Past Success Reinforcement Loops Mislead AI Venture Investors

Why the Obvious Moves in AI Venture Are the Wrong Ones and What Actually Lasts

Mike Volpi, founder of Hanabi Capital, has seen enough venture cycles to know that the most dangerous thing a firm can do is succeed. In his conversation on Uncapped, he maps out the hidden consequences of AI reshaping software, investing, and company building. The core insight isn't that AI changes everything. It's that the feedback loops from past success actively mislead the people who most need to adapt. This conversation is useful for founders and investors who sense the old playbook is breaking but haven't yet mapped the downstream effects. The advantage goes to those who can see why stage boundaries dissolve, why traditional brand building backfires, and why the biggest moat might be the willingness to feel dumb again.

The Success Trap: Why Everything You Learned Before Is Now a Liability

Volpi's central warning is about the "past success reinforcement loop." He describes how firms that dominated the SaaS era optimized for a world where software was "complicated, expensive, and takes a long time to build." That world is gone. When the cost of making software drops dramatically, every downstream assumption about go-to-market, engineering, fundraising, and customer targeting gets destabilized. The firms that can't unlearn will invest in the wrong companies using the wrong frameworks.

"The more success a firm has had, the more to use an AI term reinforcement learning there is or how things were done. And if the world shifts to a place where things are done a little bit differently that reinforcement could be applied very incorrect."

Volpi isn't just talking about old incumbents. He's talking about any investor or executive who lets past wins become cognitive anchors. The system response is insidious: the very behaviors that produced returns in the prior era feel safe and rational. But they compound risk over time as the landscape shifts. The firms that break free are the ones that actively dismantle their own confidence, what Volpi calls approaching everything with a "beginner's mind." That's not soft wisdom. It's a structural defense against the hidden cost of expertise.

When Stage Stops Mattering, Relationships Become Everything

Volpi makes a striking point: "I can invest in a company at 10 billion in valuation. And three years later, it could be worth 380 billion." That flips the traditional venture playbook on its head. Stage-based investing assumed that later-stage companies offered lower upside but more certainty. Not anymore. Companies like Anthropic produce venture-like returns even at multibillion-dollar valuations. The consequence? Firms that cling to stage labels miss the highest-return opportunities.

But here's the twist: the same person now needs to assess a $10B company and a two-person garage startup. Volpi is candid that assessment math isn't the hard part. "Looking at a spreadsheet is not rocket science." The real challenge is relationship access. A young investor can build rapport with 22-year-old founders fresh from OpenAI. They cannot walk into Dario Amodei's office. Volpi's advantage is that he can do both, but he acknowledges the friction: his job is to open pathways for his younger team to those senior relationships.

The hidden system dynamic: as stage boundaries dissolve, the bottleneck shifts from capital allocation to relationship capital. Firms that can span generational and power gaps will compound advantages. Those that can't will find themselves locked out of the best deals at every layer.

Brand That Can't Be Bought

Volpi's most counterintuitive argument is about venture brand. Traditional marketing, sponsorships, banners, polished content, doesn't just fail with young founders. It actively signals the wrong thing.

"Classical marketing efforts that have been done for brand building, particularly in venture sound off key to your average 22 year old entrepreneur. I think what does resonate is inside knowledge tips connections network all those things which are kind of indirect but organic ways of building brand."

The downstream effect is brutal: if you try to market your way to brand, you spend money on signaling that feels like an advertisement. But if you invest in being genuinely helpful, giving advice, making introductions, solving thorny problems, the brand emerges organically through references. Volpi notes that this is hard to fake because you can't pitch "I'm there when it's hard." That has to be demonstrated over time. The payoff is delayed but durable. It creates a brand that "you can't throw banners at." Most firms won't wait that long, which is precisely why it works.

Why Neo Labs and Open Source Won't Win (But Robotics Might)

Volpi's systems thinking is clearest when he traces the compute dynamics. The big labs, OpenAI, Anthropic, Google, Meta, potentially X, spend $50-100 billion annually on compute. Any new lab raising $2 billion is an order of magnitude behind. Even if a neo lab invents a better algorithm, the incumbents can "just throw compute at it and blow you away." The bitter lesson of AI is that scale dominates.

Open source? Volpi calls it "not a business." The monetizable demand is at the frontier, and the frontier models cost too