The US Justice Department is investigating Andreessen Horowitz because its partners sit on the boards of competing data companies Databricks and Fivetran.
In this article
A 1914 law bans interlocking directorates to stop the exchange of sensitive information between rivals. This rule means one person cannot lead two directly competing firms. The probe targets the firm itself rather than just the individuals involved.
The board overlap
Andreessen Horowitz manages about $90 billion in assets. It holds stakes in OpenAI, SpaceX, and ElevenLabs. Databricks was most recently valued at $190 billion.
The case focuses on two data firms backed by the venture capital group. Cofounder Ben Horowitz sits on the Databricks board. Partner Martin Casado sits on the Fivetran board. Both companies help businesses collect, organise, and analyse large amounts of data.
Casado also sat on the board of dbt Labs. Fivetran acquired that company in June. The DOJ reviewed that merger for months before clearing it without conditions. The nearly year-old probe began around the same time and is still ongoing.
Why the firm is under fire
These cases usually end with directors stepping down from one of the boards. Under the Biden administration, the DOJ forced executives to resign in about a dozen cases. What is new here is that the investigation targets the firm because several of its directors sit on competing boards.
The firm’s ties to the second Trump administration add another layer. Andreessen Horowitz has aligned itself closely with the White House. According to Bloomberg, the group successfully pushed to roll back many AI safety rules that benefit its own portfolio.
Horowitz and cofounder Marc Andreessen donated millions to a pro-Trump group in 2024.
What it means
For people making AI tools, this probe signals that regulators are watching how capital flows between competing projects. If Andreessen Horowitz faces penalties or forced divestments, other investors might rethink placing board seats in rival companies. It does not change the technology itself, but it could slow the speed at which new models reach the market.




