Nov 8, 2024 · 1h 9m · mad
Superintelligence, Bubbles And Big Bets: AI Investing in 2024 | Matt Turck & Aman Kabeer, FirstMark
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this end-of-year 2024 episode of The MAD Podcast, FirstMark venture capitalists Matt Turck and Aman Kabeer analyze the state of artificial intelligence, evaluating record startup valuations, infrastructure demands, and the debate surrounding an AI bubble. They break down investment opportunities across the AI stack, contrasting enterprise deployment realities with shifting market dynamics and long-term tech trajectories.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 84.4% of the talking time here. How this is scored →
Matt as informed peer 7.3 Guest teaching 1.7 Guest disagreement 0.3 Matt pushing back 0.3
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 26:43 Questioning Google's CodeGen Claims
Aman pushes back on Sundar Pichai's claim that AI generates over 25% of Google's code, noting internal engineers pointed out it is mostly basic auto-complete rather than true generative code creation.
Hardest push from Matt ▶ 1:01:04 Refuting the Death of SaaS PremiseMatt directly rejects the premise that AI will kill software, arguing instead that SaaS will evolve from wrappers around databases into wrappers around intelligence.
Biggest teaching moment ▶ 51:06 FirstMark CTO Guild Survey Reality CheckAman presents concrete internal survey data from FirstMark's CTO Guild showing that 62% of CTOs who adopted AI in the past year were underwhelmed by its impact.
Matt holds his own ▶ 14:09 Public SaaS vs Pre-Revenue AI Valuation MathMatt demonstrates deep market mastery by contrasting public SaaS valuation standards ($375M NTM revenue needed for a $3B valuation at 8x) against pre-revenue AI startups commanding identical valuations.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| Mega Rounds, Record Valuations, and Unprecedented Acquires | 6 | 1 | 0 | 0 | Matt Turck opens with an extensive breakdown of major AI funding milestones, citing OpenAI's $6.6B round, SoftBank's $9T capex estimates, and Elon Musk's Colossus data center. Co-host Aman Kabeer contributes complementary figures on Safe Superintelligence and Character.AI without conflict. | |
| Public Market Multiples, Nvidia/Palantir Valuations, and the AI Hardware IPO Wave | 7 | 2 | 0 | 0 | Matt details Nvidia's financial multiples relative to market averages and analyzes the Cerebras S-1 filing. Aman adds market context around Palantir's multiple and G42 CFIUS regulatory issues. | |
| Private Market AI Valuations vs. Actual Revenue Scale | 7 | 2 | 0 | 0 | Aman introduces Sierra's 225x ARR multiple and internal portfolio data showing $40B in market value across pre-scale startups. Matt analyzes the stark contrast between public SaaS median multiples (5-6x) and pre-revenue AI valuations, describing the split reality VCs face daily. | |
| Debating the AI Bubble: The Case FOR a Bubble and Spending Imbalances | 8 | 1 | 0 | 1 | Matt details the case for an AI bubble, referencing David Kahn's $600B revenue gap paper, Goldman Sachs reports, hardware timeline risks, and scaling laws. He also shares direct insights from a personal conversation with Sam Altman at OpenAI. | |
| The Case AGAINST the AI Bubble: Hypergrowth, Massive Demand, and Technical Breakthroughs | 7 | 3 | 1 | 0 | Matt outlines counterarguments to the bubble thesis using Big Tech earnings data, model reasoning breakthroughs (o1), and 90% token price declines. Aman provides context on Stripe startup acceleration metrics and highlights community skepticism regarding Google's AI code generation metrics. | |
| Does an AI Bubble Matter? The Dot-Com Analogy | 7 | 1 | 0 | 0 | Matt frames the AI spending boom using the dot-com era analogy, arguing VCs must play on the field despite risk of busts to catch generational winners like Amazon or Google. He highlights disagreement among AI luminaries (LeCun, Fei-Fei Li) on defining AGI versus ASI. | |
| Pragmatic AI Deployment vs. The Fade of AI Doomerism | 6 | 1 | 0 | 0 | Matt points out the rapid dissipation of 2023 AI doomerism as focus pivots toward enterprise deployment realities. Both speakers agree that procurement and compliance are the real friction points rather than existential risk. | |
| The AI Stack: Infrastructure & Frontier Model Competition | 8 | 1 | 0 | 0 | Matt maps out FirstMark's venture framework across the AI stack, explaining why they pass on compute and frontier foundation model rounds due to fund sizing and lack of durable differentiation. He notes competitive pressures from open-source models like Meta's Llama. | |
| Investing in Specialized Models: Modalities & Automation | 8 | 2 | 1 | 0 | Matt shares historical venture lessons from early investments in Dataiku, breaking down current dynamics in LLM evaluation and open-source agent frameworks. He provides a sharp analysis of specialized vector database risks as incumbent general-purpose databases add vector search. | |
