Martin Casado on Frontier AI Economics, Infrastructure Disruption, and Strategic M&A Dynamics
Summary
Marc Andreessen Horizon Ventures partner Martin Casado categorizes venture investing into four mindsets, advocating for a 'productive' approach focused on technical infrastructure rather than traditional financial metrics. He argues that frontier AI models have short lifespans of 3-6 months and that current market leaders hold ~80% share due to unsaturated demand, though this may shift if the market slows. Casado identifies physical hardware, power, and materials as the primary bottlenecks in AI infrastructure, predicting consolidation among large providers. He also details the strategic M&A dynamics behind the Cursor acquisition, emphasizing trust and alignment over adversarial tactics.
Episode sections
- 1:48 – 6:41
Four Investing Mindsets and Productive Methodology
Casado categorizes investing into four mindsets and describes his 'productive' approach focusing on technical infrastructure and early willingness to pay.
- 6:41 – 17:57
Deconstructing Network Effects and Defensibility
He argues that true defensibility requires superlinear growth (Metcalf's Law) and that defensibility is often overrated in rapidly growing markets.
- 17:57 – 22:33
Infrastructure Refactoring for AI
Casado explains how AI models act like ASICs, disrupting the traditional infrastructure stack and forcing a re-evaluation of databases and networking.
- 22:33 – 24:22
Frontier Model Economics and Market Slowdown
He analyzes the economics of frontier AI, noting current pricing power and market share, but raises concerns about saturation and the shift to open source.
- 24:22 – 31:02
Model Lifespan and Competitive Dynamics
Casado argues that frontier AI models have a short 3-6 month lifespan and that competitive pressure prevents downgrading to cheaper models.
- 31:02 – 34:05
Digital vs. Physical World Solvability and Cybersecurity
He asserts that digital problems can be solved by AI through simulation, while expressing uncertainty about physical world applications, and compares AI security to the early internet era.
- 34:05 – 44:04
Capital Markets and Infrastructure Supply Constraints
Casado discusses debt market constraints and identifies hardware, power, and materials as the primary bottlenecks in AI infrastructure.
- 44:04 – 46:20
Private Market Efficiency and Indexing
He states that private markets are efficient and investors should trust market pricing, aiming to index competitive deals rather than seek non-consensus opportunities.
- 46:20 – 47:00
Frontier Model Investment Strategy
Casado describes his firm's approach as stratified sampling, selecting the top company in each sector to achieve low tracking error.
- 47:00 – 51:55
Capital Creation vs. Zero-Sum Thinking
He refutes the belief that private markets are zero-sum, arguing that capital deployment expands TAM and creates value.
- 51:55 – 56:43
Strategic M&A Dynamics and the Cursor Deal
Casado outlines M&A trends, using the Cursor acquisition as a case study for successful strategic partnership based on alignment and trust.
- 56:43 – 1:01:36
M&A Negotiation Tactics and Evolution of the Investor Role
He criticizes adversarial M&A tactics and leaking, noting a mismatch between late-stage investors and early-stage company needs.
- 1:01:36 – 1:06:54
Founder-Investor Meetings and AI as Industrial Revolution
Casado advises investors to listen more during founder meetings and concludes by describing AI as an industrial revolution-level shift.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| OpenAI | private company | neutral | OpenAI, Cursor, Anthropic, Perplexity, and Vercell all have something in common. |
| Anthropic | private company | neutral | OpenAI, Cursor, Anthropic, Perplexity, and Vercell all have something in common. |
| SpaceX | private company | neutral | Again, this is one of these great examples where in the end, SpaceX had the compute and cursor did have the data, and it had a team that had very successfully built models and built this practice and code was very important. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
Ramp saves businesses 5% annually on average.
Unidentified speaker ·0:00
How to verify: Check Ramp's marketing materials or case studies published on ramp.com/invest for the specific metric source and methodology.
Rogo serves tens of thousands of bankers and investors at hundreds of leading firms.
Unidentified speaker ·0:28
How to verify: Verify via Rogo's website (rogo.com/invest) or press releases detailing their customer base size.
OpenAI, Cursor, Anthropic, Perplexity, and Vercel all use WorkOS.
Unidentified speaker ·1:03
How to verify: Check WorkOS's customer list on workos.com or public engineering blogs from these companies regarding their identity/authentication infrastructure.
Chris Dixon invested in Soylent, Oculus, Coinbase, and crypto.
Unidentified speaker ·3:20
How to verify: Review Chris Dixon’s public investment portfolio or historical press releases from Andreessen Horowitz (a16z) regarding these specific investments.
Vanta cuts audit prep by 82%.
Unidentified speaker ·16:35
How to verify: Check Vanta’s case studies or white papers cited on vanta.com/invest for the data source and sample size.
More than 16,000 companies use Vanta.
Unidentified speaker ·16:45
How to verify: Verify via Vanta’s official website or press releases announcing their customer milestone.
Ridgeline is the first end-to-end system of record with embedded AI for investment management firms.
Unidentified speaker ·16:59
How to verify: Check Ridgeline’s marketing claims on ridgeline.ai and compare against competitors in the investment management software space.
Frontier AI models tend to get 80% plus of the market dollar weighted.
Unidentified speaker ·23:26
How to verify: Analyze enterprise AI spending reports from firms like Gartner, IDC, or McKinsey to verify market share distribution among top AI model providers.
AI penetration in the enterprise is approximately 3% by headcount but 1% or less by budget.
Unidentified speaker ·23:51
How to verify: Cross-reference with enterprise AI adoption surveys from major consulting firms (e.g., McKinsey, BCG) or tech analytics firms (e.g., Gartner) for budget vs. usage penetration data.
The networking infrastructure market saw no new major entrants since Arista until the current AI-driven refresh.
Unidentified speaker ·21:57
How to verify: Review the history of enterprise networking hardware vendors and IPOs/launches from Arista’s founding to present to confirm the lack of significant new competitors in that period.
The lifespan of a frontier AI model is three to six months.
Unidentified speaker ·24:50
How to verify: Check industry reports on model release cycles and adoption rates of older models versus new releases from major AI labs.
Casado has automated 95% of his software coding tasks.
Unidentified speaker ·26:01
How to verify: This is a personal anecdote; verification would require Casado's public statements or interviews confirming this specific metric.
In competitive equilibria, startups will always pay more for the smartest model rather than using a cheaper, older one.
Unidentified speaker ·26:24
How to verify: Analyze case studies of startup software stacks and AI adoption strategies to see if cost optimization is prioritized over performance in competitive markets.
Any digital problem can be solved by AI if sufficient capital is available to create simulation and RL environments.
Unidentified speaker ·31:19
How to verify: Review academic and industry literature on the limits of reinforcement learning and simulation in solving complex digital tasks.
Windows 95 was so vulnerable that it would be infected by worms before the installation process was complete.
Unidentified speaker ·33:00
How to verify: Search historical cybersecurity records and news archives from the late 1990s regarding Windows 95 vulnerabilities and early internet worms.
There are 500 proposed legislations to stop data center building in the United States alone, including Texas.
Unidentified speaker ·43:44
How to verify: Review legislative databases and news reports regarding data center zoning, environmental regulations, and energy usage bills in the US.
Larry Page stated that five nines (99.999%) of compute will eventually be orbital.
Unidentified speaker ·43:33
How to verify: Search for transcripts or interviews with Larry Page discussing orbital compute and data center infrastructure.
Private markets are very efficient, and investors should trust market pricing.
Unidentified speaker ·44:43
How to verify: This is an opinion; verify by looking for Casado's consistent public statements on market efficiency and private market valuation strategies.
Casado owns a 5% stake in every competitive deal he participates in.
Unidentified speaker ·45:42
How to verify: Review Casado's investment portfolio disclosures or interviews where he discusses his deal terms and ownership percentages.
80% of the Neo Cloud/Neo Lab market will consolidate around scale providers, while 20% remains fragmented.
Unidentified speaker ·42:44
How to verify: Monitor market share data of major AI infrastructure providers versus niche players in the GPU and compute market.
VMware acquired Sierra for $1.26 billion, which was the highest acquisition multiple of revenue for enterprise software at the time.
Unidentified speaker ·52:10
How to verify: Search for historical press releases regarding VMware's acquisition of Sierra and financial reports on the deal value relative to Sierra's revenue.
Yammer's acquisition by Microsoft had a revenue multiple slightly lower than Sierra's, making it one of only two outliers in historical charts at the time.
Unidentified speaker ·52:20
How to verify: Review historical data on Microsoft's acquisition of Yammer and compare the purchase price to Yammer's reported revenue.
The acquisition of Tableau by Databricks beat the revenue multiple record previously held by Sierra.
Unidentified speaker ·52:42
How to verify: Check public filings or news reports regarding the Databricks acquisition of Tableau and compare the valuation multiples.
Andrew Rosen (referred to as Roloff) stated that venture capital is not an asset class and that only 20 startups matter.
Unidentified speaker ·48:52
How to verify: Search for Andrew Rosen's speeches or interviews at TechCrunch Disrupt where he discusses the size of the venture market.
Building the 'Mythos' model likely cost approximately $3 billion.
Unidentified speaker ·55:41
How to verify: This is an estimate by Casado; verify by looking for public estimates of compute costs for training large foundation models of that era, though the exact figure is internal to the speaker's calculation.
Do your own research
- Verify the claim that frontier AI models have a 3-6 month lifespan by reviewing release cycles and version updates of major models (e.g., OpenAI, Anthropic, Google) over the last 12 months.
- Check the market share data for AI model providers to validate the ~80% claim, using reports from firms like Synergy Research Group or Gartner.
- Assess the 'hardware supply, power, and materials' bottleneck by reviewing recent earnings calls from major chipmakers (NVIDIA, AMD) and energy providers for capacity constraints.
- Review the Cursor acquisition details and subsequent performance to evaluate Casado's claims about strategic M&A alignment and 'inductive proof'.
- Analyze the definition of 'superlinear network effects' versus general flywheels by examining user growth vs. value metrics for platforms like Facebook or LinkedIn.
- Investigate the regulatory environment for data center construction in key regions like Texas to assess the risk of offshore or orbital compute shifts mentioned.
- Compare the 'productive investing' methodology against traditional SaaS metrics (NRR, margins) in current VC term sheets to identify potential mismatches in evaluation criteria.
Full transcriptexpand
- 0:00
- Unidentified speaker: Ramp is the only platform built to make your finance team leaner, faster, and better, saving businesses 5% annually on average so you can stay focused on growth.
- 0:09
- Unidentified speaker: Ramp just opened for business in the United Kingdom.
- 0:12
- Unidentified speaker: So if you're running a business in the UK, you can now use Ramp's AI-powered finance platform to manage cards, expenses, bills, approvals, and accounting all in one place.
- 0:22
- Unidentified speaker: I run my business on Ramp, and so should you.
- 0:24
- Unidentified speaker: Learn more at ramp.com slash invest.
- 0:28
- Unidentified speaker: Rogo is the AI platform purpose-built for financial institutions, serving tens of thousands of bankers and investors at hundreds of leading firms worldwide.
- 0:36
- Unidentified speaker: Rogo's AI agents autonomously execute large chunk of your firm's workflows.
- 0:41
- Unidentified speaker: They can screen deals, draft STIMs, run buyer outreach, and due diligence on data rooms, all with full security and compliance handled.
- 0:49
- Unidentified speaker: Every action feeds Rogo Intelligence, the firm's context layer, turning years of deal experience into a secure, governed institutional memory that makes each new transaction stronger than the last.
- 0:59
- Unidentified speaker: To learn more, visit rogo.com/slash invest.
- 1:03
- Unidentified speaker: OpenAI, Cursor, Anthropic, Perplexity, and Vercell all have something in common.
- 1:08
- Unidentified speaker: They all use WorkOS.
- 1:10
- Unidentified speaker: And here's why.
- 1:12
- Unidentified speaker: To achieve enterprise adoption at scale, you have to deliver on core capabilities like SSO, Skim, RBAC, and Audit Logs.
- 1:20
- Unidentified speaker: That's where WorkOS comes in.
- 1:22
- Unidentified speaker: Instead of spending months building these mission-critical capabilities yourself, you can just use WorkOS APIs to gain all of them on day zero.
- 1:30
- Unidentified speaker: That's why so many of the top AI teams you hear about already run on WorkOS.
- 1:34
- Unidentified speaker: Work OS is the fastest way to become enterprise ready and stay focused on what matters most, your product.
- 1:40
- Unidentified speaker: Visit workos.com to get started.
- 1:48
- Unidentified speaker: You recently wrote about four buckets of styles of investing that might be applied, call it venture, private market tech, let's call it.
- 1:54
- Unidentified speaker: Can you enumerate those and just like say a sentence on effective and each?
- 1:58
- Unidentified speaker: I'm just in this point in my career, I've been doing this for a decade, I've been hundreds of deals that I sit in these discussions and I just realize people aren't communicating very well.
- 2:06
- Unidentified speaker: So I did this exercise where I'm like, what are the mindsets people bring to investing?
- 2:10
- Unidentified speaker: And I think there's basically four types of mindsets.
- 2:12
- Unidentified speaker: So the one that we were talking about before is like this traditional finance mindset, which is the correct proxy for the health of a business is the market it's going into and its finances.
- 2:22
- Unidentified speaker: What are the SaaS metrics?
- 2:24
- Unidentified speaker: What is the size of the market?
- 2:26
- Unidentified speaker: And that is a proxy for the entire business.
- 2:28
- Unidentified speaker: So that's one.
- 2:29
- Unidentified speaker: Clearly, growth investors do this, but like more and more you see early stage Series A people do this, which to me is a little bit discordant.
- 2:36
- Unidentified speaker: Then there's what I call productive investors, which you believe you're investing in a productive asset.
- 2:40
- Unidentified speaker: And the productive asset is what's creating value.
- 2:43
- Unidentified speaker: And if you believe in that, as it creates value, it will create market.
- 2:46
- Unidentified speaker: And as it creates value, you're gonna have to find ways for the business to catch up to that value.
