A Board Seat Signed Off as Routine
Years ago, when Ben Horowitz agreed to sit on the board of Databricks and Martin Casado agreed to sit on the board of Fivetran, both moves were logged at Andreessen Horowitz as routine governance, the kind of paperwork a VC firm signs off on without a second board meeting. Databricks was a data and AI infrastructure company. Fivetran moved data between systems. Neither looked like the other's competitor, and nobody at a16z had reason to think two of its partners were building toward a conflict.
On August 18, 2026, TechCrunch reported that the US Department of Justice has opened an antitrust probe into a16z over exactly that arrangement, examining whether it violates Section 8 of the Clayton Antitrust Act. The law bars a person from serving on the boards of two competing companies at the same time. The question the DOJ is now asking is whether Horowitz and Casado, without ever changing seats, ended up on opposite sides of a rivalry that grew up around them.
The trigger was product movement, not a boardroom decision. Databricks' pipeline product, Lakeflow, expanded over time into the same core data-pipeline business that Fivetran occupies, a business that got bigger still in June 2026 when Fivetran merged with dbt Labs. Multiple venture capitalists quoted by TechCrunch made the same point: these companies were not rivals when a16z invested.
The Roadmap Nobody Wrote Into the Contract
Lakeflow was not built to pick a fight with Fivetran and dbt Labs. It grew the way most platform products grow, adding adjacent features until an adjacent business found itself standing in a bigger company's shadow. That is normal competitive behavior, not a scheme, and nothing in the reporting suggests Databricks expanded Lakeflow in order to trap Ben Horowitz or Martin Casado in a legal problem.
But normal is exactly the trap for a founder. When a company takes venture money with a board seat attached, it is making an implicit bet that the investor's other portfolio companies will stay in their own lanes for the life of the relationship. Product roadmaps do not honor that bet. A company that was not a competitor on day one can become one years later while every person involved acts in good faith the whole way there.
Why Section 8 Suddenly Matters to Ordinary Deals
Section 8 of the Clayton Antitrust Act is not new and it is not aimed at venture capital specifically. It restricts a person from sitting simultaneously on the boards of two companies that compete with each other, on the theory that a shared board member can soften competition between rivals whether or not anyone intends it. For decades it has mattered most to large public companies with overlapping directors.
What makes the a16z probe notable is that it applies the same rule to a venture firm's ordinary practice of taking board seats across a large portfolio. A16z did not design Horowitz's Databricks seat and Casado's Fivetran seat as a single arrangement, and the two companies were not competitors when either seat was created. The DOJ's interest signals that regulators will look at where a portfolio has drifted, not just where it started.
The Decision Lesson: Build the Trigger In Before You Need It
The base-rate mistake: at the moment a board seat is signed off, "these companies are not competitors" reads as a fact. It is really a snapshot, true on the day it is written down and silent about every product decision either company will make afterward. Treating a snapshot as a permanent fact is the same species of error that shows up whenever a founder assumes today's competitive map will hold for the life of a financing round.
The fix is not vigilance. Nobody at a16z, Databricks, or Fivetran was likely to notice in real time that Lakeflow had crossed into Fivetran's territory, because nobody was assigned to watch for it. The fix is a mechanism that does not depend on anyone noticing: a contractual clause, agreed at signing, that requires disclosure or recusal if a co-investor's other portfolio company later develops a product that materially overlaps with yours.
What to Do If You Are Raising Right Now
If you are an EU or UK owner about to take institutional capital with a board seat attached, this is not a US-only story. The same investor behavior, the same portfolio drift, and the same governance gap exist wherever a fund holds board seats across companies that might one day compete. A DOJ probe is one way this problem surfaces; a blocked follow-on round or a soured co-investor relationship is another, quieter way it surfaces closer to home.
Put the review trigger in the term sheet, not in a side conversation you plan to have "if it ever comes up." Ask your lead investor to list every other board seat they hold in your category today, and agree now on what happens if one of those companies expands into yours. A clause is cheap to negotiate before a conflict exists and expensive to invent after one does.
Read next: A 190 Billion Dollar Vendor Europe Cannot Audit | Accel Raises $800M for Its Ninth Europe Fund