| Consumer AI Applications & The Billionaire Test | 8 | 2 | 0 | 0 | Matt contrasts previous mobile paradigm shifts with AI-native consumer apps, citing Suno AI as a novel model. He expands on Richard Socher's 'Billionaire Test' mental model to predict consumer software opportunities in AI tutoring and personal assistants. | |
| Enterprise AI Reality: Secret Cyborgs, Consultants & Data Readiness | 8 | 3 | 1 | 0 | Aman shares FirstMark CTO Guild data showing 62% of AI adopters felt underwhelmed by initial impact. Matt expands on this with US Census data to illustrate 'secret cyborg' shadow usage by employees versus corporate adoption delays, low-hanging fruit deployment, and massive IT consulting revenue. | |
| The Reinvention of SaaS, Outcome-Based Pricing & Agent Networks | 8 | 1 | 1 | 2 | Matt explicitly refutes the premise that AI means the 'death of SaaS', explaining the transition from database wrappers to intelligence wrappers. He illustrates the shift toward outcome-based pricing using portfolio company Ada as a concrete case study. |
Statements from this episode (39)
OpenAI's $6.6B round was the largest VC round ever
“We just had the biggest venture capital round of all time with OpenAI, which raised 6.6 billion at one hundred and fifty seven billion post-money valuation.”
Safe Superintelligence raised a $1B seed round pre-product
“There's also kind of the biggest seed round ever, right, that we saw. Safe superintelligence helmed by Former OpenAI chief scientist and co-founder Ilya Sutskover raising roughly a billion dollars in cash at a five billion dollar valuation pre-basically anythi…”
Google's $2.7B Character.AI deal was the largest acquihire ever
“And, you know, we also had the biggest acquihire of all time. So for those of you who don't know, 2.7 billion dollar acquihire by Google of the Character AI founders who, funny enough, were already at Google before they went to start Character AI.”
Meta, Google, and Amazon on track for $200B in AI capex
“The three top companies, MetaGoogle and Amazon, are on track to invest two hundred billion in AI infrastructure this year.”
Masayoshi Son estimates superintelligence will require $9T in cumulative capex
“Masayoshi Son, the CEO of SoftBank, mentioning just a few days ago that, Inez estimate to reach super intelligence that was going to be a cumulative capex, capital expenditure budget of nine trillion dollars, which you know, in typical MASA fashion he said was…”
xAI built its Memphis Colossus data center in 122 days
“Colossus, which is based in Memphis, which by the way apparently was built in a 122 days. So the fastest you know, time ever to build a data center.”
Big tech and AI companies are driving a U.S. nuclear revival
“But yeah, in 24, it turns out that big tech companies and AI companies are the ones driving this, Revival of nuclear power with, you know, Microsoft entering into a deal with Constellation to revive Three Mile Island, which was the site of the last nuclear acc…”
Nvidia generates $53 billion net income on $96 billion revenue
“It's a ninety six billion revenue company that generates fifty three billion in, in net income.”
Palantir is the most richly valued software company today
“It's the most richly valued software company today.”
Palantir has reached $2.7 billion in ARR
“They're, you know, at 2.7 billion of ARR today, which is pretty massive.”
Sierra quadrupled its valuation to $4.5B on $20M ARR
“Yes, current chairman of OpenAI as well which is more than quadrupling its valuation this year to four and a half billion. You know, on the other side of that equation is roughly twenty million of ARR.”
Private AI startups hold a $40B valuation against under $100M revenue
“And basically the kind of end, end all of the chart kind of proves the point that you have, let's say, roughly forty billion dollars in valuation across a number of these names but less than a hundred million dollars of revenue.”
Top 10 public SaaS companies trade at 14-15x revenue multiples
“The top 10 names in the public markets are somewhere in the 14 to 15 X range. And that's just the top 10 companies in the world today on the software side, but the actual median is closer to five to six X. In the past it's closer to 10 X.”
AI venture investing is rebuilding another 2021-style bubble
“Half of your time, you basically untangle the consequences of the excess period of twenty-twenty-one, where you know, you think of helping, how to help company get fit and you know, cut burn, extend runway, be efficient, all the things. That's like one part of…”
OpenAI's $6.6B round required a $250M minimum ticket
“In the OpenAI round that we're talking about, you know, the 6.6 billion dollar round. The minimum ticket size was two hundred and fifty million.”