- 2:50
- Unidentified speaker: And so the type of diligence you do if you're a productive investor is more like what is the technical approach?
- 2:55
- Unidentified speaker: What are the adjacent markets?
- 2:57
- Unidentified speaker: How fast is this likely to evolve?
- 2:59
- Unidentified speaker: How close is it to capturing value?
- 3:01
- Unidentified speaker: Like that sort of stuff.
- 3:01
- Unidentified speaker: Is this the right team to build it?
- 3:03
- Unidentified speaker: Then there's what I call frontier investing.
- 3:05
- Unidentified speaker: I think this is the romantic idealized version of VC from like the 60s and 70s, which is kind of YOLO, man.
- 3:12
- Unidentified speaker: It's like founders know better than VCs and they've seen something that I don't see, and they're gonna go out there and do it.
- 3:18
- Unidentified speaker: And like, who am I to second guess them?
- 3:20
- Unidentified speaker: And they almost look like talent networks, right?
- 3:21
- Unidentified speaker: Like I would say Chris Dixon.
- 3:22
- Unidentified speaker: I mean, remember when he was investing in nootropics, he invested in Soylent, he invested in Oculus, he invested in Coinbase and in crypto.
- 3:29
- Unidentified speaker: I mean, it's like every science fiction trope from the 70s and 80s you kind of invest in, and you just don't overthink any of these things.
- 3:36
- Unidentified speaker: So it's almost like the strong form of the productive.
- 3:39
- Unidentified speaker: And then the last one, and it is always so surprising how prevalent it is, is speculative investors.
- 3:45
- Unidentified speaker: They don't even care about the asset, they just want to time a market.
- 3:49
- Unidentified speaker: You'll see this like in crypto and in secondary markets.
- 3:51
- Unidentified speaker: They don't know about the underlying asset.
- 3:53
- Unidentified speaker: They probably don't even know about the actual business metrics, but they're like, well, this looks cheap.
- 3:59
- Unidentified speaker: It's clearly going to go up in value.
- 4:01
- Unidentified speaker: Oh, this is the top of the market, you know, like I'm gonna get out.
- 4:04
- Unidentified speaker: They're trying to like time markets.
- 4:05
- Unidentified speaker: They're actually just doing speculative.
- 4:07
- Unidentified speaker: They're not really interested in like, will the business grow?
- 4:10
- Unidentified speaker: And again, the problem is if you have all of these archetypes, you can justify any deal or not justify any deal.
- 4:16
- Unidentified speaker: So if you don't know what your beliefs are, you'll always justify whatever personal bias you have.
- 4:20
- Unidentified speaker: If you think about yourself as a productive investor, focused on, I love this idea that disruptive technologies, you should not look at some of these traditional things that so many investors talk about.
- 4:31
- Unidentified speaker: Talk about what you do look at.
- 4:33
- Unidentified speaker: Like you look at so many companies.
- 4:35
- Unidentified speaker: What is your method by which to assess whether or not someone might be creating a truly disruptive piece of technology, a valuable asset?
- 4:43
- Unidentified speaker: So I only do computer science investing and I only do infrastructure investing.
- 4:46
- Unidentified speaker: And I think infrastructure as an industry has a certain shape.
- 4:51
- Unidentified speaker: Certain buying behaviors, the technology looks a certain way and the industry acts a certain way.
- 4:56
- Unidentified speaker: And so what you're trying to do is you're trying to take a technical approach and you're trying to map it into that techno-strategic framework.
- 5:03
- Unidentified speaker: And the way you do that, there's a number of approaches.
- 5:05
- Unidentified speaker: One of them, you're like, what are the leading companies doing internally?
- 5:08
- Unidentified speaker: And this will give you a glimpse on maybe this is how platforms evolve.
- 5:12
- Unidentified speaker: Or perhaps the team that is doing this didn't it another context.
- 5:17
- Unidentified speaker: And now they're actually bringing it to this context.
- 5:20
- Unidentified speaker: I think it's one of these things where, like, how much can you diligence without actually just using the numbers for a proxy?
- 5:26
- Unidentified speaker: You can talk to customers, will they use this?
- 5:28
- Unidentified speaker: You can see if there's like adjacent budgets that are growing that could arguable fragment to become this.
- 5:35
- Unidentified speaker: Or maybe you have early signs of willingness to pay, but they actually haven't figured out the business model.
- 5:39
- Unidentified speaker: Like I always think every tech wave, you know, I've been in the industry for 30 ways.
- 5:43
- Unidentified speaker: Every tech wave, people like hand-ring about margins.
- 5:47
- Unidentified speaker: Everyone, like I remember for the cloud, and I remember for mobile, I remember for the internet.
- 5:51
- Unidentified speaker: Clearly, now people worry about margins.
- 5:53
- Unidentified speaker: So maybe you're like, there's a willingness to pay and there's a lot of interest, but we haven't figured out the cogs yet and the actual viable business model.
- 6:00
- Unidentified speaker: So, like you're using partial indicators, not full indicators in order to do this.
- 6:03
- Unidentified speaker: And so my team is a lot of technical people, product people, many have PhDs.
- 6:09
- Unidentified speaker: We don't second guess the tech.
- 6:10
- Unidentified speaker: Like the founder will know the tech better than us, but but we can ask the question how does this tech fit into the broader platforms, fit into spending?
- 6:17
- Unidentified speaker: How does this tech fit in with our understanding of stickiness and defensibility?
- 6:21
- Unidentified speaker: And then we look at early market numbers and we'll say, hey, listen, a financial investor would puke at this, but it actually shows that the customers are really interested to figure something out and will they be able to figure something out?
- 6:33
- Unidentified speaker: So there's a lot of diligence you can do.
- 6:35
- Unidentified speaker: Where with frontier investing, I don't think you do any of that.
- 6:37
- Unidentified speaker: It's people, is this the best team to do this crazy thing and then you just yellow?
- 6:41
- Unidentified speaker: You were saying before we hit start that nobody understands network effects.
- 6:44
- Unidentified speaker: Can you can you explain what you meant?
- 6:46
- Unidentified speaker: I've realized that even the most common terms I use, there's no consensus on.
- 6:50
- Unidentified speaker: And so, for example, network effects is the favorite one that I pick on.
- 6:54
- Unidentified speaker: I asked you before.
- 6:55
- Unidentified speaker: I messed it up.
- 6:56
- Unidentified speaker: No, you didn't mess it up.
- 6:57
- Unidentified speaker: You said you said what like a third of people say.
- 6:59
- Unidentified speaker: So I said, So what is a network effect?
- 7:01
- Unidentified speaker: And what did you say?
- 7:01
- Unidentified speaker: I said a product that gets better as it adds a marginal user.
- 7:05
- Unidentified speaker: Right.
- 7:05
- Unidentified speaker: So to me, this is almost a tautological definition because most products will get better as you add a user, right?
- 7:12
- Unidentified speaker: They will find a bug, they will provide feedback, they will tell other people.
- 7:16
- Unidentified speaker: That doesn't say anything about defensibility.
- 7:18
- Unidentified speaker: But other people, when they use the term network effects, they actually mean the original meaning of the term network effect from the 70s, which is a telephone network where when users join, they interact with each other and that grows super linearly, and that's why you get like natural monopolies.
- 7:33
- Unidentified speaker: And other people use the term network effect more to mean like an actual flywheel, so like a self-reinforcing flywheel.
- 7:40
- Unidentified speaker: So not just like things get better, but like it self-reinforces in some sort of a way.
- 7:44
- Unidentified speaker: This is where you hear the term like data network effects.
- 7:46
- Unidentified speaker: But even in that sense, like you don't know if that self-reinforcement is superlinear or not.
- 7:51
- Unidentified speaker: So you don't know if it makes it defensible or not.
- 7:53
- Unidentified speaker: So you have this very common word that's on 70% of pitch decks, where a founder is sitting in front doing a pitch and nobody has consensus on what it means.
- 8:03
- Unidentified speaker: And so, how would you define it?
- 8:04
- Unidentified speaker: So I think the root of the confusion is not a definitional problem.
- 8:07
- Unidentified speaker: I think the root of the confusion is conceptually we conflate the mechanism with the shape of the moat.
- 8:15
- Unidentified speaker: There's many things that could create a moat, right?
- 8:18
- Unidentified speaker: Like a brand, scale, uh two-sided marketplace, uh network effect.
- 8:23
- Unidentified speaker: But that moat to reason about that is only really useful if it has some properties that makes the company defensible against somebody else.
- 8:32
- Unidentified speaker: If that moat grows superlinearly with scale, like a classic Metcalf's law network effect would, that's really defensible.
- 8:40
- Unidentified speaker: If it grows sublinearly with scale, it may not be useful at all.
- 8:44
- Unidentified speaker: And so I think to you and I have a conversation of defensibility, I think the first thing we need to do is not conflate the shape of the moat with the mechanism that caused it.
- 8:53
- Unidentified speaker: So first, I think we should say the moat that we have is superlinear, it's linear, or it's sublinear with scale.
- 8:59
- Unidentified speaker: And then the second thing, a lot of things actually contribute to a moat.
- 9:02
- Unidentified speaker: For example, like we mentioned, network effects will, or scale effect.
- 9:06
- Unidentified speaker: What's a scale effect?
- 9:07
- Unidentified speaker: A scale effect to me is as the company scale, it can reduce unit economics normally because it has pricing power, for example, or maybe CAC goes down because of brand.
- 9:16
- Unidentified speaker: What's a two-sided marketplace?
- 9:17
- Unidentified speaker: That's very different to me than a network effect.
- 9:19
- Unidentified speaker: That's where basically you've got self-reinforcing supply and demand.
- 9:22
- Unidentified speaker: So I think that there are many mechanisms, and then there's the shape.
- 9:26
- Unidentified speaker: And then to have a discussion, I think you want to make sure that the shape is something that actually provides long-term defensibility.
- 9:32
- Unidentified speaker: Then the last thing I'll say is it's almost impossible to have these conversations in isolation because we're actually talking about competitive equilibria.
- 9:39
- Unidentified speaker: Because if you have a network effect, then maybe the competitor has a network effect too.
- 9:43
- Unidentified speaker: And maybe if they have more capital than you and grow faster, that probably impacts the shape of your mode.
- 9:49
- Unidentified speaker: You may not grow as fast marginally because they're stealing your customers or something like that.
- 9:54
- Unidentified speaker: And so I just think that we need like very, very basic concepts.
- 9:57
- Unidentified speaker: We need to not conflate the notion of the actual shape with the actual mechanism.
- 10:01
- Unidentified speaker: And then I think we should enumerate the mechanisms because something like the definition that you had just doesn't really help anybody.
- 10:08
- Unidentified speaker: Can you give examples of superlinear, linear, or sublinear just to bring it to life?
- 10:12
- Unidentified speaker: Yeah, for sure.
- 10:13
- Unidentified speaker: A common view of where network effects came from was from the 70s.
- 10:16
- Unidentified speaker: I forgot the name of the economist that did it, but he worked at Bell Labs.
- 10:19
- Unidentified speaker: And he noticed for train networks and communications networks, the interactions of users grow with the square of the number of people that join.
- 10:30
- Unidentified speaker: As a user joins, they can communicate to many other people.
- 10:33
- Unidentified speaker: So the actual number of connections grows more than the number of users.
- 10:37
- Unidentified speaker: Therefore, the value grows more than just linearly with them adding.
- 10:42
- Unidentified speaker: The same thing with destinations on a highway or destinations on a train.
- 10:46
- Unidentified speaker: So that would mean that if you have two competing companies, one takes $5 million of funding and the other takes $5 million of funding, but this one is marginally better, marginally larger.
- 10:55
- Unidentified speaker: This one's a little bit larger.
- 10:57
- Unidentified speaker: For the same amount of investment, it will grow faster than a challenger.
- 11:01
- Unidentified speaker: The growth actually increases over time, and that's why you get a monopoly.
- 11:05
- Unidentified speaker: Whereas sublinear means if I'm a leader and I invest the same amount as a challenger, I grow marginally slower than the challenger.
- 11:12
- Unidentified speaker: That would be sublinear.
- 11:13
- Unidentified speaker: Clearly, this is very important when it comes to modes because if you have two competing companies and there's a sublinear mode, the challenger will always catch up to the leader because the leader has to spend ever more money to grow all things being equal.
- 11:28
- Unidentified speaker: Favorite recent company examples of these different types?
- 11:31
- Unidentified speaker: How many actual network effects in the Metcalf's law of superlinear do you think there have actually been?
- 11:36
- Unidentified speaker: Very few.
- 11:36
- Unidentified speaker: I don't know.
- 11:37
- Unidentified speaker: I feel like this almost never happens.
- 11:39
- Unidentified speaker: Like I think we can all agree the telcos.
- 11:41
- Unidentified speaker: That's why we had to break them up.
- 11:42
- Unidentified speaker: We could probably all agree LinkedIn.
- 11:44
- Unidentified speaker: They haven't had to change very much and they seem to have a permanent status.
- 11:48
- Unidentified speaker: Social networking.
- 11:49
- Unidentified speaker: But all of these examples, it's like the interconnections of people.
- 11:53
- Unidentified speaker: I would say GitHub was another one that had a interconnection.
- 11:57
- Unidentified speaker: It's interconnections of users, right?
- 11:59
- Unidentified speaker: So I think there are not very many examples, but when you have them, these are these very durable, high-value companies.
- 12:05
- Unidentified speaker: So that to me would be like a classic network effect.
- 12:08
- Unidentified speaker: Other companies probably are more scale effects.
- 12:11
- Unidentified speaker: For example, people will say Uber.
- 12:13
- Unidentified speaker: I'm not sure that that would fall the definition of a network effect, maybe a two-sided marketplace because you have the drivers and you have the passengers.
- 12:20
- Unidentified speaker: Maybe you could use that.
- 12:21
- Unidentified speaker: But even then, you don't have a lock on either of them.
- 12:24
- Unidentified speaker: Like all drivers drive for everybody, and all passengers could use anybody.
