Mike Volpi | Uncapped with Jack Altman Summary

1. Key Themes

AI Has Fundamentally Broken the Core Assumptions of Venture Capital

The entire software business model — high fixed costs, low marginal costs, sell to as many as possible — is being inverted by AI, which dramatically lowers the cost of software creation. This doesn't just affect software companies; it cascades through every VC decision including go-to-market, team structure, which customers to target, and whether to be product vs. service-centric.

"Every business model starts looking like that. Then you move into an AI era, which takes the cost of making software way down. You're completely shifting the core assumptions on how a business is built. And that then extends into everything from go-to-market, engineering, fundraising, which customers you target first." - Mike Volpi 00:03:19

Stage-Centric Investing Is an Obsolete Framework in the AI Era

Traditional VC stage discipline — seed, Series A, growth, crossover — is a relic. In the current environment, a $10B valuation company can still produce venture-like returns. The right framework is magnitude of opportunity, not stage.

"I could invest in a company at $10 billion in valuation and three years later, it could be worth $380 billion. In essence, what you have is a suite of companies that we would now consider growth stage, but that represent venture-like return characteristics." - Mike Volpi 00:06:17

"Would you own 20% of some schmo company or 1% of Anthropic?" - Mike Volpi 00:15:00

The Frontier Labs Have Won — The Moat Is Compute, Not Algorithm

The five winners in the core model layer (OpenAI, Anthropic, Google, Meta, xAI) are already determined. Capital-compute correlation is so dominant that any architectural innovation from a new entrant can simply be outspent. The only exception is proprietary data that the big labs cannot access.

"Even if you came up with a far better model and you could argue that the way OpenAI and Anthropic do their things are a little more clunky, they can just throw compute at it and blow you away." - Mike Volpi 00:22:43

2. Contrarian Perspectives

Open Source AI Is Not a Sustainable Business Model — It's Already Dying

Most people believe open source AI creates durable competition and commoditizes the market. Mike argues it's fundamentally uneconomic at frontier scale and that the most committed open source players are quietly closing their models.

"The latest Quinn model is closed. MuseSpark is closed. That's not a coincidence. They're spending a lot of money training those models. They're not just going to, like, open them up." - Mike Volpi 00:24:30

"Who's going to go spend $50 billion to give it away? I mean, it just doesn't make sense." - Mike Volpi 00:26:33

Board Seats Are Overrated — High-Frequency 1-on-1s Are Superior

The VC industry treats board seats as a badge of honor and a signal of engagement. Mike argues formal board meetings are low-bandwidth and largely theatrical, and that weekly or biweekly 1-on-1s are dramatically more valuable for founders.

"I've done enough board meetings to last a few lifetimes at this point, and I don't find them as productive for knowledge extraction... Once a quarter you prepare a deck and you sit around and read the deck and all the three guys that are observers want to say something useless." - Mike Volpi 00:16:19

"Professional Services" Is Now a Feature, Not a Bug

For a decade, VCs penalized software companies for having professional services revenue as a sign of poor scalability. Today the FDE (Field Deployment Engineer) model — which is essentially professional services — is the highest-value go-to-market for AI companies.

"10 years ago when I was doing VC and we said like, 'oh, they have a professional services business model.' You're like, 'oh, no, no, no. Pass, pass.' Now they're like called FDEs. Yeah, they're really cool. Palantir, cool." - Mike Volpi 00:41:36

Venture Firms Should Be Leaner, Not Larger

The industry trend has been toward scale — large platforms with hundreds of people, multi-strategy funds. Mike argues the optimal structure is a lean, well-rounded team where every partner can source, evaluate, sell, and support — and that specialization destroys the feedback loop that builds brand.

"The firms of the future are probably leaner and more concentrated to groups of people that are well-rounded at that suite of four skill sets... once you start trying to over-specialize... you're trying to stitch together like five people to do one job." - Mike Volpi 00:18:28

Defense Investing Is Morally Net Positive, Not Ambiguous

Most ESG-oriented VCs avoid defense. Mike's evolved view is that weapons investment is a deterrent that reduces violence, not causes it — and he regrets passing on Anduril as a result of an overly narrow ethical framework.

