Benchmark: “you start to run into each other” as free AI building breaks the no-rivals pact

Let me catch you up on a shift that cuts across your product reviews, your hiring pitch, and your competitive deck. AI made building features cheap. That broke the thing that used to protect you. If a rival can copy your best screen in a weekend, what keeps customers from leaving? That question is now live in boardrooms, venture firms, and open-source repos all at once. Here is where we are.

The deep cut

• A moat is a loop, not a feature. Dan Hockenmaier splits the fuel like speed and brand from the four advantages that actually compound.
• AI makes copying cheap, so switching has to be costly. Rippling and Stripe win by getting embedded, not by shipping prettier screens.
• Loyalty bends when the cost of building hits zero. Benchmark admits the no-rivals pact is straining as portfolio companies collide.

What copying for free does to your product

Peter Yang built a replacement for a $200-a-year YouTube tool in five minutes with Claude. A small team rebuilt seven Adobe apps with Opus, including a Photoshop clone that pulled 45,000 GitHub stars. If your product is bloated with features power users wanted and nobody else touches, that is now the soft spot. Someone will rebuild the 20 percent that matters and skip the rest.

The warning for your roadmap is simple. A pile of features is not protection. Retool found 35 percent of enterprises have swapped a SaaS tool for an internal app. Curative dropped a $600,000 Salesforce contract after building its own CRM. Your biggest competitor might be your customer's own engineering team.

The four things that actually hold

So what does hold when copying is free? Dan Hockenmaier argues only four advantages compound: switching costs, network effects, scale economies, and data effects. Speed, brand, distribution, and funding are fuel. They get the car moving. They do not keep rivals out.

His sharpest point is about switching costs in the AI era. AI makes the migration faster, sure. But the real cost is the "workflow tax," all the time your customer's team spent learning the tool and wiring it into how they work. Rippling gets ripped out rarely because it sits across HR, payroll, finance, and more on shared infrastructure. Stripe Radar has seen 70 trillion data points, so 92 percent of transactions use a card it already knows. Copy the screen all you want. You cannot copy that.

When the AI money keeps flowing anyway

The commodity-features story runs into a wall at the top. Anthropic posted $16.8 billion in Q3 revenue, nearly 4x all of 2025, with net revenue retention near 900 percent. Customers who spent a dollar a year ago now spend nine. At the application layer, Ramp hit a $60 billion valuation four months after a $44 billion one, crossing $2 billion in annualized revenue.

Ramp is a live test of the embed strategy. It started in expense management and now runs payments, procurement, accounting, even a corporate card for AI agents. Investor Keith Rabois skipped past the obvious rival: "Ramp is not worried about Rippling. We're going to compete with Anthropic." The game is owning the whole finance workflow, not winning one feature fight.

The pact your investors stopped promising

The old venture rule was that a firm would not back two companies that compete head-on. Benchmark now says that rule cannot be counted on. Chetan Puttagunta tied it straight to AI: as the cost of building software "effectively goes to zero," founders are ambitious, they see a surface area, and "you start to run into each other." Peter Fenton called conflicts the firm's "kryptonite."

This matters to you beyond cap tables. When features commoditize, companies expand into each other's turf, and the people who funded you may fund the team coming for your lane. OpenAI and Anthropic already share dozens of investors. The Factory-Cognition fight ended with a board adviser jumping to the rival. Do not assume shared investors means shared loyalty. Assume overlap.

Three questions for your team

• Which of the four compounding advantages are we actually building, and what on next quarter's roadmap makes that loop turn faster?
• If a rival cloned our top three screens this weekend, what would still keep our customers from leaving, and is that a real workflow tax or wishful thinking?
• Where are our customers most likely to build an internal tool instead of renewing, and are we API and agent ready before they decide?