- 12:27
- Unidentified speaker: That was more like of a funding scale effect because they could raise more money and therefore the unit economics went down.
- 12:33
- Unidentified speaker: My meta observation.
- 12:34
- Unidentified speaker: So, first off, I actually think the network effecting was more of like a personal annoyance and me being a pedantic.
- 12:40
- Unidentified speaker: But I actually think defensibility in most is not really even worth talking about most of the time.
- 12:44
- Unidentified speaker: Let me explain why.
- 12:45
- Unidentified speaker: Here's my experience for what it's worth.
- 12:47
- Unidentified speaker: If a market you're investing in is growing, then companies are not going to want to compete.
- 12:52
- Unidentified speaker: There's like a lot of new white space that gets created.
- 12:55
- Unidentified speaker: And so what they'll try and do is they'll try and just move into other white space.
- 12:58
- Unidentified speaker: Talking about defensibility is kind of a second or third order thing if the market is growing.
- 13:03
- Unidentified speaker: And if the market is slowing down or shrinking, normally it consolidates anyways, independent of modes or not.
- 13:10
- Unidentified speaker: And so I do think we spend a lot of time as investors hand-wringing about defensibility when in practice it doesn't really matter, especially in growing markets, which we invest in.
- 13:19
- Unidentified speaker: And we've seen this with AI.
- 13:20
- Unidentified speaker: I remember early discussions on models.
- 13:23
- Unidentified speaker: I mean, I can't tell you how many investors are like, I will never invest in a model.
- 13:26
- Unidentified speaker: It'll never accrue value.
- 13:28
- Unidentified speaker: It's not defensible.
- 13:29
- Unidentified speaker: That's fine.
- 13:29
- Unidentified speaker: But as long as the market's growing in a large market, it's going to accrue a lot of value.
- 13:32
- Unidentified speaker: I remember this about the neo clouds.
- 13:34
- Unidentified speaker: They're never defensible.
- 13:35
- Unidentified speaker: The quality of revenue is bad.
- 13:36
- Unidentified speaker: I think that we kind of trick ourselves as investors by thinking in terms of defensibility when if there's a growing market, it's kind of not even worth considering.
- 13:46
- Unidentified speaker: Do you think that's a unique privilege of your stage of investor?
- 13:50
- Unidentified speaker: And then once you get later stage public echo, like someone that's buying the anthropic S1 needs to probably think about defensibility.
- 13:57
- Unidentified speaker: I mean, it depends on your thesis on the size of the market.
- 13:59
- Unidentified speaker: I mean, if you believe we're 0.1% penetrated, this has got a long way to go and value will accrue.
- 14:04
- Unidentified speaker: And if you're liquid, you can go in and out of the market if values accrues.
- 14:08
- Unidentified speaker: I think the nature of venture has really changed, actually, in general.
- 14:13
- Unidentified speaker: If you read Arthur Rock, it was really frontier-like investing.
- 14:18
- Unidentified speaker: And people didn't think about quality of revenue, you know, they didn't think about defensibility so much.
- 14:23
- Unidentified speaker: It was like people, and it was tech and it was new markets.
- 14:26
- Unidentified speaker: Over the last 30 years, the tech market has grown so large.
- 14:31
- Unidentified speaker: The amount of money that's available to investors has grown so much.
- 14:34
- Unidentified speaker: And then the size of the private companies has grown.
- 14:36
- Unidentified speaker: Like a private company today rivals public companies of just 10 years ago.
- 14:40
- Unidentified speaker: And so I say, on some level, VCs have actually become public investors.
- 14:45
- Unidentified speaker: And then a lot of the people that have entered VC have come from more business school and traditional finance backgrounds.
- 14:51
- Unidentified speaker: So I think a lot of the mindset comes from that world, but I think it gets applied too early.
- 14:57
- Unidentified speaker: Financial mindset gets applied too early.
- 14:59
- Unidentified speaker: Like a traditional investing financial mindset gets applied too early.
- 15:04
- Unidentified speaker: And it's very often, for example, I talk to like a junior investor and they're like, oh, the quality of revenue is bad.
- 15:08
- Unidentified speaker: I'm like, homie, we don't even know how to monetize the internet.
- 15:12
- Unidentified speaker: Like business models always lag disruptive technologies.
- 15:17
- Unidentified speaker: If you truly believe you're investing in a disruptive technology, and that's what you truly believe in, A, you should believe it creates TAM.
- 15:25
- Unidentified speaker: So why would you worry about market size?
- 15:27
- Unidentified speaker: Because it's creating market size.
- 15:28
- Unidentified speaker: And B, it's disruptive.
- 15:29
- Unidentified speaker: So we don't even know what the pricing model is.
- 15:31
- Unidentified speaker: Like, remember like how long it took for us to go from like perpetual to like recurring years of hand-wringing, et cetera.
- 15:38
- Unidentified speaker: And so my caution is like most investors should know what kind of investor they are.
- 15:42
- Unidentified speaker: Maybe you're a financial investor.
- 15:44
- Unidentified speaker: That's how you view the world.
- 15:45
- Unidentified speaker: But then you're not actually evaluating tech.
- 15:47
- Unidentified speaker: You're not evaluating the disruptive power tech.
- 15:49
- Unidentified speaker: You're literally looking at the quality of revenue, which, by the way, is going to naturally move you towards non-disruptive technology.
- 15:56
- Unidentified speaker: Because if it's disruptive, you probably don't know what the business model is.
- 16:00
- Unidentified speaker: You probably don't know what the TAM is.
- 16:01
- Unidentified speaker: On the other hand, if you believe I invest in tech and tech creates value, and that's the type of invest you are and you do that early, you should not consider these other things.
- 16:11
- Unidentified speaker: I think it's less about is there a right way to invest?
- 16:14
- Unidentified speaker: No, these are all legitimate, coherent right ways to invest.
- 16:16
- Unidentified speaker: It's more about don't be confused in your basic approach because otherwise you become incoherent.
- 16:23
- Unidentified speaker: You're like, oh, I'm a disruptive tech investor, but I'm looking for a good NRR.
- 16:26
- Unidentified speaker: Like it doesn't work that way.
- 16:29
- Unidentified speaker: Vanta automates compliance so your team can spend less time on security reviews and more time getting customers.
- 16:35
- Unidentified speaker: Vanta cuts audit prep by 82% and gives you instant, up-to-date proof of trust without all the manual work, so you can close deals faster and with less friction.
- 16:45
- Unidentified speaker: Customers report a 526% return on investment, and more than 16,000 companies use Vanta, including Ramp, Harvey, and Snowflake.
- 16:54
- Unidentified speaker: Get started with $1,000 off at vanta.com slash invest.
- 16:59
- Unidentified speaker: Ridgeline is the first end-to-end system of record with embedded AI for investment management firms running portfolio accounting, reconciliation, reporting, trading, and compliance on one unified platform.
- 17:10
- Unidentified speaker: Firms are moving off legacy technology and onto Ridgeline because of how far ahead Ridgeline's AI features are compared to anything else in investment management software.
- 17:20
- Unidentified speaker: I've been hearing from a lot of investment managers about AI, and they fall roughly into two camps, with some unsure where to even start, and others convinced they can build their own order management system over a weekend.
- 17:30
- Unidentified speaker: The reality is that running an investment firm will always require governance, controls, and a single source of truth for your data.
- 17:36
- Unidentified speaker: And no amount of AI enthusiasm changes that requirement.
- 17:40
- Unidentified speaker: Ridgeline is built on exactly that foundation, which is why I believe that the firms that come out ahead in the AI era will be the ones running on Ridgeline's unified platform.
- 17:49
- Unidentified speaker: If you're serious about your firm's AI strategy, Ridgeline should be part of that conversation.
- 17:53
- Unidentified speaker: And you can request a demo at ridgeline.ai.
- 17:57
- Unidentified speaker: How would you describe the state of the union today in your world in infrastructure?
- 18:01
- Unidentified speaker: Like obviously it's changing really, really fast.
- 18:03
- Unidentified speaker: There's new kinds of companies, but there do seem to be some layers or some categories.
- 18:08
- Unidentified speaker: Yeah.
- 18:08
- Unidentified speaker: How would you describe what it looks like today?
- 18:10
- Unidentified speaker: Yeah.
- 18:10
- Unidentified speaker: So for me, the definition of infrastructure is like the differentiation is actually technical.
- 18:16
- Unidentified speaker: Where I don't view SaaS or apps as that way.
- 18:19
- Unidentified speaker: If you invest in a SaaS or an apps company is like a technical investment, it's like a business model innovation.
- 18:24
- Unidentified speaker: So because for us as a technical innovation, it has to sit kind of in the traditional technical stack.
- 18:29
- Unidentified speaker: So the traditional technical stack for infrastructure is compute, network, storage, and databases, right?
- 18:35
- Unidentified speaker: And each one of those has a subcomponent and they all have buyers.
- 18:38
- Unidentified speaker: We're seeing that being kind of reformed with AI a bit.
- 18:42
- Unidentified speaker: And so all of them are being disrupted, but also now you have models, which I would consider infrastructure.
- 18:47
- Unidentified speaker: They look to me more like ASICs than they look like traditional software.
- 18:51
- Unidentified speaker: I mean, like the capital needs, the competitive pressures on them.
- 18:55
- Unidentified speaker: And then I think they're disrupting the entire application stack above them.
- 18:59
- Unidentified speaker: Like we don't even know if they will layer.
- 19:01
- Unidentified speaker: I mean, maybe you could argue with JEV, you can kind of put them in software, but traditionally they're their own separate entity that you can't marry to software very well.
- 19:07
- Unidentified speaker: And so I would say we're in kind of a point of maximum disruption, but all of the infrastructure supporting software, whether it's vibe coded or not, or whether it's to build models or not, I mean, that doesn't change like the major components.
- 19:18
- Unidentified speaker: And so I think we do have a bit of a traditional mindset of compute network storage databases.
- 19:23
- Unidentified speaker: But we also throw up our hands and like, this is all changing so fast.
- 19:26
- Unidentified speaker: Because, for example, when you worried about dev tooling in the past, you're worried about ergonomics for a human.
- 19:31
- Unidentified speaker: You don't really worry about that.
- 19:32
- Unidentified speaker: And then actually a lot of the design decisions that we made as technologists were like, well, a human's going to be using this so it can't be too complicated.
- 19:38
- Unidentified speaker: And then you would trade off things like safety or you trade off things like reliability, but we don't really have to do that anymore because the models are doing it.
- 19:44
- Unidentified speaker: So I think this is kind of the biggest refresh of computer science infrastructure we've seen, maybe since the internet, maybe forever.
- 19:50
- Unidentified speaker: Would the shorthand for this be that each of those five portions of infrastructure are being refactored for agents?
- 19:56
- Unidentified speaker: Is that overly simplifying?
- 19:58
- Unidentified speaker: So I split it in two pieces.
- 20:00
- Unidentified speaker: People are vibe coding a lot of stuff.
- 20:02
- Unidentified speaker: There's more software being written ever before.
- 20:04
- Unidentified speaker: And that stuff has to run somewhere.
- 20:06
- Unidentified speaker: That's not agents, that's just normal software.
- 20:08
- Unidentified speaker: Maybe it's an agent that created it, maybe it's a human being that created it, but it needs databases and it needs compute.
- 20:13
- Unidentified speaker: So that's one thing.
- 20:14
- Unidentified speaker: It's like the product of agents.
- 20:16
- Unidentified speaker: And then the other one is like we're building these new systems that have AI at the heart of them.
- 20:20
- Unidentified speaker: And then how does that change it?
- 20:22
- Unidentified speaker: Those are the two.
- 20:22
- Unidentified speaker: So just assume way more software being built and has to be run and deployed somewhere.
- 20:26
- Unidentified speaker: And then think that now we also have systems that actually create agents.
- 20:30
- Unidentified speaker: It seems like each new wave of technology, there's a new kind of database that's needed.
- 20:36
- Unidentified speaker: Are you seeing that across all the areas that you care about that because of AI, there's whole new entrants?
- 20:42
- Unidentified speaker: Yeah, I think so.
- 20:43
- Unidentified speaker: I mean, I think a lot of infrastructure is getting a bit sleepy.
- 20:46
- Unidentified speaker: And now because agents are very sophisticated, that's no longer the case.
- 20:50
- Unidentified speaker: Like, so for example, let's take OLTP.
- 20:51
- Unidentified speaker: What is OLTP?
- 20:52
- Unidentified speaker: OLTP is a transaction database.
- 20:54
- Unidentified speaker: I think the oracles of the world.
- 20:55
- Unidentified speaker: That's a pretty mature industry.
- 20:57
- Unidentified speaker: You didn't see a lot of movement.
- 20:58
- Unidentified speaker: Maybe you saw like movement to the cloud or something like that.
- 21:00
- Unidentified speaker: But now we're seeing really interesting companies that are like, you know what?
- 21:05
- Unidentified speaker: The hard thing about using a database from a programmer's standpoint is databases have always had their own languages and they've had their own languages for technical reasons.
- 21:13
- Unidentified speaker: And it's really, really hard to have them work with normal programming languages.
- 21:18
- Unidentified speaker: And one of the reasons is if you change that language, like anybody that knows the databases wouldn't know how to use it.
- 21:23
- Unidentified speaker: And you'd have to like train a whole version of programmers to use this new language.
- 21:27
- Unidentified speaker: But now we're actually seeing databases come out, which are an entirely new way of thinking about databases, which have way better properties.
- 21:34
- Unidentified speaker: They don't really have to train anybody because the agents already know how to do it.
- 21:37
- Unidentified speaker: I think these things are path dependent.
- 21:39
- Unidentified speaker: Like some sectors, a new entrance is going to come and change it.
- 21:42
- Unidentified speaker: Like databases, OTP may look entirely different in five years.
- 21:46
- Unidentified speaker: And some, the incumbent may figure it out and be like, hey, listen, we're just going to add a few things.