Sam Altman asserts that AI scaling laws will absolutely continue
“I got a chance to ask that question to Sam Altman a few weeks ago. I was at an event at an event at OpenAI. Not surprisingly, when I asked him, you know, will scaling laws continue, he looked me straight in the eye and said, absolutely”
OpenAI aims to reach $100B in ARR by 2029
“The company has extraordinary ambitions to be at a hundred billion dollar in ARR in 29”
OpenAI's funding deal was priced at 13.5x forward revenue
“The multiple on the deal on a forward revenue basis was 13.5 X.”
AI startups reach $30M ARR five times faster than legacy SaaS
“The AI startups in the cohort that Stripe measured took 11 months to reach one million in ARR after their first sales on Stripe versus 15 for the previous generation of SaaS businesses, and then maybe more impressively, you know, scaling to more than thirty mi…”
Over 25% of new code at Google is AI-generated
“Sundar Pashai said, you know, more than a quarter of all code written at Google is now generated by AI before it's reviewed and accepted by engineers”
AWS AI business is growing 3x faster than original AWS
“AWS's AI business is a multi-billion dollar revenue run rate business that continues to grow at a triple digit year over year percentage, and is growing more than three X faster at this stage of its evolution as AWS itself grew”
GPT-4 price per token dropped roughly 90% in one year
“The, I think the price per token of GPT-IV dropped something like 90%. Over the last year”
Billions will likely be incinerated in the AI boom
“It's possible, perhaps likely, that this is going to be the same thing here, that the same thing is going to happen here, where, like, another few billion dollars are going to be incinerated in, in this moment.”
Sitting out the dot-com boom meant missing foundational tech companies
“And if you would have sat out then you would have, you know, maybe avoided a couple of busts, but you would have also missed on some of the most foundational companies of our lives.”
VCs are suckers for big visions like artificial superintelligence
“You know, we VCs are suckers for a big vision, so super from general is, is, is a VC magnet.”
Great companies can be built using current AI without further progress
“Actually just with what we have, there's Plenty we can do. Plenty of great companies we can build. Just taking the current state of models and sort of deploying them in a way that's, you know, integrated in workflows and, you know, verticalized per industry.”
Enterprise AI deployment is hindered mostly by non-AI organizational hurdles
“Even deploying AI there's a lot of things that have nothing to do with AI itself. Like, you know, to get you know, procurement authorization and compliance and all the things. So you know, far from killing us all just like deploying it in your department is is…”
Nvidia released an AI model that outperformed GPT-4
“Just a couple of weeks ago, Nvidia announced that In addition to having the best chips, they just released a model that was actually better than GPT-IV.”
LLM performance evaluation is arguably more critical than for traditional software
“Those things are stochastic, not deterministic, meaning that they don't get the right answer a hundred percent of the time, and you don't get the same answer each time you ask. Therefore, evaluating performance is more important arguably, than any other kind o…”
Open-source AI popularity is fleeting and does not equal commercial traction
“As we've seen over the past year popularity doesn't necessarily translate into commercial traction, and popularity is pretty fleeting in the world of AI.”
Suno AI is an early candidate for an AI-native app
“One company that we discussed a couple of times on the Mad podcast that could be an early candidate for what a truly you know, AI-native consumer app what it might be would be Suno.”
Many people use ChatGPT as an informal therapy tool
“A lot of people use ChatGPT to just, like, sort of vent and talk.”
FirstMark survey: 62% of CTOs adopting AI were underwhelmed by impact
“While 64% of, you know, CTOs in our guild had adopted AI in some form or the other in the last 12 months, 62% of those that had adopted AI were underwhelmed by the impact it had on their organization.”
39% of Americans use ChatGPT and 28% use it at work
“39% of Americans say that they use ChatGPT, including 28% that say that they use it at work, and 11% use it every day.”
Census data shows only 5% of U.S. businesses use AI directly
“Only five percent of American businesses say that they are using AI technology to produce goods or services.”
Accenture reported $3B in annual generative AI consulting fees
“Famously Accenture reported three billion dollars in in annual fees which You know, made the headlines and all the things.”
SaaS will not die, but evolve into wrappers around AI intelligence
“We think that SAS It's not going to be dead. We think that SAS is just going to have to evolve dramatically. And if you think of SAS being largely a wrapper around a database, and you could argue that's what Salesforce is, you know, a lot of, like, workflow an…”
Customer service AI startup Ada shifted pricing to per-resolution billing
“And we saw, saw that in our portfolio with Ada, which is one of the leaders in AI chatbots for customer service, where, you know, not that long ago, they would sell on a per seat basis, and, you know, occasionally, you know, per conversation basis, sort of sel…”