The old no-rivals rule is buckling under AI
For years, venture capital’s had a pretty simple social contract: if one firm backs a startup, it usually avoids funding a direct rival inside the same portfolio. The rule was never written into stone tablets. But it was treated like one of those quiet industry customs that everyone understands and most people pretend not to discuss. If a VC wants to keep founders talking to them, it can’t look as if it’s cheerfully placing bets on both sides of the same knife fight.
That convention has been under strain for a while, and AI has made the strain obvious. At TechCrunch Disrupt, Benchmark’s partners spoke publicly about the problem, which is telling on its own. Benchmark has a reputation for being unusually founder-friendly and loyal once it commits. So when a firm like that talks openly about pressure around venture capital rivals, it usually means the issue’s moved from awkward edge case to daily annoyance.
In venture capital, the old rule was simple: don’t back two companies that may end up selling the same thing. AI has made that rule much harder to keep without some awkward exceptions.
The reason is pretty practical. AI startups move fast, and their products can drift across category lines almost by accident. A team that starts with a narrow tool may find demand pulling it into adjacent workflows. Another company may launch with a broader pitch and then chip away at a neighbor’s market before anyone in the boardroom has time to react. What looked like a clean separation in January can feel fuzzy by summer.
That pace leaves investors with fewer neat choices. They can try to stay strict and risk missing good companies, or they can tolerate overlap and risk making founders nervous. Neither option is lovely, which is probably why the old taboo is being discussed so publicly now. The pressure isn’t really about ideology. It’s about how quickly AI companies can change shape once they get real users, real revenue and real pressure to expand.
Benchmark’s comments also suggest that the industry’s unwritten pact still exists, just less comfortably than before. Firms still prefer to avoid direct competition where they can. Founders still expect loyalty. Nobody has suddenly decided that backing rivals is a fine idea and can’t possibly cause trouble. But the assumption that the line will always hold’s looking shaky, and that matters because expectations in venture tend to travel fast through founder circles.
The next question, of course, is what happens when those lines are crossed for real. That’s where things get messier, and where the neat theory starts rubbing up against actual people, board seats and recruiting decisions.
The Factory-Cognition dispute shows how messy overlap gets
If the first part of this story was about a rule that now feels stretched thin, Factory’s fight with Cognition shows what that strain looks like in practice. The abstract version is easy enough to discuss at a conference, even one like TechCrunch Disrupt. Advisers, and recruiters all start bumping into one another, given the real version gets uglier once the same investors.
That overlap is already baked into frontier AI. OpenAI and Anthropic share a long list of investors, which tells you this isn’t some rare, one-off exception. Money has flowed into a small cluster of large, ambitious AI companies and that means backers can end up with seats near more than one camp at once. When the companies start chasing the same customers or the same people, the arrangement gets awkward fast.
In AI, the problem isn’t just that rivals compete. It’s that the same adults can be in the room for both sides of the fight.
In late September, Factory said it removed Chris Degnan from his roles as board observer and adviser after what it described as unethical conduct tied to Cognition, a direct rival. That kind of move doesn’t happen casually. A board observer can sit close enough to hear planned plans, product direction, hiring plans and the sort of half-finished ideas that never make it into a press release. An adviser can occupy a similar gray zone. Put the two together, then add a rival company in the same market, and the trust issue stops being theoretical.
Factory’s chief executive went further and said Degnan had been talking about a job at Cognition while he still had access to board-level conversations. That is the part that makes founders bristle. It’s one thing to lose a team member to a competitor after they’ve stepped away cleanly. It’s another to wonder whether someone who could hear sensitive discussions was also exploring an exit to the other side. Even if no secret was passed along, the optics alone can sour relationships.
Degnan pushed back. He said he had already stepped down and denied sharing confidential information with Cognition. That denial matters, of course, because these disputes often turn on what was actually said, when it was said and who knew what. But the timing still raised eyebrows. He later took the job of chief revenue officer at Cognition, which made the whole episode look even more tangled from the outside.
Vinod Khosla then stepped into the mess on X, where he defended Degnan and criticized Factory. That’d be a fairly normal bit of investor commentary if Khosla Ventures didn’t back both companies. Because it does, the public defense carried a sharper edge. One investor, two portfolio companies, one public dust-up. For founders watching from the sidelines, that’s the kind of moment that makes AI startup conflicts feel less like an edge case and more like a recurring nuisance.