- 21:50
- Unidentified speaker: Networking, of course, which is listen, I did my PhD in networking.
- 21:52
- Unidentified speaker: So it's an area of special interest.
- 21:54
- Unidentified speaker: Like that's been so sleepy and boring for the last 15 years.
- 21:57
- Unidentified speaker: And now, because these clusters that we're building are so large and they've got like such rigid performance constraints, we're seeing net new networking companies come out for the first time since Arista.
- 22:09
- Unidentified speaker: You're seeing changes in the materials used to build them.
- 22:12
- Unidentified speaker: You were seeing optical come up.
- 22:13
- Unidentified speaker: And so there's a lot of interesting stuff in networking.
- 22:15
- Unidentified speaker: Compute, the whole notion of what are the abstractions to run an agent on?
- 22:21
- Unidentified speaker: Is it like a container?
- 22:22
- Unidentified speaker: Is it a VM?
- 22:23
- Unidentified speaker: Is it a durable workflow?
- 22:24
- Unidentified speaker: Something like temporal that is changing.
- 22:27
- Unidentified speaker: But I do think we're at a point of maximum disruption where many points of the infrastructure stack can be rebuilt.
- 22:33
- Unidentified speaker: Can we spend a few minutes on models?
- 22:35
- Unidentified speaker: Dropic S1 came out this week, obviously, a beautiful view into the business itself.
- 22:39
- Unidentified speaker: But I'm actually more interested because the infrastructure stuff in what you think about just the state of these things today.
- 22:46
- Unidentified speaker: I'm interested in all the debates, closed versus open, the public frontier versus the private frontier that the companies keep for themselves, movement of talent, opportunities, things that don't exist that should, whatever's on your mind about these things and how you see them evolving.
- 23:00
- Unidentified speaker: I'm fascinated to hear.
- 23:01
- Unidentified speaker: So I think the biggest question is what happens if and when the market slows down?
- 23:07
- Unidentified speaker: Because I think we've got three years of data of what happens in the existing very big, fast growing market.
- 23:14
- Unidentified speaker: And like what happens there is being on the frontier means you've got tremendous pricing power.
- 23:19
- Unidentified speaker: Even if you're just an epsilon better than the next guy, you've got tremendous pricing power.
- 23:23
- Unidentified speaker: There's no demand problem.
- 23:24
- Unidentified speaker: You've got basically infinite demand.
- 23:26
- Unidentified speaker: And so these frontier models that are the best tend to get 80% plus of the market dollar weighted.
- 23:32
- Unidentified speaker: And that's happened for three years.
- 23:34
- Unidentified speaker: But you could argue the reason that's happening is because we've not saturated the market at all.
- 23:38
- Unidentified speaker: And people always gonna want the smartest thing.
- 23:41
- Unidentified speaker: And so then, under that view, there's nothing structural that's gonna change as long as the market's growing.
- 23:46
- Unidentified speaker: And you could argue this is gonna go on for quite a while because what is a penetration 0.1% or something like that?
- 23:51
- Unidentified speaker: They say 3% of the enterprise, but actually from budget-wise, it's like 1% or less, right?
- 23:55
- Unidentified speaker: And so there's just a lot to grow given how useful the technology is.
- 23:59
- Unidentified speaker: The more interesting question to me is what happens if the market starts to saturate in pockets?
- 24:04
- Unidentified speaker: So if that happens, things shift to open source?
- 24:07
- Unidentified speaker: Will you lose pricing power on the frontier?
- 24:10
- Unidentified speaker: One thing that we don't really do a good job of accounting for when we model out the numbers for these companies is what happens if you invest in a model and then you don't get the next model out really quickly?
- 24:22
- Unidentified speaker: Let's assume that Anthropic stumbles and they don't release a model for six months.
- 24:26
- Unidentified speaker: What is the lifespan do you think of the previous model?
- 24:29
- Unidentified speaker: A very short period of time.
- 24:30
- Unidentified speaker: It's very, very short, which is entirely different than we've ever had in software.
- 24:34
- Unidentified speaker: In software, we've always assumed some long tail.
- 24:36
- Unidentified speaker: But it turns out what happens if you put out the soda model and everybody loves it, then something internal happens.
- 24:42
- Unidentified speaker: I don't know if it's regulatory, I don't know if it's technical, I don't know if it's a breach, whatever.
- 24:46
- Unidentified speaker: And then you slip three months, right?
- 24:48
- Unidentified speaker: It could be the case that you lose the entire market.
- 24:50
- Unidentified speaker: In my experience, and I see a lot of these model graphs from apps companies and from pitches, I would say that the lifespan of a model is three to six months.
- 24:58
- Unidentified speaker: So that means you have to release the next model.
- 25:00
- Unidentified speaker: So again, for this conversation, I think it's a very interesting conversation.
- 25:02
- Unidentified speaker: We should tease it apart.
- 25:03
- Unidentified speaker: Like, what happens if the market continues to grow and doesn't saturate?
- 25:07
- Unidentified speaker: I think we already know because we have three years of data and it just means the frontier wins and they do a great job and they create great models.
- 25:12
- Unidentified speaker: Then what happens if either the use cases saturate, meaning you don't need more intelligence, or the market starts slowing down for another reason?
- 25:19
- Unidentified speaker: Nobody knows the answer to that.
- 25:20
- Unidentified speaker: I think there's a very real world where the value comes crashing back down into open source.
- 25:26
- Unidentified speaker: And we'll see a shift that we've never seen because software has always had a long tail of retention.
- 25:31
- Unidentified speaker: On this trip, I've been asking a lot of people what percent of software coding tasks are already saturated.
- 25:36
- Unidentified speaker: And the typical answer is very, very high percentage.
- 25:39
- Unidentified speaker: That's been the biggest market.
- 25:40
- Unidentified speaker: And yet it still seems like people are spending like crazy on frontier tokens to do coding, which it seems a little at odds.
- 25:48
- Unidentified speaker: Like if I could do it with a six-month-old model or something that's vastly cheaper.
- 25:52
- Unidentified speaker: Well, these are two different statements.
- 25:54
- Unidentified speaker: This is a great thing to dig in on, but we need to be very careful.
- 25:57
- Unidentified speaker: So statement number one is I am not using more tokens.
- 26:00
- Unidentified speaker: I have enough tokens.
- 26:01
- Unidentified speaker: I've automated 95% already, so I can't automate anywhere.
- 26:04
- Unidentified speaker: And I think that we're getting pretty close to that.
- 26:06
- Unidentified speaker: Comment number two is therefore I can use a cheaper model, does not follow from comment number one.
- 26:12
- Unidentified speaker: So if you and I are competing startups, we've got Martine startup, we've got Patrick startup, right?
- 26:16
- Unidentified speaker: So we've got competing startups.
- 26:18
- Unidentified speaker: And because we're in competitive equilibria, are you going to choose a dumber model to compete with me?
- 26:24
- Unidentified speaker: In competitive equilibria, you wouldn't, you would always pay 10 times more for the smartest model.
- 26:29
- Unidentified speaker: Yeah, always, always, always.
- 26:30
- Unidentified speaker: And so there's this very open question of will you saturate tasks so that you don't need a smarter model?
- 26:38
- Unidentified speaker: And I think it's very hard to do that in a Darwinistic economic equilibria when you're actually competing.
- 26:45
- Unidentified speaker: And some of them may, but I think at the end of the day, it's about can you get software out faster?
- 26:49
- Unidentified speaker: Can you make it better?
- 26:50
- Unidentified speaker: And can you compete in the marketplace?
- 26:51
- Unidentified speaker: And that will necessarily push you to the more intelligent thing.
- 26:54
- Unidentified speaker: Now, that said, clearly things will slow down because you're actually not spending more tokens.
- 27:00
- Unidentified speaker: And so either you have to increase prices or you're gonna have to go broad.
- 27:05
- Unidentified speaker: But in my experience, the pricing is maintained for sure, and because of the competitive equilibria, like you'll always want the smartest thing.
- 27:12
- Unidentified speaker: And then we're just seeing a broadening use of these outside of code, general knowledge work.
- 27:17
- Unidentified speaker: I'm a huge grok by user.
- 27:19
- Unidentified speaker: It's like my favorite thing ever.
- 27:20
- Unidentified speaker: And so I think we're starting to see, yeah, take over more and more of these.
- 27:24
- Unidentified speaker: Say more about your personal experience with that, what you're doing with it, what you're learning from that experience.
- 27:28
- Unidentified speaker: Like my new grok is so funny.
- 27:30
- Unidentified speaker: I know that home automation thing going on.
- 27:32
- Unidentified speaker: I know it's so good.
- 27:33
- Unidentified speaker: Like I just feel like I've waited my entire life for this power, this technology to show up, right?
- 27:39
- Unidentified speaker: Like I want Star Trek.
- 27:40
- Unidentified speaker: I want to go home and I want my house to greet me, and I want to talk to it, and I want to have the face show up on the monitor, and I want the whole thing.
- 27:48
- Unidentified speaker: We've never been able to do it, and now it's kind of trivial.
- 27:51
- Unidentified speaker: So I use GrokBot for a lot of stuff.
- 27:52
- Unidentified speaker: I use it for personal finances.
- 27:53
- Unidentified speaker: My big hobby is modular synthesis.
- 27:55
- Unidentified speaker: Like I like synthesizers like making music.
- 27:57
- Unidentified speaker: Oh, cool.
- 27:57
- Unidentified speaker: But like the 1970s, like patching, tweaking type stuff.
- 28:00
- Unidentified speaker: It helps me with a lot of that stuff.
- 28:02
- Unidentified speaker: So I use it for a lot of like standard things, but my new passion project is I want GrokBot to embody itself.
- 28:08
- Unidentified speaker: I say, I want you to be the living AI in my house.
- 28:10
- Unidentified speaker: That's what I want.
- 28:11
- Unidentified speaker: And I wanted to, just like Star Trek, help me make this happen.
- 28:14
- Unidentified speaker: And it's literally buy stuff.
- 28:16
- Unidentified speaker: It's buying stuff.
- 28:17
- Unidentified speaker: So I was like, oh, let's try this.
- 28:19
- Unidentified speaker: Yesterday I bought a few hundred dollars worth of gear that's coming to the house.
- 28:23
- Unidentified speaker: And like, I'm gonna have it embody itself so that what gear is it buying?
- 28:26
- Unidentified speaker: I don't even know.
- 28:29
- Unidentified speaker: It's like these weird, like these weird open source projects.
- 28:33
- Unidentified speaker: And like, I mean, it's like this really obscure stuff.
- 28:35
- Unidentified speaker: Like, I know some home automation stuff.
- 28:37
- Unidentified speaker: Like I know like Zigbee Life, I know some home, but it's like really obscure stuff that it's decided to buy.
- 28:42
- Unidentified speaker: But like I just have in my head this vision of that kind of vintage view of what the future looked like, like the 70s view, like you walk through the house.
- 28:49
- Unidentified speaker: Yeah, like I just want that.
- 28:51
- Unidentified speaker: I've basically just been feeding it pictures and descriptions, and I want it to create that by itself.
- 28:56
- Unidentified speaker: And so I'll let you know how that goes.
- 28:58
- Unidentified speaker: What is this and other experiences using this technology?
- 29:02
- Unidentified speaker: Since you're probably in the 0.0001% of creativity using it, what does it tell you about what's coming?
- 29:09
- Unidentified speaker: Yeah, this is it.
- 29:10
- Unidentified speaker: I just feel people keep thinking of these things as software and it's not.
- 29:13
- Unidentified speaker: And I think we just keep mapping it to things like search and it's not.
- 29:17
- Unidentified speaker: And we really have created general flexible technology that can bring us many of these future things that we've always dreamed about.
- 29:26
- Unidentified speaker: We just didn't have the technology to implement.
- 29:28
- Unidentified speaker: And I just feel like we're so constrained from our presuppositions of what software can do that people aren't doing as many of these things.
- 29:35
- Unidentified speaker: I'll give you another example.
- 29:36
- Unidentified speaker: You know, I worked with Faith and World Labs, great company.
- 29:39
- Unidentified speaker: And one thing I realized is part of that, and what they did is they created a model that would generate 3D scenes.
- 29:44
- Unidentified speaker: They released a model called Atlas a few months ago, and that would like generate basically any scene, but also supported dynamics so it would move.
- 29:52
- Unidentified speaker: And I realized we're this close to being able to step into a movie.
- 29:56
- Unidentified speaker: Imagine, like, Miyazaki probably wouldn't be happy with it, but imagine like having spirited away running, and you throw on your VR glass and you walk in.
- 30:02
- Unidentified speaker: You're there, and like, did you ever see Sleep No More?
- 30:05
- Unidentified speaker: Like the Of course, yeah.
- 30:06
- Unidentified speaker: I went to it.
- 30:07
- Unidentified speaker: Why isn't there a Sleep No More in VR that you're walking in, like the whole thing is going on?
- 30:12
- Unidentified speaker: And like, just for those that are listening, Sleep No More is an immersive play, which is three hours.
- 30:17
- Unidentified speaker: It was in this three-story hotel or something like that building in New York, and you're like running around trying to figure out and like Hamlet is being played out or whatever.
- 30:24
- Unidentified speaker: We're starting to realize how powerful this stuff is.
- 30:27
- Unidentified speaker: A lot of these tropes and visions of the future from the 50s and 60s, we can actually make.
- 30:33
- Unidentified speaker: And I just think that the biggest constraint is no longer the tech.
- 30:36
- Unidentified speaker: It's literally our creativity.
- 30:38
- Unidentified speaker: And it's just because we've been beaten down by 30 years of software being shit.
- 30:42
- Unidentified speaker: I mean, let's be honest, software has sucked, technology's been hard to use, old people hate it.
- 30:47
- Unidentified speaker: It's most often broken, and now we're actually in a point where this stuff could actually start working.
- 30:51
- Unidentified speaker: And so for me, it's like, I'm so excited for what this brings.
- 30:54
- Unidentified speaker: It's tempting to say you could just freeze the models and their current capability and still get all of what you described.
