Mar 5, 2025 · 25m · tbpn
"Work at an outstanding organization first" - Keith Rabois Advice on Starting Companies
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
Venture capitalist Keith Rabois shares deep insights on Ramp's rapid ascent and contrarian company building, offering foundational advice for emerging founders alongside critical analyses of AI durability, robotics economics, and tech governance.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The hosts as informed peer 4.8 Guest teaching 5.0 Guest disagreement 2.4 The hosts pushing back 1.6
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 24:07 Keith instantly reframes the question on Sundar Pichai
Rather than accepting the host's premise about Sundar failing compared to Satya, Keith dismisses it by pointing out Apple is performing far worse with multi-year delays.
Hardest push from the hosts ▶ 14:37 Host challenges assumptions on robotics economics via DJI precedentThe host directs the robotics discussion to hardware dumping, detailing how Unitree and DJI price units below component cost to capture market share.
Biggest teaching moment ▶ 5:45 Keith's empirical breakdown of NFL rookie quarterbacks and startup timingKeith educates the host on career timing with data on rookie quarterbacks who start immediately versus those who sit on the bench before leading.
The host holds their own ▶ 14:37 Host demonstrates deep knowledge of bill-of-materials and market subsidizationThe host demonstrates deep analytical command by citing teardown component costs and structural geopolitical advantages in Chinese hardware distribution.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Ramp's Momentum and Transforming the Modern CFO Suite | 3 | 4 | 1 | 1 | The host warmly invites Keith to give an update on Ramp's valuation and progress. Keith provides a detailed thesis on modernizing the archaic CFO suite and maintaining product velocity. | |
| The Rare Synergy of Ramp's Co-Founders | 3 | 6 | 2 | 1 | The host asks about founder archetypes and whether Gen Z founders should build simple wrappers or swing big. Keith educates the host using an empirical NFL quarterback career progression study to explain when to start a company. | |
| Ramp's Inception and Contrarian Investing in Crowded Markets | 4 | 5 | 2 | 1 | The host inquires about contrarian market categories ripe for disruption. Keith recounts his 2019 thesis on replacing corporate credit cards and convincing Founders Fund despite existing competitors like Brex and Amex. | |
| Next-Generation AI Shifts and the Durability of AI Revenue | 5 | 6 | 2 | 2 | The host contrasts dot-com bubble valuations with modern nine-figure AI revenues. Keith breaks down the durability hierarchy across consumer subscriptions, pilot programs, and deep enterprise process integrations. | |
| Humanoid Robotics, Capital Influx, and Geopolitical Stakes | 7 | 4 | 3 | 2 | The host brings deep hardware and market knowledge by comparing Unitree's pricing strategy to DJI's below-cost hardware dumping. Keith strongly agrees and outlines the national security and military implications of humanoid robots. | |
| Warehouse Robotics, Human Labor Realities, and Hardware Decay | 6 | 5 | 2 | 2 | The host references Ryan Petersen's views on human labor efficiency and asks about hardware depreciation curves. Keith illustrates wear-and-tear using a commercial gym equipment metaphor to explain unproven economic payback periods. | |
| Critique of the Strategic Bitcoin Reserve and Sovereign Funds | 5 | 4 | 3 | 2 | The host questions the implementation of a strategic Bitcoin reserve and suggests utilizing seized assets. Keith expresses skepticism toward government-run funds and advocates strict limitations on federal capital expenditure. | |
| Washington D.C. as a Startup Hub and Big Tech AI Leadership | 5 | 6 | 4 | 2 | When the host asks if Sundar Pichai will be fired due to AI underperformance compared to Satya Nadella, Keith sharply pivots to critique Apple's multi-year delay and challenges assumptions about Microsoft's long-term dominance. |
Statements from this episode (17)
Rabois: Even cutting-edge companies run archaic, analog CFO functions
“Even the most cutting edge organizations typically have a very archaic, call it analog CFO function.”
Rabois: The best CEOs serve as their company's chief marketing officer
“The best CEOs are the chief marketing officer for the company.”
Rabois: Founders usually benefit from working at an outstanding organization first
“I think it's most useful most of the time for founders to work in an outstanding organization for a year or two before starting their own company. You get a feel for a taste, taste of what great is, like what excellent really, excellence really means, excellen…”
Rabois: Ramp was a non-standard investment for Founders Fund
“When I was working at founders fund at the time, There's a very non-standard Founders Fund investment in many, many ways. The most acute one being, if you have all these vendors that are doing well, what's the monopoly story?”
Rabois wants to fund next-generation AI that completely obsoletes LLMs
“I would love to invest in something that's a next generation replacement for LLM. It's not a better LLM, but something that will, you know, transforms a pretty good double tundra, but that's something that would Be a you know, complete it would make LLMs obsol…”
Rabois: Most OpenAI revenue comes from durable ChatGPT subscriptions
“OpenAI, most of their revenue comes from ChatGPT subscriptions. That's pretty durable. Consumers or people using it, you know, for business applications, but on an individual basis.”
Rabois: Enterprise-wide AI deployments yield durable 5+ year revenue
“If you sell to an enterprise and they roll out to the entire organization, this AI based product, the chance that they're going to swap, you know, in any short term is very, very low. So if you get through it, you get through procurement, you know, you have a …”
Rabois: Current AI fine-tuning and training methods may be obsolete within two years
“The advances in AI are so fast. That the way we build models today with fine tuning and different kinds of training may be also obsolete a year, two years from now. So there may be models that are built that don't require any human training. So revenues that a…”
Rabois: China is ahead of the US in humanoid robotics
“Well, spicy is that look, Chinese are ahead of us and it's very dangerous politically, geopolitically.”
Rabois: High startup proliferation and valuations make investing in robotics hard
“I think as a VC right now, it's really hard to invest in a sector, given the proliferation of startups, all with high potential, the prices you'd have to invest that money.”
Rabois: Trump administration will find a way to ban Chinese robots
“I would be shocked if this administration doesn't figure out a way to ban Chinese robots before it's too late.”
Rabois: Outsider robotics firms will fail to replace warehouse workers
“Some company from afar that doesn't understand how a warehouse works, that doesn't understand real workers and what real workers, you know, needs are and tries to run a mixed system without having any understanding of real people. I don't think it's going to b…”
Rabois: Nobody understands humanoid robots' physical decay curve
“Well, I don't think anybody really knows the decay curve, you know, because what is wear and tear due to these robots? Like, and what functions, too? It may be very different.”
Rabois opposes government business functions and a US sovereign wealth fund
“But I'm not the biggest fan of like the government doing performing business functions, period. So I don't like the sovereign wealth fund either.”
Rabois: Founding a non-government startup in DC is probably a mistake
“But I think if you're an early stage company, that's not really targeting the
The government is your first customer, federal.
You know, in some way, then I think it's probably a mistake right now.”
Rabois: Apple's reported two-year AI delay is a devastating, generational setback
“Two years is like, you know, as other people have pointed out, two years in an AI world is an eternity. You might as well say I'm delaying something for 20 years or, you know, like two generations at this point. It's crazy. And so I think that's a serious prob…”
Rabois: Microsoft will not hold as much of a strategic lock on AI
“I think Sundar made several miscalculations, and I actually do think that Microsoft and, you know, Satya got a lot of credit, but I think they're seeing some of the downsides to some of their strategies now. Like I think they are going to wind up not having qu…”