DOJ probes a16z over seats on rival AI boards

Share
DOJ probes a16z over seats on rival AI boards

The Justice Department has spent nearly a year looking at the most ordinary thing a venture firm does: putting partners on boards.

Andreessen Horowitz is the focus of a Justice Department antitrust probe over whether its investment partners are improperly serving on the boards of competing AI companies, according to people familiar with the matter. The investigation, which had not been previously reported, is nearly a year old. Among the companies at issue are Databricks — one of the world's most valuable privately held technology firms, where co-founder Ben Horowitz sits on the board — and Fivetran, where partner Martin Casado is a director. Both companies sell the layer that makes enterprise AI usable: they help businesses collect, organize, and analyze huge amounts of data.

The overlap is not theoretical. Casado was also on the board of dbt Labs, a similar data company that Fivetran acquired in June after a deal first announced last October. The Justice Department spent months reviewing that merger and ultimately cleared it with no conditions. The board-seat investigation opened around the same time as that review and kept going after the deal closed. The people who described the matter asked not to be named because it is confidential. Resolving probes like this typically means a director steps down from one of the competing boards.

That sequence is the story. Clearing a merger answers whether two companies may combine. It does not answer whether the same partnership should keep a window into companies that still compete — or that just swallowed a competitor. Databricks and Fivetran are not chatbots, but they are the plumbing chatbots run on: pipelines, warehouses, transformation, the unglamorous work of turning a company's mess into something a model can use. If those businesses are close enough that overlapping directors raise an antitrust question, the AI data stack has crossed a line the government is willing to police.

Why it matters: a16z is not a silent check-writer in this cycle. Its partners sit on the boards of the companies that own the data pipelines, the model shops, and the application layer. That is how venture firms claim to add value, and it is also how they create the conflict antitrust law has always worried about — a small group of people seeing competitive plans at two firms that sell into the same customer. A year-old probe that survived an unconditional merger clearance is not a rumor that faded. It is a signal that the department will keep pulling on board composition even after it lets a deal through.

The take is simple. The AI boom packed the same partners onto the same category of boards because there were only so many people who understood the category. That looked like expertise. It now looks like an exhibit. If Databricks and Fivetran are in scope, every multi-board AI partner at every large firm should assume the next letter is not hypothetical.

What to watch: whether Horowitz or Casado leaves a board — and whether the probe names companies beyond the two already identified.

Should the same VC partners sit on the boards of competing AI companies? Tell us in the comments.

Read more

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kuaishou's Kling AI has spent two years becoming the video model people actually finish projects with. Its new flagship, Kling 4.0, went fully live on October 15 — and this round is less about prettier frames than about the two things that make a generated shot usable in a real edit: length and control. Kling 4.0 generates up to 30 seconds in a single native pass, outputs 10-bit HDR at 4K, 1080p and 21:9, and takes up to 15 reference assets in one task. That budget splits across ten images, fi

Brain organoids classified Braille at 83% — in a trio, not alone

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Microsoft finished pre-training its biggest in-house model yet — then used it to make a point about the company it bankrolls. Microsoft AI completed pre-training of MAI-Base-2, a mixture-of-experts foundation model with 80 billion active and 2.8 trillion total parameters, and says the recipe is 4.2–12.3x more pre-training FLOPs-efficient than DeepSeek-V4-Pro-Base. The run itself is the number worth keeping: 16,000 Nvidia GB200 GPUs for 38 days across August and September, at 21% average model

Amazon's new Bee devices listen all day — for $11.99 a month

Amazon's new Bee devices listen all day — for $11.99 a month

Amazon turned its always-listening Bee line into a two-device family on Thursday, and NVIDIA opened a 2.8-trillion-token Indic corpus recipe. The connecting thread: both are bets that the next AI money is in memory — yours, or your model's. Amazon put two new always-listening devices on sale, and both now lead to a subscription. Bee Frontier, the next-generation wristband that also clips to clothing, ships now at $74.99 in Onyx, Lavender and Sage — up from the $50 predecessor — with seven-day