- 30:59
- Unidentified speaker: So there's like a diffusion question I'd like to explore.
- 31:02
- Unidentified speaker: Before we do that, what if anything do you think is missing from the models?
- 31:05
- Unidentified speaker: Like, how do you hope or wish or dream that the models themselves can do?
- 31:09
- Unidentified speaker: So I think I think if you step back and you look at like what we know as an industry, the current tech stack, anything that's digital with enough capital, you can solve.
- 31:19
- Unidentified speaker: And the reason is is because anything you can simulate on a computer, you can create RL environments for, whether that's 3D using Blender, whether that's general application use of a computer, whether it's code.
- 31:30
- Unidentified speaker: You and I have talked about this before.
- 31:31
- Unidentified speaker: I think when it's all said and done, the most interesting thing about AI, the meta thing, is that like you can take a money kid and then you can point it at a problem and it will solve that problem.
- 31:42
- Unidentified speaker: Like maybe it's expensive, but at least you can solve it.
- 31:43
- Unidentified speaker: In the past, you'd be like blocked in engineering and you die.
- 31:46
- Unidentified speaker: Now you can actually just direct money at problems and solve these problems.
- 31:49
- Unidentified speaker: So I feel very confident saying anything in the digital world, because you can create an environment and you can RL it and you can train on it, we can solve.
- 31:57
- Unidentified speaker: So I just feel very comfortable.
- 31:58
- Unidentified speaker: Anything you do with a computer, AI will be able to do.
- 32:00
- Unidentified speaker: So maybe it doesn't do it now, but like you can just see that path happening.
- 32:03
- Unidentified speaker: What I don't know is does the same thing apply to the physical universe?
- 32:08
- Unidentified speaker: Sure, you could set up a lab to like nick speakers or whatever, you know, to do that.
- 32:12
- Unidentified speaker: But like, are the scaling curves the same?
- 32:14
- Unidentified speaker: Is the distribution the same?
- 32:16
- Unidentified speaker: So anything to do with robotics and the physical world and science and medicine, I have no idea if it follows the same laws.
- 32:22
- Unidentified speaker: And so I think that's still an unlock.
- 32:24
- Unidentified speaker: But I feel very comfortable that the current digital side is done.
- 32:27
- Unidentified speaker: How have you processed all this interesting security stuff?
- 32:30
- Unidentified speaker: Breaches, I've heard it called breach pocalypse.
- 32:33
- Unidentified speaker: Uh it it's obvious that we're like entering a very interesting era in cyber.
- 32:37
- Unidentified speaker: What do you make of all this?
- 32:38
- Unidentified speaker: I lived through on the ground with formal studies and security, I lived through the internet.
- 32:44
- Unidentified speaker: And it was way worse than this.
- 32:45
- Unidentified speaker: Like by this time with the internet, we had worms that had cost billions of dollars of economic damage.
- 32:53
- Unidentified speaker: We'd taken out critical infrastructure.
- 32:56
- Unidentified speaker: Remember Windows 95?
- 32:57
- Unidentified speaker: You're probably too.
- 32:57
- Unidentified speaker: No, I remember.
- 32:58
- Unidentified speaker: Okay.
- 32:58
- Unidentified speaker: Remember Windows 95?
- 33:00
- Unidentified speaker: So it used to be that if you took Windows 95 and you installed it and it was connected to the internet, before the installer was done, it'd be infected.
- 33:10
- Unidentified speaker: Like then take it away.
- 33:11
- Unidentified speaker: Like, not a joke, right?
- 33:12
- Unidentified speaker: Like, so like you didn't even have time to patch it.
- 33:15
- Unidentified speaker: It was so chaotic and there was so much mayhem.
- 33:17
- Unidentified speaker: And this is what happens every time you've got a new technology that changes the nature of how we run computers in the security posture.
- 33:26
- Unidentified speaker: And we're definitely dealing with another one of those moments.
- 33:28
- Unidentified speaker: And so I think we can look to what happened with the internet, which is like there's a few years of mayhem.
- 33:34
- Unidentified speaker: You do have breaches that cause real issues, and then you know, the industry evolves.
- 33:37
- Unidentified speaker: And normally the way you think about security is like, will the good guys be able to spend more money than the bad guys?
- 33:43
- Unidentified speaker: If the answer is yes, which is typically the case, you will come up with a technical solution.
- 33:46
- Unidentified speaker: And I think that will be the case here.
- 33:48
- Unidentified speaker: And you're already starting to see it.
- 33:49
- Unidentified speaker: These vulnerabilities that you see showing up, this is good.
- 33:52
- Unidentified speaker: That means that somebody used the model to make things safer by pointing out the vulnerabilities.
- 33:57
- Unidentified speaker: I think this is going exactly like you would expect it to go.
- 34:00
- Unidentified speaker: But I think the net impact itself will be much more secure because AI is also securing it.
- 34:05
- Unidentified speaker: Can you talk about capital?
- 34:06
- Unidentified speaker: There's very interesting things happening in capital markets.
- 34:09
- Unidentified speaker: There's clearly lots of demand destruction happening.
- 34:11
- Unidentified speaker: We're not funding things that we would have in a different environment because the cost of capital to fund them is too high.
- 34:17
- Unidentified speaker: And the cost is rising because the returns are so high on anything AI.
- 34:20
- Unidentified speaker: So you've got these comp I was looking at this company this morning where if you and I had seen the deck five years ago, we'd be like, this is an amazing software business and we should fund it.
- 34:28
- Unidentified speaker: And now it seems to pale in comparison to other things that we could put our capital to.
- 34:33
- Unidentified speaker: Capital itself feels like this interesting variable in this whole equation as a constraint.
- 34:38
- Unidentified speaker: I'm curious how that feels to you.
- 34:40
- Unidentified speaker: I sit on the boards of a lot of great companies that have the problem that you have.
- 34:44
- Unidentified speaker: Profitable, growing 40, 50, 60% at scale, 300, 400, 500 million in ARR.
- 34:51
- Unidentified speaker: What do those companies do?
- 34:52
- Unidentified speaker: You can't really IPO because they're subscale, but they're great companies.
- 34:55
- Unidentified speaker: And I think about my own personal portfolio.
- 34:57
- Unidentified speaker: I'm like, whatever, I'm in Munibonds, man.
- 34:59
- Unidentified speaker: I'd rather be in these companies, right?
- 35:01
- Unidentified speaker: And so I think that everybody would be happy with an asset that grows 40%.
- 35:07
- Unidentified speaker: Like you would be happy with your personal money being in there if you could guarantee it.
- 35:11
- Unidentified speaker: And so I still think these are very good companies.
- 35:15
- Unidentified speaker: I feel it's more a question, again, this is my naivety of not understanding this of like, are we gonna see a shift in the market so that people can get exposure to this asset class?
- 35:25
- Unidentified speaker: Because I think the risk profile is very, very different than AI.
- 35:27
- Unidentified speaker: Nobody knows what's gonna happen to AI when the market slows down.
- 35:30
- Unidentified speaker: Nobody knows.
- 35:31
- Unidentified speaker: It could be easy come, easy go.
- 35:32
- Unidentified speaker: It could keep going forever.
- 35:33
- Unidentified speaker: Like nobody really knows.
- 35:34
- Unidentified speaker: And so the only free lunch I've ever seen in investing is diversification.
- 35:38
- Unidentified speaker: And I would love to personally have like a basket.
- 35:41
- Unidentified speaker: A basket that includes these slower growing, but these assets that don't disappear.
- 35:47
- Unidentified speaker: Are most of the ones though that are in your mind extremely subjected to this disruption risk?
- 35:52
- Unidentified speaker: There's lots of companies with that rough profile, which seem terrifying in the face of how easy it's going to be to create software.
- 35:59
- Unidentified speaker: I'm an infra guy.
- 36:00
- Unidentified speaker: So for me, more software means you need more infra.
- 36:03
- Unidentified speaker: We don't vibe code databases yet.
- 36:04
- Unidentified speaker: Like we do if you're a database.
- 36:06
- Unidentified speaker: Can Snowflake be a well-known example of kind of what you're talking about?
- 36:09
- Unidentified speaker: Yeah, it's a great example.
- 36:10
- Unidentified speaker: It's not just building it.
- 36:11
- Unidentified speaker: It's really hard to have the long tail of features, the long tail of performance tricks to actually operate these things as really, really tough.
- 36:17
- Unidentified speaker: And so let's say that Snowflake is growing 30 to 40%, and the economists are good.
- 36:22
- Unidentified speaker: Like I don't think that they're at any risk of being disrupted by AI, maybe by databricks, but that is like pre-AI type of thing.
- 36:30
- Unidentified speaker: And so again, I'm very naive when it comes to these things.
- 36:32
- Unidentified speaker: I'm a very simplistic thinker.
- 36:34
- Unidentified speaker: My view is like, I don't think AI is going to disrupt these.
- 36:37
- Unidentified speaker: I think they are good businesses.
- 36:38
- Unidentified speaker: I think more and more the industry is going to want to wrap these up and get exposure.
- 36:43
- Unidentified speaker: And I don't know what that looks like.
- 36:43
- Unidentified speaker: It's kind of interesting.
- 36:44
- Unidentified speaker: Like we're even seeing specs come back.
- 36:46
- Unidentified speaker: I get a lot of calls now for companies that sit on the board, which are good companies.
- 36:49
- Unidentified speaker: Were these specs are like, hey, listen, like we can make this stuff go public.
- 36:52
- Unidentified speaker: So I think that the financial markets have to figure that out.
- 36:55
- Unidentified speaker: There's a lot of companies that will be disrupted by AI, but there's a whole bunch that will not.
- 36:59
- Unidentified speaker: And I just think the market doesn't know what to do with them right now.
- 37:02
- Unidentified speaker: Most meetings that you have around here, at least for me in the last, let's say, three months, half the meeting is about compute and that the product across categories just needs more compute that is available to accomplish its goals.
- 37:15
- Unidentified speaker: So this is another way of asking about the capital thing.
- 37:17
- Unidentified speaker: You guys manage a ton of money.
- 37:19
- Unidentified speaker: Everyone's deploying their funds at very fast paces.
- 37:22
- Unidentified speaker: Each company that's working requires lots of capital.
- 37:25
- Unidentified speaker: I'm trying to get a sense of the feel of how much capital itself is the constraint that, like, if we just airdropped a trillion dollars more in all the funds, that would be a good thing.
- 37:37
- Unidentified speaker: I said this before, but I'm going to reiterate because it's so important.
- 37:39
- Unidentified speaker: I really think the big disruption of AI is the fact that you can throw capital at problems.
- 37:46
- Unidentified speaker: In the history of tech, you haven't been able to do that because it was always an engineering problem, which just doesn't scale.
- 37:51
- Unidentified speaker: The mythical man month is a mythical man month for the reason.
- 37:54
- Unidentified speaker: Ten years ago, if I had five founders show up and they ask for a billion dollars and I gave it to them, what would happen?
- 38:00
- Unidentified speaker: Sits there.
- 38:01
- Unidentified speaker: Or they would detonate, right?
- 38:02
- Unidentified speaker: They would like hire a bunch of engineers, and then you're really fucked.
- 38:05
- Unidentified speaker: You can't throw capital in it.
- 38:08
- Unidentified speaker: Can you explain the mythical man month?
- 38:09
- Unidentified speaker: That's a very interesting idea.
- 38:11
- Unidentified speaker: Yeah, so the mythical man month basically says that you can't increase the speed of engineering.
- 38:18
- Unidentified speaker: It's like a natural.
- 38:20
- Unidentified speaker: It's very sublinear.
- 38:22
- Unidentified speaker: And it has to come with organizational communication, which is one of the conclusions.
- 38:25
- Unidentified speaker: But the takeaway for those that are listening is that this comes from, I think, Fred Brooks, right?
- 38:30
- Unidentified speaker: And like the 70s, which is basically like you had these software projects and you wanted to get them out faster.
- 38:35
- Unidentified speaker: And so they would throw more people at, and then it would go slower.
- 38:38
- Unidentified speaker: Engineering has always been this bottleneck.
- 38:40
- Unidentified speaker: And in my experiences, and I had my own startup, I did it for 10 years, right?
- 38:44
- Unidentified speaker: It ended up being a very large business.
- 38:45
- Unidentified speaker: This is totally true.
- 38:47
- Unidentified speaker: You can't scale engineering.
- 38:49
- Unidentified speaker: And to me, the biggest shift, like if we step back, like industrial revolution level shift with AI is I can take 20 people and I can give them a billion dollars and they can use that.
- 39:00
- Unidentified speaker: Now, do they make 500 million from that billion or two billion?
- 39:03
- Unidentified speaker: I have no idea.
- 39:04
- Unidentified speaker: But the fact that you can productively use that money is so new and it's so interesting.
- 39:09
- Unidentified speaker: And so at some level, the capital markets need to figure this out because they are taking this money and using it.
- 39:16
- Unidentified speaker: And if you look at Anthropic and OpenAI, they've probably raised more money in aggregate than the entire downstream ecosystem combined by a factor of whatever.
- 39:23
- Unidentified speaker: They are these mechanisms of turning money into growth.
- 39:27
- Unidentified speaker: So that's one side.
- 39:28
- Unidentified speaker: But there's another side, which is the debt side, which is quite a bit different than the equity side, which is I do think that currently, I'm not a growth investor, not a public markets guy, I'm not a debt guy.
- 39:37
- Unidentified speaker: But here's what I'm seeing on the ground.
- 39:39
- Unidentified speaker: We're running out of available debt.
- 39:41
- Unidentified speaker: And if you're the Norway pension fund, are you gonna give debt to a random NeoCloud or buy T-bills?
- 39:48
- Unidentified speaker: And I think that we're just at a point where you don't have enough money to go around.
- 39:52
- Unidentified speaker: And so it's gonna prefer the investment grade companies.
- 39:55
- Unidentified speaker: And then those that are not investment grade will have a harder time.
- 39:59
- Unidentified speaker: So scale wins again.