What makes this episode hard to wave away is the collision of roles. In software, investors often want the luxury of being close enough to help but distant enough to avoid real responsibility. In AI, especially when companies are growing fast and hiring from the same small pool, that balance gets tested constantly. Board access can blur into recruiting conversations. Recruiting can blur into competitive intelligence. A defense on social media can land like a rebuke to one of your own portfolio companies. Nobody gets to stay neatly in one lane for long.
That’s why the Factory-Cognition dispute matters as a case study, even if the facts keep being debated by the people involved. It shows how overlap becomes personal once a founder, a board insider and a rival employer all meet in the same week. The rivalry isn’t just about products anymore. It’s about who sat where, who heard what, and who decided to leave for the other side.
Why AI startups keep running into each other
The Factory-Cognition dispute was loud, but the pattern behind it’s quieter and harder to avoid. Eric Vishria’s point was basically this: companies don’t stay where they started. A startup can look cleanly separated from another one at the time of the first check, then spend the next year creeping into adjacent territory because the market moves fast and the product surface area keeps growing. What seemed like two neat boxes in a pitch deck can turn into a messy Venn diagram by the time a team has shipped three new features and hired half a dozen people.
That’s part of why these conflicts feel less like a moral failure than a byproduct of the current market. Nobody’s sitting around hoping for a portfolio fight. Investors would rather avoid it entirely. But when AI companies expand quickly, and when the category itself shifts under everyone’s feet, overlap shows up whether anyone likes it or not.
In AI, the competitor you declined at launch can become the neighbor you backed six months later.
But Benchmark also pointed out something a lot of people forget when they imagine venture firms as all-seeing referees. A VC shop isn’t a product lab or a research group. It does not sit inside every engineering decision, and it usually doesn’t have perfect visibility into how a model gets tuned, which customer requests are landing, or which adjacent problem a founder will chase next quarter (and that’s no small thing). That matters. Even if a firm wanted to draw a bright line between one company and another, it often wouldn’t have the inside technical knowledge to do it cleanly. It can read the memo, and it can’t read the future.
Chetan Puttagunta pushed the point even further. In his view, the cost of building software is drifting close to zero, or at least close enough to change founder behavior. When shipping code becomes cheaper and faster, teams get bolder about the scale of the business they want to build. They don’t just ask, “Can we make this feature work?” They ask, “Why stop there?” That shift sounds simple, but it changes the way startups move. If it takes less effort to launch another module, another workflow, or another product line, then founders have a stronger incentive to stretch into more territory and claim more of the market while they can.
That ambition creates a very specific kind of conflict. Puttagunta drew a line between an obvious competitor at the start and a portfolio company that later wanders into someone else’s lane. The first case is easy to spot. Two founders are plainly chasing the same customer with nearly the same pitch. A VC can see that on day one and make a choice. The second case is uglier. A company may begin in one corner of AI, raise money with no real overlap, then grow into a neighboring category because the economics make that move tempting. By the time the overlap is obvious, the investor is stuck explaining why yesterday’s harmless bet now looks like a rival.
That’s why the OpenAI and Anthropic investors conversation keeps coming up in these debates. Shared backers are everywhere in frontier AI, and that alone doesn’t create a conflict. The conflict appears when the companies start brushing up against each other in product, hiring, or distribution. The same thing showed up in the Factory Cognition dispute. A board-level relationship that looked manageable on paper suddenly looked fraught once one company believed an adviser had crossed a line toward a direct competitor.
The uncomfortable part is that this kind of collision may become more common, not less. AI lets small teams move faster, launch broader products, and enter neighboring markets with less friction than older software categories allowed. That makes the old neat separation between “this investor backs this lane” and “that investor backs that lane” harder to preserve. The lines keep shifting. Sometimes they shift before anyone has time to blink.
And that’s the real reason the rivalry problem feels larger now. It isn’t just about a single board seat or one awkward recruiting story. It’s about a market where company boundaries are soft, product scopes widen quickly and investors keep discovering that today’s unrelated bets can become tomorrow’s awkward cousins. The next question, naturally, is what a firm does once it knows that collisions are no longer rare.
How Benchmark tries to manage conflicts without breaking trust
Once a firm starts backing companies that may end up eyeing the same customer, the question stops being theoretical. It becomes a matter of who knew what, when they knew it and whether they said anything before things got awkward.