"Weapons are a system that deters violence because the more weapons one side has as the other side have it, the less likely they are to go to war... I think the right kind of defense investing is the right thing to do." - Mike Volpi 00:47:24

3. Companies Identified

Hanabi Capital Mike Volpi's new AI-focused venture firm, managing $175M. Structured as a lean fund targeting AI natively across all stages, with emphasis on proprietary data moats and regular founder engagement.

"Hanabi is a small fund and we manage $175 million... almost every investment that I've made as part of Hanabi, I will do weekly, bi-weekly, or at least monthly check-ins with founders." - Mike Volpi 00:14:07

Anthropic AI safety-focused lab considered one of the definitive winners of the core model layer alongside OpenAI and Google.

"You would get into Anthropic at $60 billion and you're going to get like 10X plus return on it." - Mike Volpi 00:07:04

Cerebras Systems Wafer-scale AI chip company. Highlighted as a compelling alternative to NVIDIA specifically for inference workloads — not general-purpose training, but extremely fast and cost-efficient for inference.

"In inference, it's a godsend. It's super fast and it's made certain compromises that make it very, very good for that." - Mike Volpi 00:35:18

Periodic Labs Portfolio company of Hanabi. Scientific AI lab creating proprietary training data through their own lab experiments rather than relying on internet-available data — cited as a model for defensible AI application companies.

"They create their own data through their own labs and they do training and reinforcement based off of that." - Mike Volpi 00:28:47

Mind Robotics Robotics spin-out of Rivian, Hanabi portfolio company. Uses its own manufacturing facility as a proprietary data collection instrument for robotics pre-training — seen as a durable moat in the robotics space.

"Their own factory will be their data collection instrument and they will have a proprietary funnel, which will be materially differentiated than other people." - Mike Volpi 00:32:49

Scale AI Labeled data company. Mike is on the board and has been a long-time investor. Used as a case study for why commodity data provision is not a defensible position — even Scale faces constant re-bids.

"Every single time a new contract will come up, it's a dogfight. It's Scale. It's Surge. It's Merkur. It's Handshake. It's Turing." - Mike Volpi 00:30:26

Etched / Talus Next-generation inference chip companies that bake neural network weights into hardware, making them faster but less flexible than Cerebras. Cited as examples of the move toward mission-specific silicon.

"If you look at Etched or Talus, these are companies that are fundamentally baking in some parts of the neural network weights into the chip itself, which will make it even faster than Cerebras." - Mike Volpi 00:35:46

Anduril Defense technology company. Used as the prime example of a venture-backed defense company that successfully broke open the DoD market and validated the sector for future investment.

"Companies like Anduril and to a degree Palantir being successful at breaking into the market of selling into the U.S. defense infrastructure will be an inspiration for both sides." - Mike Volpi 00:46:44

Figma Design software company. Cited as an example of a pre-AI SaaS company with a founder (Dylan Field) capable of navigating the AI transition — contrasted with less adaptive peers like Workday.

"I think there's a big difference between, I don't know, Figma and Workday. I think the chances that Dylan's going to figure something out are much higher." - Mike Volpi 00:44:30

Thrive Capital Josh Kushner's growth equity firm. Highlighted as the model for organic brand building in venture — highly ranked by young founders despite minimal traditional marketing.

"I think Josh and his team over there are probably the prototypical built a brand on the down low." - Mike Volpi 00:12:04

4. People Identified

Mike Volpi GP of Hanabi Capital; former partner at Index Ventures; longtime Cisco executive (joined when hundreds of employees, left at 55,000). One of the most experienced voices combining deep operator background with multi-decade venture investing, now building an AI-native firm from scratch.

"Having had that experience of growth, recruiting, customers, relationships, all that was enormously useful for me." - Mike Volpi 00:49:40

Peter Fenton Benchmark partner. Cited as proof that operational experience is not a prerequisite for world-class venture investing.