- 40:00
- Unidentified speaker: Scale will win, and then it'll be a harder time raising debts.
- 40:03
- Unidentified speaker: If your business model requires you to raise debt, there's just not enough debt to go out.
- 40:07
- Unidentified speaker: So I know you're talking about capitalism, like general capital, you're gonna see a ton of liquidity.
- 40:11
- Unidentified speaker: I think that that feels fine to me.
- 40:13
- Unidentified speaker: The debt side, I don't know these markets, but I definitely know it's harder to get debt now.
- 40:18
- Unidentified speaker: And I think what this will mean, if we want to have smaller players around that are not IG that get debt, they're gonna need to prepay, which means their customers are gonna need to prepay.
- 40:27
- Unidentified speaker: So I think over time this all kind of flows back to the availability of cash and capital.
- 40:31
- Unidentified speaker: Then you do have limited amounts of debt.
- 40:33
- Unidentified speaker: This reminds me to ask about Neo Labs, Neo Clouds, inference providers and companies.
- 40:38
- Unidentified speaker: I don't know if you call these infrastructure or what part of the stack you can do.
- 40:41
- Unidentified speaker: Yeah, yeah.
- 40:41
- Unidentified speaker: I've I'll view those all as infrastructure.
- 40:43
- Unidentified speaker: So if you take those categories in turn, there's lots of them.
- 40:46
- Unidentified speaker: Lots of them have been funded.
- 40:48
- Unidentified speaker: Very few of them are at scale.
- 40:50
- Unidentified speaker: I'm curious how you think this plays out, like what roles they should and will play, what the market feels like.
- 40:57
- Unidentified speaker: These things are growing so unbelievably fast in most cases.
- 41:01
- Unidentified speaker: I'm very interested by these companies and I'm curious what you think of these categories.
- 41:04
- Unidentified speaker: Aaron Powell Well, let's put neo labs aside, because I think that's the least well-defined of all of them.
- 41:08
- Unidentified speaker: I included them to be clear because so many of them seem to be building their own.
- 41:13
- Unidentified speaker: They're becoming neoclassic.
- 41:14
- Unidentified speaker: Well, like every pitch, so many pitches turn into some version of like we're building a token factory with some degree of vertical integration.
- 41:21
- Unidentified speaker: And some neolabs are going all the way down to like buying land and power and all the way up through the token.
- 41:26
- Unidentified speaker: You know what I mean?
- 41:27
- Unidentified speaker: So there's this weird like, yeah, I think so.
- 41:29
- Unidentified speaker: In my world, that wouldn't be considered a neol lab.
- 41:31
- Unidentified speaker: That would be it.
- 41:31
- Unidentified speaker: Like all neo clouds want to be token factories.
- 41:33
- Unidentified speaker: So I would view if like your output is like a token of an existing model architect.
- 41:38
- Unidentified speaker: Kimmy Patriot or something, yeah.
- 41:39
- Unidentified speaker: Yeah, yeah, neo cloud.
- 41:40
- Unidentified speaker: Like all of them want to do that and they will all verticalize.
- 41:43
- Unidentified speaker: The actual real limitation right now is hardware supply.
- 41:45
- Unidentified speaker: That's the it's not capital, it's not debt.
- 41:47
- Unidentified speaker: That's the actual real limitation and power.
- 41:50
- Unidentified speaker: So Neo Cloud to me is one that actually provides the physical infrastructure that ends at tokens.
- 41:54
- Unidentified speaker: And so they can provide GPUs, they can provide fine-tuning services, or they can provide tokens.
- 42:00
- Unidentified speaker: But it's like think of like a data center that actually has the GPUs in it.
- 42:03
- Unidentified speaker: That's what a Neo Cloud is to me.
- 42:04
- Unidentified speaker: A Neo Lab to me is give me a billion dollars, we're gonna come up with a new model architecture.
- 42:10
- Unidentified speaker: I love a lot of these business plans, are like if we solve ASI, that's our business model.
- 42:14
- Unidentified speaker: Like, literally, like that.
- 42:15
- Unidentified speaker: And by the way, to be fair, that's what OpenAI's original thesis was, and to some extent, anthropics, which is like if you solve the intelligence problem, the business model part is easy.
- 42:24
- Unidentified speaker: So that's to me a new lab is new architecture, model training, and sometimes they'll buy their own infrastructure and buy their own GPUs, but like that's not what they're building towards or selling.
- 42:33
- Unidentified speaker: The real constraint right now is power, it is chips, it is materials.
- 42:38
- Unidentified speaker: With that sort of a limitation, 80% of these things are gonna go to the scale providers.
- 42:44
- Unidentified speaker: 20% are just gonna be kind of hard-to-reach places, niche type places.
- 42:49
- Unidentified speaker: Like I have the relationship with a random government that has the power.
- 42:52
- Unidentified speaker: You know, it's kind of funny.
- 42:53
- Unidentified speaker: I've met this company, uh GPU marketplace.
- 42:56
- Unidentified speaker: That's what it does, but it's like the most low-level version of it.
- 42:59
- Unidentified speaker: And like many of their GPUs converted chicken sheds, you know, like like chicken coops where like people in farms are like finding ways to provide it.
- 43:08
- Unidentified speaker: So the market is so large, there will always be a significant portion that it just doesn't make sense for the SpaceXs to do.
- 43:13
- Unidentified speaker: Like, SpaceX is not gonna go and find out how to create a marketplace of chicken coop GPUs.
- 43:18
- Unidentified speaker: I think this is what ended up happening is the 20% is large enough to have a very wide, fragmented array of folks, but then the 80% of the market are gonna be these homogeneous things like SpaceX.
- 43:28
- Unidentified speaker: I heard this story recently that someone asked Larry Page how much of compute will be orbital.
- 43:33
- Unidentified speaker: And he said five nines.
- 43:34
- Unidentified speaker: And the reason was it's just physics.
- 43:36
- Unidentified speaker: Eventually it's all going to be in space.
- 43:38
- Unidentified speaker: The actual space thing actually pencils out, but the reason it makes the most sense is just regulatory.
- 43:43
- Unidentified speaker: Just because it skirts it?
- 43:44
- Unidentified speaker: Yeah.
- 43:44
- Unidentified speaker: There's 500 proposed legislations to stop data center building in the United States alone, including Texas, right?
- 43:49
- Unidentified speaker: Like if you can't build a data center in Texas, I don't know what we're doing anymore.
- 43:53
- Unidentified speaker: If that regulatory pressure continues and you actually can't break ground, and I've heard from many of the top data center people, like it's very tough to break ground right now, then either you go international or you go into space.
- 44:04
- Unidentified speaker: Your comment on these software companies growing 60% reminds me to ask about prevailing valuations in the market.
- 44:11
- Unidentified speaker: And as a productive investor that's focused on disruptive assets being created, even if the disruptive asset is truly incredibly valuable and great, every one of those has a price that will ruin the return of investing in it if it's too high.
- 44:25
- Unidentified speaker: So this is another one of my bugaboos.
- 44:27
- Unidentified speaker: I don't know how you can be a VC and not believe tech creates value, which is the first conversation we had.
- 44:31
- Unidentified speaker: Now I'm gonna have a second conversation, which is I don't know how you can be a VC and not believe markets are efficient.
- 44:35
- Unidentified speaker: What are we doing?
- 44:36
- Unidentified speaker: Like we're capitalists, you have to believe markets are efficient.
- 44:39
- Unidentified speaker: I think private markets are very efficient, and private markets figure out the price.
- 44:43
- Unidentified speaker: And so I just don't try and bet against the market.
- 44:46
- Unidentified speaker: I think a lot of people think that private markets aren't efficient, and so you can find the deal and find the non-consensus deal.
- 44:52
- Unidentified speaker: And all of those deals were incredibly competitive.
- 44:55
- Unidentified speaker: And I didn't win by looking for a deal or whatever.
- 44:59
- Unidentified speaker: I pay the market price and maybe I'll get a little discount for negotiating or whatever, but I never really worry about that.
- 45:04
- Unidentified speaker: And so as a productive investor, I just trust that the markets are actually pretty smart and I trust that the markets will value it given risk adjusted.
- 45:12
- Unidentified speaker: And I believe that my diligence reduces that risk for me.
- 45:16
- Unidentified speaker: I don't do it on market value.
- 45:18
- Unidentified speaker: I'm not trying to find a better price.
- 45:19
- Unidentified speaker: I'm just trying to find a less riskier asset.
- 45:22
- Unidentified speaker: Doesn't it follow, though, that then the approach would be to actually try to index the private markets?
- 45:26
- Unidentified speaker: Like you just try to own literally the market weights of.
- 45:29
- Unidentified speaker: Oh, I would love to, but you can't.
- 45:31
- Unidentified speaker: I would love to, of course.
- 45:32
- Unidentified speaker: Yeah.
- 45:32
- Unidentified speaker: So this is right.
- 45:32
- Unidentified speaker: So this is where this all lends to.
- 45:34
- Unidentified speaker: And this is kind of how we think about investing.
- 45:36
- Unidentified speaker: Of course, if I can have an index of every competitive deal, like literally, my my dream would be I get five percent.
- 45:42
- Unidentified speaker: You've owned cognition when you buy cursor.
- 45:44
- Unidentified speaker: I have every I've gotten five percent of every competitive.
- 45:48
- Unidentified speaker: Like I really believe private markets are super efficient.
- 45:50
- Unidentified speaker: I do.
- 45:51
- Unidentified speaker: And I think that the consensus is actually smart because I'm a capitalist and I believe in markets.
- 45:55
- Unidentified speaker: And so if I could have 5% of every competitive deal where you've got legit investors in it, I'd be happy.
- 46:00
- Unidentified speaker: But you simply cannot do that.
- 46:01
- Unidentified speaker: You can't do that for competitive reasons.
- 46:03
- Unidentified speaker: You can't do that for capacity reasons.
- 46:05
- Unidentified speaker: So as an investor, you have to come up with how do I not run into the competitive bus hub, but even more importantly, how do I not run out of our ability to keep track and help these companies?
- 46:15
- Unidentified speaker: Because I do take the boards, I do get very involved, very involved.
- 46:19
- Unidentified speaker: That's just impractical.
- 46:20
- Unidentified speaker: So the way that we think about it, which you've heard before, but I'll restate it, is for every legitimate sector, and by that legitimate sector, we define it as there's good founders in it because we don't second guess founders.
- 46:31
- Unidentified speaker: If they're risking their futures and the opportunity costs and their family's wealth, then it's a good sector.
- 46:36
- Unidentified speaker: We do the work to try and decide for every good sector, which is the top company.
- 46:41
- Unidentified speaker: And that means you look at the team, you look at the tech, and you look at the monetization potential, you look at the market growth potential, you look at all of that stuff, and then we choose the best one.
- 46:49
- Unidentified speaker: And then we assume that the market is efficient.
- 46:50
- Unidentified speaker: That's kind of the approach.
- 46:51
- Unidentified speaker: So it's kind of an index, but we're actually segmenting it by market.
- 46:55
- Unidentified speaker: Sometimes we get it right and sometimes we get it wrong, but that way you're indexing the interesting markets.
- 47:00
- Unidentified speaker: Yeah, there's this thing in indexing called stratified sampling.
- 47:02
- Unidentified speaker: It's from my old quantitative base.
- 47:04
- Unidentified speaker: You can buy a small subset of the index and have your tracking error to the index be very low, because actually, literally, what you're describing is effectively stratified sampling.
- 47:13
- Unidentified speaker: Get exposure to the best, biggest, whatever per category, and that will perform similarly to the I think venture capital early on, and I wasn't there, I just read the books.
- 47:25
- Unidentified speaker: I think it believed that tech created value.
- 47:27
- Unidentified speaker: I mean, often there was no business to evaluate.
- 47:30
- Unidentified speaker: And I just don't see so much hand-wringing about bubbles or this or that or the other thing.
- 47:34
- Unidentified speaker: The market price is the market price.
- 47:36
- Unidentified speaker: To me, there's always this huge dissonance when I'm talking with other investors.
- 47:40
- Unidentified speaker: They're always decrying the bubbles, which nobody could ever predict.
- 47:43
- Unidentified speaker: By definition, if everybody's calling it a bubble, it's probably not a bubble.
- 47:46
- Unidentified speaker: Like it's almost a bubble like when everybody's investing because they think there's still upsides.
- 47:51
- Unidentified speaker: And then they don't somehow think private markets are efficient.
- 47:53
- Unidentified speaker: Remember, I got VC canceled because I'm like, non-consistent investing is dangerous.
- 47:58
- Unidentified speaker: I once on Twitter was like, hey, listen, I think it's dangerous to invest against consensus because basically there's follow-on capital risk.
- 48:05
- Unidentified speaker: And from saying that, many prominent VCs came out and they were like, oh, like all of these deals that we've done were non-consensus.
- 48:12
- Unidentified speaker: And I think there's something very tied to the ego and the identity of investors that they have to believe that they can truffle hunt and find the non-consensus thing.
- 48:20
- Unidentified speaker: They're smarter than everybody else.
- 48:21
- Unidentified speaker: But I think the reality is certainly if you look at our portfolio, the ones that return money were actually very, very competitive in general.
- 48:28
- Unidentified speaker: I just think these markets are actually quite efficient.
- 48:30
- Unidentified speaker: I love the idea of market efficiency and private markets being one area that you disagree with lots of well-known markets.
- 48:36
- Unidentified speaker: Apparently a lot.
- 48:37
- Unidentified speaker: Apparently, all of them.
- 48:38
- Unidentified speaker: Are there any other areas like that where you feel especially alone?
- 48:41
- Unidentified speaker: Yes.
- 48:42
- Unidentified speaker: There's another one that drives me nuts.
- 48:43
- Unidentified speaker: This is my last one, I promise.
- 48:45
- Unidentified speaker: Another thing that drives me nuts is this belief that if you put money in private markets, it doesn't somehow grow the tablet private markets.
- 48:52
- Unidentified speaker: I love Roloff.
