Benchmark Admits Rival-Backing Pact Is Dead, Blames AI

Benchmark Capital, the firm that has built its entire reputation on surgical decision-making and long-term discipline, has now admitted something Silicon Valley has known for two years: the unwritten rule against backing direct competitors no longer exists. The confession arrived not with dramatic fanfare but as a quiet acknowledgment that even the most stalwart keepers of VC propriety cannot resist the gravitational pull of artificial intelligence money madness. When Benchmark—a shop known for saying no—starts explaining why it can no longer afford to say no to conflicts, you are witnessing the death rattle of an entire ethical infrastructure.

The pact in question was never a formal agreement, which is precisely why it was so delicious as a mechanism of social control. For decades, the unspoken rule held that serious venture firms would not simultaneously back two companies in direct competition, on the theory that doing so created irreconcilable conflicts of interest. A partner could not advocate equally for both, could not share insights that benefited one without harming the other, could not claim to have a real conviction about either founder's vision. It was gentlemen's agreement disguised as prudence. Now, with every VC from Sand Hill Road to Toronto convinced that the next $100 billion company will run on large language models, that gentlemen's agreement looks like a luxury good nobody can afford.

Benchmark's pivot is not surprising—it is inevitable. When a thesis becomes hot enough, when the opportunity cost of sitting out becomes visible enough, principles don't erode gradually; they evaporate. The firm has watched rival shops like Sequoia and a.16z pursue aggressive AI portfolios that deliberately court overlap and direct competition. To maintain discipline while competitors abandon it is to accept being left behind. Benchmark has chosen, rationally, to abandon a position that only works if everyone maintains it. The moment one prestigious firm defects, the equilibrium collapses.

What's remarkable is the candor of the admission itself. Rather than quietly start backing AI rivals while maintaining the pretense of restraint, Benchmark has simply announced that the old rules no longer apply. This is not a rationalization—it is a surrender. The firm is not arguing that conflicts of interest have somehow become less problematic; it is arguing that fear of missing out is a more powerful force than integrity. At least they are honest about the hierarchy.

The practical implications are substantial and mostly bad. Without the rival-backing constraint, VC firms now have even fewer incentives to act as genuine stewards of capital allocation. A partner's duty to the firm's portfolio—to the LPs who funded it—will now routinely pit them against their duty to individual founders. The conflicts will not disappear; they will simply be priced in as a cost of doing business. Founder meetings will become Kabuki theater, with partners carefully managing what information can be shared with whom. Trust, already thin in venture, gets thinner still.

What we are witnessing is the predictable collapse of a system that relied on voluntary compliance and social shame. Both have proven powerless against FOMO and the belief that missing one AI wave means irrelevance. Benchmark's confession is not the cause of this breakdown; it is merely evidence that it was already complete. The venture pact against rivals did not die because AI changed the economics. It died because, for the first time in forty years, enough money moved fast enough that everyone realized simultaneously: principles are negotiable when the stakes are this high and the herd is moving this fast.