"You look at Peter Fenton, the biggest thing he's operated was a lemonade stand. But he is one of the greatest VCs of all time." - Mike Volpi 00:50:00

Trey Stevens Investor and early Anduril backer. Gave Mike the framing that weapons are a deterrence system, not an aggression tool — a conversation that changed Mike's view on defense investing.

"I had a conversation with Trey Stevens. And he was like, 'I don't think you get the mix here. Weapons are a system that deters violence.'" - Mike Volpi 00:47:24

Dylan Field Co-founder of Figma. Singled out as an example of a pre-AI SaaS founder with the intelligence and adaptability to transition his company into the AI era.

"I think the chances that Dylan's going to figure something out are much higher, a lot higher." - Mike Volpi 00:44:30

5. Operating Insights

The Four Non-Negotiable Skills of a Great Venture Capitalist

Mike distills the entire job to four skills and warns against firms that over-specialize. Any operator building a high-performance team — whether in VC or beyond — should apply this framework: hire and develop for the full loop, not isolated functions.

"You need to find opportunities. So discovery sourcing is important. You need to have good judgment to decide whether something is good or not. You need to convince the entrepreneur to take your money. So you've got to be a salesman. And then lastly, you've got to help them develop their business so there'll be a reference so that the other things work." - Mike Volpi 00:17:32

The FDE Model Is the Highest-Leverage GTM for Enterprise AI

For any AI company selling into large enterprises, the FDE (Field Deployment Engineer) role — bridging business problems with technical solutions — is not overhead, it's the product. The contract sizes this unlocks make the cost irrelevant.

"When you're talking about the CEO of T-Mobile saying like, I need to reduce my churn by 2% and you do that for them, that's hundreds of billions... no surprise, the contract values that these people are willing to offer are huge. Because you're actually solving a business problem for them. You're not providing them with a technology." - Mike Volpi 00:43:06

Generational Bridging Is a Core Firm Design Principle

For any firm or company led by experienced operators, the design principle should be: hire cohort-relevant talent for front-line relationships, use experienced leaders to open doors upward, then transfer access over time. Teaching someone the VC craft is easier than teaching an experienced person to think like a 25-year-old.

"It's much easier to teach somebody of the current cohort a little bit about VC rather than teach an old dog how a 25-year-old thinks." - Mike Volpi 00:56:05

6. Overlooked Insights

Compute Acquisition Timing Is Now a Permanent Competitive Moat

This was mentioned almost in passing, but it's significant: OpenAI's early and aggressive compute procurement may have locked in the lowest cost-of-compute of any player in the market — possibly permanently, given TSMC wafer constraints. This is not just a balance sheet advantage; it's a structural cost moat that compounds with every inference query and cannot be replicated by simply raising more capital later.

"One of the very smart things that OpenAI did is they bought a lot, a lot, a lot of compute ahead of time. As a result, I think Google aside, because we talk about TPUs in a second, but Google aside, they probably have the lowest cost of compute of any player in the market right now." - Mike Volpi 00:34:14

The implication for investors: when evaluating AI infrastructure companies or labs, ask when and at what price they locked in compute — this may be a more important indicator of long-run margin structure than any product metric.

The Legacy Partner Economics Problem Is the Single Biggest Failure Mode in Venture

Mike identifies this almost casually as "the quintessential failure mode of every venture firm," but the insight deserves much more attention: because the VC return loop is so long (5-10 years), legacy partners systematically harvest carry from work done by successors, then resist giving away economics to the next generation. This is not a governance issue — it's a structural misalignment baked into how carry vesting and fund sequencing works, and it explains firm mortality better than almost any other variable.

"Legacy partners who no longer do useful investments take the disproportionate portion of the promotion. That is the quintessential failure mode of every venture firm." - Mike Volpi 00:20:28

The non-obvious implication: when evaluating a VC firm's longevity as a platform or as a potential LP, audit the carry distribution model and the transition economics — not just the portfolio.