- 48:53
- Unidentified speaker: I think he's fantastic.
- 48:54
- Unidentified speaker: He's an incredibly intelligent guy.
- 48:56
- Unidentified speaker: He's a great board member.
- 48:57
- Unidentified speaker: But he gave this talk.
- 48:58
- Unidentified speaker: Do you remember at TechCrunch Disruptory?
- 49:01
- Unidentified speaker: Many investors believe this.
- 49:02
- Unidentified speaker: So this isn't to call off Roloff, because I was actually a huge fan of his, but he was like, venture capital is not an asset class.
- 49:08
- Unidentified speaker: There's only 20 startups that matter, and that's it.
- 49:10
- Unidentified speaker: If there's too much money going in, you've overfunded private markets.
- 49:14
- Unidentified speaker: And to me, this is ludicrous.
- 49:16
- Unidentified speaker: Capital creates opportunities and markets and value.
- 49:19
- Unidentified speaker: And even though there is so much more money that has gone into the private markets in the last 20 years, the value in the private markets has kept pace, right?
- 49:29
- Unidentified speaker: You've gotten massive private companies, tons of opportunities, very large exits.
- 49:33
- Unidentified speaker: And so this zero-sum thinking about private markets is so bizarre.
- 49:38
- Unidentified speaker: Like we're venture investors.
- 49:40
- Unidentified speaker: We create value, like tech creates value.
- 49:42
- Unidentified speaker: If you put more money in it, it'll grow.
- 49:45
- Unidentified speaker: And you will be surprised, and maybe you won't be surprised, but I've been very surprised at how common this mindset is.
- 49:50
- Unidentified speaker: And very often I'll talk to a pretty senior GP who's been investing for 20 years.
- 49:54
- Unidentified speaker: And they'll be like, oh, there's too much money.
- 49:55
- Unidentified speaker: I'm like, what are you talking about?
- 49:56
- Unidentified speaker: There should be more money in there.
- 49:58
- Unidentified speaker: We need more ecosystem, more value, more investment.
- 50:01
- Unidentified speaker: So I do believe money in private markets creates TAM.
- 50:04
- Unidentified speaker: It creates the investable TAM.
- 50:06
- Unidentified speaker: I think it creates value.
- 50:07
- Unidentified speaker: And there's more opportunity to deploy money now in private markets than ever ever has before.
- 50:12
- Unidentified speaker: Even very simple examples of that are like companies not going public.
- 50:17
- Unidentified speaker: Like if there's more private money available, why would you even go public?
- 50:20
- Unidentified speaker: Again, we need to all step away from zero-sum thinking.
- 50:23
- Unidentified speaker: I think we all need to believe that technology creates value.
- 50:26
- Unidentified speaker: Now, especially with AI, we need to believe that capital enables technology to create value.
- 50:29
- Unidentified speaker: And if that's the case, more money in private markets, we should deploy the money and we should create the value.
- 50:37
- Unidentified speaker: Your finance team isn't losing money on big mistakes.
- 50:39
- Unidentified speaker: It's leaking through a thousand tiny decisions.
- 50:42
- Unidentified speaker: Nobody's watching.
- 50:43
- Unidentified speaker: Ramp puts guardrails on spending before it happens.
- 50:46
- Unidentified speaker: Real-time limits, automatic rules, zero firefighting.
- 50:50
- Unidentified speaker: Try it at ramp.com slash invest.
- 50:53
- Unidentified speaker: As your business grows, Vanta scales with you, automating compliance and giving you a single source of truth for security and risk.
- 51:00
- Unidentified speaker: Learn more at vanta.com slash invest.
- 51:03
- Unidentified speaker: Ridgeline offers one unified platform that automates away the complexity across portfolio accounting, reconciliation, reporting, trading, compliance, and more, all at scale.
- 51:12
- Unidentified speaker: Schedule a demo at ridgeline.ai.
- 51:15
- Unidentified speaker: Every investment firm is unique, and generic AI chatbots don't understand your process.
- 51:21
- Unidentified speaker: Rogo is built by finance professionals.
- 51:23
- Unidentified speaker: It's connected to your data, learns your process, and quickly produces client-ready outputs.
- 51:28
- Unidentified speaker: Check them out at rogo.com slash invest.
- 51:32
- Unidentified speaker: The best AI and software companies from open AI to cursor to perplexity use workOS to become enterprise ready overnight, not in months.
- 51:39
- Unidentified speaker: Visit workos.com to skip the unglamorous infrastructure work and focus on your product.
- 51:45
- Unidentified speaker: You've been on in MBA GM terms, you're like the player that's like on fire, can like jump really high between cursor and Fay Fey's company and open router, you know, all of these in a very short period of time.
- 51:55
- Unidentified speaker: Well done.
- 51:56
- Unidentified speaker: But I'm more curious about lessons you learned about MA and what's going on in this market, having been a part of maybe the three biggest major scale MAs in the last couple months.
- 52:06
- Unidentified speaker: So actually, I I kind of have a history with strategic MA going back to my company.
- 52:10
- Unidentified speaker: So when I sold my company to VMware, it was a $1.26 billion exit, which at the time was the highest.
- 52:17
- Unidentified speaker: I think I beat David Sachs's Yammer.
- 52:20
- Unidentified speaker: It was the highest acquisitions on.
- 52:21
- Unidentified speaker: Just counting, yeah.
- 52:25
- Unidentified speaker: It was kind of funny because Catalyst, Catalyst showed me this chart.
- 52:28
- Unidentified speaker: It was just multiples of revenue, like the total exit.
- 52:30
- Unidentified speaker: And there's like two outliers, and it was like yammering company.
- 52:33
- Unidentified speaker: And they're like really close to each other.
- 52:34
- Unidentified speaker: I was like marginally better.
- 52:36
- Unidentified speaker: Just a small bit.
- 52:38
- Unidentified speaker: At the time, I think that was the highest acquisition of multiples of revenue enterprise software.
- 52:42
- Unidentified speaker: And then actually, I was part of a deal that beat that, which was the tabular acquisition to Databricks.
- 52:47
- Unidentified speaker: And then, of course, Kershaw, I think it was the largest price fit acquisition ever.
- 52:50
- Unidentified speaker: And then, of course, you know, Faye Fey's acquisition is a very large acquisition.
- 52:53
- Unidentified speaker: Have this been a part of a lot of these strategic MA.
- 52:56
- Unidentified speaker: And so I think my comment only goes to strategic MA.
- 52:59
- Unidentified speaker: I've noticed a few trends.
- 53:01
- Unidentified speaker: One of them is there's almost always deep personal alignment because it's a real risk for both sides.
- 53:08
- Unidentified speaker: I think like Lisa Sue and Fay Fey, or think Michael Trule and Elon, or think like Alex and Patrick.
- 53:13
- Unidentified speaker: I mean, you now almost always have people that have the same visions of the future and they think about it the same way, whether it's you think in terms of markets or you think in terms of infrastructure or you think in terms of data, that's almost always present.
- 53:27
- Unidentified speaker: There's some existing alignment.
- 53:29
- Unidentified speaker: The second one is you have to take it from the mind of the buyer.
- 53:32
- Unidentified speaker: So what the buyer is doing is they're basically buying an option on a large market.
- 53:37
- Unidentified speaker: So in the case of VMware buying the Sierra, they're buying a $1 billion option on a $40 billion market.
- 53:43
- Unidentified speaker: And so, in order for that to make sense, you kind of have to solve the inductive proof for them.
- 53:49
- Unidentified speaker: You have to show them the base case, which is we have this business that sells in the networking and we can get that budget.
- 53:54
- Unidentified speaker: And then you have to do the inductive step.
- 53:56
- Unidentified speaker: And you guys have virtual machines.
- 53:59
- Unidentified speaker: So if we attach this to a virtual machine, we can sell more.
- 54:03
- Unidentified speaker: And then therefore, for part of you, we can sell a whole bunch of these things.
- 54:05
- Unidentified speaker: So you almost, because they're buying a future option on a market, the company's job is to show the inductive step.
- 54:10
- Unidentified speaker: The company being acquired, or both of them, the the discussion clearing has to Yeah, the discussion for the MA has to be that you have the base case, you have the inductive step, therefore you can get this to this market, which is like the large markets that happens.
- 54:23
- Unidentified speaker: And that conversation, in my experience, is one kind of only founders can do, and maybe investors that really understand strategically what is going on.
- 54:34
- Unidentified speaker: But those conversations are almost like really high-stakes enterprise sales.
- 54:38
- Unidentified speaker: I've been a part of a lot of more auction-y type MA.
- 54:41
- Unidentified speaker: You never have those conversations.
- 54:43
- Unidentified speaker: You're looking at the financial metrics and the spreadsheets and can they buying an asset, et cetera.
- 54:47
- Unidentified speaker: And they've got a very different flavor.
- 54:49
- Unidentified speaker: But I think that, especially with the timelines going down on tech, I think we're going to see more and more strategic exits.
- 54:56
- Unidentified speaker: And so I think it's probably important for people to get pretty smart of how these things look.
- 55:00
- Unidentified speaker: And it's very different than they assume.
- 55:01
- Unidentified speaker: And most people assume you should hire a banker and you should have an auction and get the highest price.
- 55:06
- Unidentified speaker: Hiring a banker is fine.
- 55:07
- Unidentified speaker: You can do that if you want.
- 55:08
- Unidentified speaker: But the idea that you're going to go to the apex predator, someone like whatever, Mark Zuckerberg, and play poker, like you're kidding yourself, right?
- 55:15
- Unidentified speaker: Like these things are built on trust and telling out a story, not playing games or playing poker.
- 55:20
- Unidentified speaker: And so I just think that there's a mindset that you need to have if you're going to engage in these things that's quite different than the intuition.
- 55:26
- Unidentified speaker: So can you tell from your perspective the story of, let's say, the cursor, since that's the biggest one, of that acquisition, since you're such an involved investor?
- 55:32
- Unidentified speaker: Give us the story.
- 55:34
- Unidentified speaker: We're just about ready to get a raise done.
- 55:36
- Unidentified speaker: We'll raise a few billion dollars that was getting done, but this is about the time that Mythos came out.
- 55:41
- Unidentified speaker: And if you worked backwards, you realize it probably costs, I don't know how much it costs, but it probably costs, say, $3 billion to make Mythos.
- 55:48
- Unidentified speaker: And so we're like, if Cursor is going to build models, we need to raise more money and get access to those compute.
- 55:54
- Unidentified speaker: And so then, of course, that kicks off strategic discussions.
- 55:56
- Unidentified speaker: Like there's not a lot of people that have access to that much compute.
- 55:58
- Unidentified speaker: And so then you have those conversations.
- 56:00
- Unidentified speaker: And then, of course, there's quite a bit of interest in the company as a result, just because the company had by far the most users and by far the most data.
- 56:07
- Unidentified speaker: Again, this is one of these great examples where in the end, SpaceX had the compute and cursor did have the data, and it had a team that had very successfully built models and built this practice and code was very important.
- 56:19
- Unidentified speaker: So there was just a lot of alignment.
- 56:21
- Unidentified speaker: And it always helps to have a raise because that helps actually set the value and set the price of these things.
- 56:26
- Unidentified speaker: So that was actually a fairly straightforward deal.
- 56:29
- Unidentified speaker: I'm curious if there's anything similar, lesson-wise, from OpenRouter or from Faith Face Company World Labs.
- 56:36
- Unidentified speaker: Anything else that is shareable from those two experiences that might bear some lessons from this kind of strategic MA?
- 56:43
- Unidentified speaker: So I'm going to go back to what I said, which is for these strategic MA, the most important thing is the intuition, especially of a lot of investors, is there's a deal going down.
- 56:51
- Unidentified speaker: We need to play games, we need to leak it to the press.
- 56:55
- Unidentified speaker: Drives me fucking nuts.
- 56:56
- Unidentified speaker: We need to get other companies in.
- 56:59
- Unidentified speaker: And you don't realize, like you're literally dealing with the most sophisticated people on the planet.
- 57:03
- Unidentified speaker: I just don't think you walk up to Zuckerberg and you're like, oh, like I'm going to beat you in poker, dude.
- 57:07
- Unidentified speaker: That guy'll fucking take out your sucker.
- 57:08
- Unidentified speaker: Every one of these deals, and these are world record deals.
- 57:12
- Unidentified speaker: The companies in very good faith engage with the buyers.
- 57:16
- Unidentified speaker: They didn't play any games.
- 57:18
- Unidentified speaker: And they knew that they were worth a certain amount of money because they had done raises or there were market comps and there was a lot of trust.
- 57:27
- Unidentified speaker: Having now done so many of these deals, I've just found that that approach for strategic MA is the right approach.
- 57:35
- Unidentified speaker: And I've been part of deals that are that size that have fallen apart because somehow a board member got into the head of the founder that, oh, you need to get competitors, you got to leak into the price.
- 57:45
- Unidentified speaker: I mean, God, the open router deal drove me nuts.
- 57:47
- Unidentified speaker: Like I got leaked.
- 57:48
- Unidentified speaker: Did you notice, by the way, that world labs did not get leaked?
- 57:50
- Unidentified speaker: Yeah, interesting.
- 57:51
- Unidentified speaker: Didn't get leaked.
- 57:52
- Unidentified speaker: Because we didn't tell anybody.
- 57:52
- Unidentified speaker: It was like me, you know, it was a few people.
- 57:55
- Unidentified speaker: Open router was leaking constantly, right?
- 57:57
- Unidentified speaker: And I think a number of VCs think, oh, this is how you drive up the price, or this is how you get it done.
- 58:02
- Unidentified speaker: You see these things leak all the time.
- 58:03
- Unidentified speaker: That destabilizes deals.
- 58:05
- Unidentified speaker: Again, I think the most important thing is a structural thing, like your intuition as an investor in particular, that this is like some commodity asset and it's about playing poker.
- 58:14
- Unidentified speaker: It's just wrong.
- 58:15
- Unidentified speaker: If it's strategic deal for a lot of money, I think you need to view it very professionally.
