The seat is the conduct

Ben Horowitz sits on the board of Databricks, valued at $190 billion. Martin Casado sits on the board of Fivetran, which combined with dbt Labs in June. Same firm, two partners, two companies that now sell into the same market. The Justice Department has been examining that arrangement for nearly a year, and the venture industry's response, reported this week, is that it doesn't understand the question.

The confusion is the finding. Section 8 of the Clayton Act has no conduct element. It bars an individual or entity from serving on the boards of competing companies — no bad act to prove, no intent to establish, no information anyone has to show was misused. Several VCs told TechCrunch they were surprised by the probe. They're looking for the wrongdoing. There isn't any. That is what the statute was built to make unnecessary.

Two defenses are on offer and both concede the point. The first is that Databricks and Fivetran weren't rivals when a16z invested — true, and irrelevant. Databricks was a cloud storage company until Lakeflow walked it into AI data pipelines and connectors, which is Fivetran's business. Section 8 attaches to the configuration that exists now, not the one that existed at the term sheet. The second defense is an information wall between Horowitz and Casado, so neither partner sees the other's confidential material. That is a remedy imported from securities compliance, where the harm is the information flow. Here the seat is the harm. And the law reaches the entity, not only the person — a wall between two partners of the same firm answers a question nobody asked.

The industry knows what's actually at stake, and TechCrunch printed it plainly:

That is the honest objection, and it is a real cost. A board seat from a top-decile firm is sold to founders as a commitment — the partner who stays through the down round, who takes the 11 p.m. call. Repricing it as revocable makes it worth less, and founders pay for that, not the DOJ.

The DOJ isn't creating the liability. It's disclosing it.

But the commitment was already conditional; nobody had written the condition down. A firm with hundreds of portfolio companies is running a permanent overlap surface, and the defense being offered in public is that overlap is inevitable at that scale. Take that seriously. If pivots into adjacent markets are structurally certain across a portfolio that size, then the Databricks–Fivetran pair isn't an unlucky draw. It's the expected outcome, arriving on schedule. The founder who took Casado's seat this year has always been accepting that some product decision at a company he's never heard of could pull his director away. He just wasn't quoted a price for it.

Venture has spent a decade arguing that scale is what makes a firm useful to founders — the network, the platform team, the partner who has seen this movie forty times. Section 8 says scale is what makes the seat unlawful. Both can be true. Only one of them is 112 years old and on the books.