of 2027.
- Businesses should record which open models their automated work depends on and keep a tested copy.
- The regulator's outcome is not known; watch for official statements and the closing announcement.
What happened
According to Reuters, as cited by CNBC, the US Department of Justice has asked Nvidia how its acquisition of Hugging Face could affect competitors. CNBC said it confirmed the inquiry with one person familiar with the matter. Reuters, citing unnamed sources, reported on 28 October 2026 that officials are asking whether the deal would let Nvidia control a key distribution point for AI models. Reuters also reported that the review covers Nvidia's other acquisitions, investments and sales practices, including whether the company has put pressure on customers. Neither the Justice Department nor Nvidia commented, according to PYMNTS. A Justice Department spokesperson has said government investigations are not evidence of wrongdoing.
The inquiry runs alongside a probe into Nvidia's conduct that began under the Biden administration and has continued under the Trump administration. The Hugging Face questions are part of that wider review, not a separate case that has been filed in court.
The deal in brief
Nvidia announced the purchase on 3 September 2026. TechCrunch reported that Jensen Huang said Nvidia compute would not be required to build on or deploy through Hugging Face. Nvidia has said it expects the deal to close in the first half of 2027, according to TNW. Hugging Face's own site describes more than 2 million models and about 500,000 datasets on its platform, which is the layer a business's automation tools often reach into.
Why it matters for businesses that automate work
Many automation projects use open models downloaded from a shared hub. A model may read invoices, sort customer emails, or summarise contracts. If one company came to own the hub where those models are published and updated, the terms of access could change over time. The regulator's concern, as reported, is exactly that kind of control.
The review does not mean the platform will change. It means the question is open. Nvidia's public position is that the platform stays open to rival chips, clouds and software. That position is a company statement, and the outcome of the review is not known.
What changes in practice
For most businesses, the first change is a list. Write down which models each automated task uses, where each model was downloaded from, and the version you tested. A five-person firm can do this in a spreadsheet. A large group should hold it in its IT asset register.
Next, keep a second copy of any model you depend on. Open model licences differ, so read the licence attached to each model before you rely on it. Store the approved version inside your own systems so a change on the hub does not change your workflow without a test.
Finally, separate the model from the workflow. If an invoice-matching task calls one model by name in several places, a change means editing many steps. If the model name sits in one setting, a switch is a test and an approval, not a rebuild.
Limits and what to watch next
Automation cannot tell you what a regulator will decide. It can only show you which of your processes depend on a given model, and whether a replacement gives the same output on your own documents. Those results depend on your own data, and they need a person to check them before anything goes live.
| Item | Status on 30 October 2026 |
|---|---|
| Justice Department review | Reported, not confirmed by the department; no filing announced |
| Nvidia's open-platform pledge | Stated by Nvidia; the review may test it |
| Expected closing | First half of 2027, according to Nvidia |
Watch for an official statement from the Justice Department, any request for changes to the deal's terms, and Nvidia's closing announcement. Any change to access terms on the platform would be the signal to test your models again.
How AiStaffo would automate this
AiStaffo would first map each automated task to the model it calls, the source it was downloaded from and the version that was tested. Routine checks would then run automatically, so a model change on a shared hub triggers a test batch of past invoices, emails or reports. The results are compared against the outputs the business already accepts, and a person reviews any difference. The owner still decides which model to use and approves any switch. Book a free automation audit.
Questions people ask
Has the US Justice Department blocked Nvidia's Hugging Face deal?
Will Hugging Face stop being open after Nvidia buys it?
Does this affect my business if we only use AI tools built on Hugging Face models?
When is the Nvidia and Hugging Face deal expected to close?
Should a business stop using open AI models because of this news?
Book a free automation audit
Thirty minutes. We look at one process you run every week and tell you exactly what an AI worker would take off your desk, and what it would not.

