Peter Fenton called that problem Benchmark’s hardest one and, with a bit of dramatic flair that still felt more candid than theatrical, compared it to kryptonite. That’s a useful word here. Not because it sounds cool in a boardroom, but because it suggests a weakness that can’t be waved away with a polished memo. The firm can’t pretend rivalry never happens. It can only decide how it’ll behave when it does.
Fenton also drew a line between two styles of investing that are often talked about as if they’re the same thing. Not ideal. At one end sits the co-founder-like investor, the person who works closely with a founder, takes part in the messy middle, and knows the company’s plans well enough to spot trouble early. At the other end is the public-market stock picker, who buys in, watches from a distance and mostly cares whether the numbers work out. Benchmark, at least in its own telling, puts itself much closer to the founder side of that spectrum.
Trust gets tested when the awkward part arrives, not when everyone is still smiling through the first pitch meeting.
That framing matters because it shapes how the firm says it handles overlap. Transparency is not treated as a nice-to-have. It’s the rule. If a conflict emerges, the founder shouldn’t hear about it after the fact, once decisions have already been made and the story has hardened into “we mentioned it, sort of.” Fenton’s point was plain enough: nobody should be surprised later by a relationship, a side conversation, or a competing investment that was kept vague at the time.
That’s easier to say than to do, of course. Venture firms live in a world where the same people move between startups, boardrooms and new jobs with surprising speed. A case like Chris Degnan’s departure from Factory and his move to Cognition’s exactly the sort of thing that can turn a loose acquaintance into a real mess. Vinod Khosla’s public defense of Degnan, even with Khosla Ventures backing both companies, shows how quickly these disputes can spill beyond the companies themselves and into the investor circle around them.
Benchmark’s answer isn’t that it can prevent every collision. It seems more interested in proving that it’ll not act slippery when one shows up. If a founder and a portfolio company overlap, the firm wants the founder to understand the situation before it becomes gossip or a surprise boardroom scene. That may sound basic, but plenty of conflict problems become worse because someone waits too long to explain the obvious.
Still, Benchmark does draw a firmer line at the earliest stage. When it’s first building a relationship with a founder, exclusivity matters more than it does later on. That makes sense. Early on, the investor and founder are deciding whether they can actually work together, and trust at that point’s fragile. The partnership starts on a bad foot, if the first few conversations already feel crowded or divided. Nobody likes finding out their “special relationship” came with a bunch of invisible asterisks.
So the firm’s position is not that conflicts are harmless. It’s that they have to be handled openly, quickly and without pretending the founder should simply shrug and carry on. The pitch is simple enough: if overlap is unavoidable, don’t play games with it. Say what’s happening. Say it early. Then let the founder decide whether the arrangement still makes sense.
What founders should read into the new reality
For founders, the useful takeaway is fairly plain: Benchmark is not declaring the anti-rival rule dead, but it is admitting that AI keeps pushing firms into exceptions they can’t easily dodge. That matters because startup competition in AI rarely stays parked in one neat category. A company may begin in one lane, then expand into adjacent territory a few months later, and suddenly the old comfort blanket of “these two bets don’t touch” feels a lot thinner.
If your company grows into a neighbor’s lane, yesterday’s clean separation can disappear faster than anyone wants to admit.
That’s the part founders should internalize. The main risk isn’t that every investor will start backing obvious rivals on purpose. It’s that portfolio conflicts will show up more often because AI startups move fast, copy borders faster than lawyers can label them and often decide they’d rather own a wider slice of the market than stay politely boxed in. A company that looked harmless at seed stage may look a lot less harmless after it adds a new product line, a new buyer, or a new layer of the stack.
So the practical lesson is less glamorous than a lot of venture mythology. Ask early how a firm handles overlap. Ask what happens if another portfolio company starts circling the same customer. Ask how board access is managed, who gets looped into sensitive conversations, and whether the firm will say out loud when those lines start to cross. Those questions can feel a little awkward at first. They’re cheaper than finding out later that everyone assumed a different set of rules.
The other open question’s whether more venture firms will speak this plainly. Benchmark’s chosen to discuss the problem instead of pretending it won’t happen, and founders may want to notice that. Transparency won’t erase tension, but it can prevent the sour aftertaste that comes when expectations were fuzzy from day one. In practice, that may matter more than any old promise that investors will keep their portfolios neatly separated and everyone will live happily in their own little silo.
That’s the irony at the end of all this. A firm with a reputation for loyalty’s now telling founders, in effect, that rivalry conflicts are coming anyway. Not because loyalty’s vanished, but because AI startup competition keeps forcing the issue into view.