- 58:20
- Unidentified speaker: I think you need to develop a trust relationship.
- 58:22
- Unidentified speaker: I think you need to do it in good faith.
- 58:24
- Unidentified speaker: You should not leak these deals and realize that the people that are counterparties to this are the most sophisticated people in the industry.
- 58:30
- Unidentified speaker: The founders in every one of these cases treated it that way.
- 58:33
- Unidentified speaker: Michael Trul treated it that way in the team.
- 58:35
- Unidentified speaker: Alex and Chris treated that way for Open Router.
- 58:37
- Unidentified speaker: And for sure, Faye Fey treated it.
- 58:39
- Unidentified speaker: How are you seeing the role of investors evolve?
- 58:41
- Unidentified speaker: Both literally how it's evolving across these many, many boards that you're on, but also some commentary on if you think investors generally of your type are doing a good job or actually helping companies.
- 58:53
- Unidentified speaker: It seems like because the stakes are so high, growth is so big, the capital dollar amounts are so big that all these investors are overseeing, it's kind of crazy.
- 59:01
- Unidentified speaker: Some commentary on like the role of the investor.
- 59:03
- Unidentified speaker: One thing I will say is venture is not what it used to be.
- 59:06
- Unidentified speaker: The startup ecosystem is not what it used to be.
- 59:08
- Unidentified speaker: I mean, if you look at the largest private companies right now, they would rival any public companies 20 years ago, right?
- 59:14
- Unidentified speaker: Things like Stripe or Databricks or like SpaceX before it IPO or Anderil or these companies are massive.
- 59:20
- Unidentified speaker: And so one of the problems is what it means to be a startup has shifted so much that now you have this undefined quagmire of who should probably be public investors investing in early stage startups.
- 59:36
- Unidentified speaker: And I just don't think those intuitions are right.
- 59:38
- Unidentified speaker: And so I think that's kind of the root of the confusion.
- 59:41
- Unidentified speaker: I think maybe it just actually makes sense to start delineating a bit more on the private side.
- 59:47
- Unidentified speaker: Or if you have later stage people investing in earlier stage companies, maybe a little bit more self-awareness.
- 59:54
- Unidentified speaker: Like, for example, you wouldn't put me on a mining company board.
- 59:57
- Unidentified speaker: So remember 2021, we'd have these PE investors and they would invest in these Series B companies.
- 1:00:02
- Unidentified speaker: I would sit on the board with these people.
- 1:00:04
- Unidentified speaker: And it's like, it's literally like putting me on the board of a mining company.
- 1:00:08
- Unidentified speaker: No idea.
- 1:00:08
- Unidentified speaker: Like trying to like reason about the company through like burn.
- 1:00:12
- Unidentified speaker: Like, dude, you're gonna be miserable here because all we do is burn cash for a long time.
- 1:00:18
- Unidentified speaker: It's more kind of the blending of the lines because private companies are a real asset class for late-stage investors.
- 1:00:27
- Unidentified speaker: A lot of this is just growing pains.
- 1:00:28
- Unidentified speaker: Like, how much larger do you think the market is now than it was 30 years ago?
- 1:00:32
- Unidentified speaker: A thousand times.
- 1:00:34
- Unidentified speaker: So, of course, of course, the industry is gonna have to mature around that.
- 1:00:38
- Unidentified speaker: And of course, you're gonna have to view it like a kind of later stage investor.
- 1:00:41
- Unidentified speaker: Of course.
- 1:00:42
- Unidentified speaker: But I just think that there's this funny identity crisis with a lot of later stage investors who view the world that way, just because they're in private markets and their VCs think that, for example, I would never invest in a late-stage company.
- 1:00:54
- Unidentified speaker: Like, I don't think it goes together.
- 1:00:55
- Unidentified speaker: Like, I'm gonna think, what would I do?
- 1:00:57
- Unidentified speaker: I go there and tell you how to like run a finance opera.
- 1:01:00
- Unidentified speaker: It's just not me.
- 1:01:01
- Unidentified speaker: So I think that's it.
- 1:01:02
- Unidentified speaker: It's basically the encourage of traditional banking and traditional finance and venture, which absolutely belongs there.
- 1:01:07
- Unidentified speaker: I just think sometimes it kind of reaches a little bit too low.
- 1:01:10
- Unidentified speaker: And then you would have these issues.
- 1:01:11
- Unidentified speaker: You have these issues in MA where they're like, whatever, let's run an auction and let's do all this other stuff.
- 1:01:16
- Unidentified speaker: You have it in the operations of a company where like you're 20 people and they're worried about burn, or they're over rotating on metrics for disruptive technology.
- 1:01:25
- Unidentified speaker: I just feel like we need to kind of bring back traditional productive VC for some subset of this, and there's enough market for everyone.
- 1:01:32
- Unidentified speaker: Has anything changed about how when you meet somebody for the first time?
- 1:01:36
- Unidentified speaker: Like if I came and watched 10 meetings in a week with you 10 years ago when you started, and then I came today and watched 10 meetings with you.
- 1:01:43
- Unidentified speaker: Have you changed how you conduct your early interactions with founders?
- 1:01:49
- Unidentified speaker: Yeah, a lot.
- 1:01:50
- Unidentified speaker: I mean, I think most investors go through this, especially if you come from the operating world, you have to learn how to listen as an investor.
- 1:01:58
- Unidentified speaker: And then you also have to learn how to ask the questions that you really care about so that you can answer the questions to yourself.
- 1:02:06
- Unidentified speaker: And it's a very delicate balance.
- 1:02:08
- Unidentified speaker: I do feel like a lot of meetings with founders tend to be pretty performative on both sides.
- 1:02:13
- Unidentified speaker: Like you don't know what to ask.
- 1:02:15
- Unidentified speaker: And so you'll be like, who's your competition?
- 1:02:18
- Unidentified speaker: And then a lot of it is just asking questions for asking questions' sake, but like it doesn't actually inform your decision.
- 1:02:24
- Unidentified speaker: And then, especially if you're an operator, you'll hear something you disagree with and you kind of want to fix that problem, but that's kind of not your job.
- 1:02:31
- Unidentified speaker: I have learned, and I'm not perfect, but the primary job is just to listen.
- 1:02:36
- Unidentified speaker: A founder will tell you what they believe.
- 1:02:38
- Unidentified speaker: If you disagree, you disagree, but like that's what they believe.
- 1:02:41
- Unidentified speaker: So often they'll say something like, Well, but I would do it better and therefore like it's a good investment.
- 1:02:46
- Unidentified speaker: It's truly an art.
- 1:02:47
- Unidentified speaker: You have to make a decision based on that meeting.
- 1:02:50
- Unidentified speaker: There's a lot of silly questions you could ask that are not really going to inform that decision.
- 1:02:54
- Unidentified speaker: What if Amazon does that?
- 1:02:55
- Unidentified speaker: What is OpenAI?
- 1:02:55
- Unidentified speaker: Like, whatever.
- 1:02:56
- Unidentified speaker: You can answer that yourself.
- 1:02:57
- Unidentified speaker: Maybe you can like figure out how they think, but there are key questions that you really want to know and kind of knowing which ones those are during the time of the meeting and getting those in is a hard thing to do.
- 1:03:07
- Unidentified speaker: And I think I've really kind of refined my ability to do that.
- 1:03:10
- Unidentified speaker: Maybe a fun place to close is just to ask what has you most excited?
- 1:03:15
- Unidentified speaker: We've talked about lots of the magic of this new era of technology and things that weren't possible before bottlenecks that are now gone.
- 1:03:22
- Unidentified speaker: I know all of that excites you, but if you had to just step all the way back and describe why this job is so incredibly fun right now and why you're most excited.
- 1:03:31
- Unidentified speaker: I'm more and more of the belief that this is an industrial revolution level shift.
- 1:03:36
- Unidentified speaker: And for me, the industrial revolution went from basically a zero-sum world to like a non-zero sum where value is being created, net new value is being created.
- 1:03:44
- Unidentified speaker: And I think this is a very similar shift where before, to solve problems, you had to solve very complex engineering problems, which basically didn't scale and were very tough, and then you'd have to maintain them, and that's why Alsoftware has always sucked.
- 1:03:56
- Unidentified speaker: To basically being a capital problem.
- 1:03:58
- Unidentified speaker: I don't know about the real world yet.
- 1:04:00
- Unidentified speaker: I think we'll get there, but I think we've turned this into a capital resourcing problem.
- 1:04:04
- Unidentified speaker: And we're capital allocators.
- 1:04:06
- Unidentified speaker: And so I just feel like we're just in this very, very strategic position to help drive the future in a much, much more direct way than we ever have been before.
- 1:04:16
- Unidentified speaker: Because I think this is a scale-up capital game that causes that to happen.
- 1:04:20
- Unidentified speaker: I don't believe private markets have limited TAM.
- 1:04:22
- Unidentified speaker: I think they have unlimited TAM, and the more money that goes into them, the more value that'll be created.
- 1:04:27
- Unidentified speaker: So I just feel like we're in this very privileged position, very key to the innovation chain.
- 1:04:31
- Unidentified speaker: And we're both very lucky to be here.
- 1:04:34
- Unidentified speaker: I love the every two-year schedule that we're on.
- 1:04:36
- Unidentified speaker: I hope we get to do it again in another two years.
- 1:04:38
- Unidentified speaker: I can only imagine what the world's going to look like two years from now.
- 1:04:41
- Unidentified speaker: We'll have our Grokbots talk.
- 1:04:43
- Unidentified speaker: Thank you so much for your time.
- 1:04:44
- Unidentified speaker: Thank you.
- 1:04:45
- Unidentified speaker: If you enjoyed this episode, visit Colossus.com.
- 1:04:48
- Unidentified speaker: You'll find every episode of this podcast complete with hand-edited transcripts.
- 1:04:52
- Unidentified speaker: You can also subscribe to Colossus, our quarterly print, digital, and private audio publication featuring in-depth profiles of the founders, investors, and companies that we admire most.
- 1:05:01
- Unidentified speaker: Learn more at Colossus.com slash subscribe.
- 1:05:06
- Unidentified speaker: Patrick O'Shaughnessy is the CEO of Positive Sum.
- 1:05:09
- Unidentified speaker: All opinions expressed by Patrick and podcast guests are solely their own opinions and do not reflect the opinion of Positive Sum.
- 1:05:16
- Unidentified speaker: This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions.
- 1:05:22
- Unidentified speaker: Clients of Positive Sum may maintain positions in the securities discussed in this podcast.
- 1:05:26
- Unidentified speaker: To learn more, visit psum.cc.
- 1:05:50
- Unidentified speaker: You know how small advantages compound over time?
- 1:05:52
- Unidentified speaker: That's true in investing and just as true in how you run your company.
- 1:05:55
- Unidentified speaker: Your spending system is your capital allocation strategy.
- 1:05:58
- Unidentified speaker: Ramp makes it smarter by default.
- 1:06:00
- Unidentified speaker: Better data, better decisions, better economics over time.
- 1:06:02
- Unidentified speaker: See how at ramp.com slash invest.
- 1:06:05
- Unidentified speaker: As your business grows, Vanta scales with you, automating compliance and giving you a single source of truth for security and risk.
- 1:06:11
- Unidentified speaker: Learn more at vanta.com slash invest.
- 1:06:14
- Unidentified speaker: The best AI and software companies, from open AI to cursor to perplexity, use WorkOS to become enterprise ready overnight, not in months.
- 1:06:21
- Unidentified speaker: Visit workos.com to skip the unglamorous infrastructure work and focus on your product.
- 1:06:25
- Unidentified speaker: Ridgeline is redefining asset management technology as a true partner, not just a software vendor.
- 1:06:30
- Unidentified speaker: They've helped firms 5x in scale, enabling faster growth, smarter operations, and a competitive edge.
- 1:06:36
- Unidentified speaker: Visit RidgelineApps.com to see what they can unlock for your firm.
- 1:06:40
- Unidentified speaker: Every investment firm is unique and generic AI doesn't understand your process.
- 1:06:44
- Unidentified speaker: Rogo does.
- 1:06:45
- Unidentified speaker: It's an AI platform built specifically for Wall Street, connected to your data, understanding your process, and producing real outputs.
- 1:06:51
- Unidentified speaker: Check them out at rogo.aiv.
More from Invest Like the Best with Patrick O'Shaughnessy
Zach Reitano - Building Ro and Why Healthcare Should Be 50% of GDP - [Invest Like the Best, EP.500]
Episode discussion: Zach Reitano on Ro's 10x growth via GLP-1s, culture reset, and healthcare's future
Speakers’ claims: Zach Reitano argues healthcare should be 50% of GDP, citing misaligned incentives where patients are not the customers. He details Ro's turnaround from a 'zombie company' to a $3B+ run rate through a 'strong attract, strong repel' culture and operational efficiency. Reitano predicts GLP-1s will see 50%+ US penetration, driving prices down via direct-to-consumer models. He contrasts this with AI, noting GLP-1 revenue exceeds AI revenue and impacts more individuals. Reitano advocates for prioritizing health outcomes over job preservation and leveraging AI to make concierge care a commodity.
Ramtin Naimi on the State of Venture Investing - [Invest Like the Best, EP.499]
Episode discussion: Ramtin Naimi on VC market speed, 'cuck tranches', and why proactive investing in non-obvious markets generates alpha
Speakers’ claims: Ramtin Naimi describes the current venture market as extremely fast-paced, with deal cycles compressed and top companies receiving disproportionate demand. He highlights a 'cuck tranche' phenomenon where late entrants pay 3x-5x premiums to join primary rounds, driven by capital-intensive needs in AI. Naimi contrasts current high-quality, product-market-fit companies with the 2021 SaaS boom, noting that while valuations are high, the underlying business fundamentals are stronger. He argues that proactive investing in non-obvious markets yields better returns than chasing consensus late-stage deals. Naimi also discusses the impact of LP concentration in mega-companies and the rising costs of compute for private AI labs.