Partnership

TL;DR

AI startup that attracts and retains generational talent for the long-haul via KKR / McKinsey / Cravath / Benchmark / Millennium-style partnership

Companies

Relevant Models: KKR, Cravath, McKinsey, Benchmark, Millennium

Companies Experimenting with the idea: Rippling, Scale, Palantir, Netflix

Overview

Top 0.01% engineers are orders of magnitude more productive than those elsewhere on the bell curve See the resumes of Jeff Dean, Dennis Ritchie, Ken Thompson, John Carmack, Bret Taylor, Linus Torvalds, Michael Stonebraker, Geoffrey Hinton, Noam Shazeer, Diederik P. Kingma, John Schulman, etc.
Hiring top 0.01% engineers’ has gotten exponentially more expensive for decades:
From science being only for the pay-less hobbies of landed aristocracy → fairchild offering early employees intellectual autonomy → Google offering sleep pods & 20% Time → Facebook inventing RSUs → Ramp offering college dropouts $700k → early Ramp employees being offering offered $2M checks from VCs to drop out of Ramp → acqui-hires Meta offering individual engineers $1B comp and concentrating computing resources of entire countries in the top 20 people
Combine those two and it’s possible that the way to create a new AI startup with generational talent that remains committed for the long-haul instead of leaving after their equity vests to start their own thing is a partnership modeled after industries with a few extremely high leverage individuals like PE, hedge funds, IB, law, and consulting where partners get a direct share of the 10-100s of millions in revenue they bring in a year:

if you were andy jassy at Amazon in 2006 in today’s funding environment, what incentive would you have to create AWS, making Jeff Bezos 1000x more wealth than you, when you have so little exposure… if you can just get any VC to write you 20 at 200 post to start aws yourself? btw jassy’s not even a billionaire while software engineering is traditionally thought of as “high leverage”, the revenue commanded per employee at google or facebook is only around ~$1.5M. this pales in comparison to the $20M every mckinsey partner is expected to bring into the firm. today, the best engineers are JUST starting to rival the leverage commanded by a general partner at KKR or cravath partner. of course, there are going to be challenges in structuring R&D as an partner w/ revenue exposure - when it’s not directly revenue generating. there will need to be more modular ways of valuing engineering output. it’s not obvious why software assets couldn’t be treated like other assets like real estate, just make the asset amortization faster and treat it like super-high maintenance costs. these partnerships could be more decentralized or more centralized, and the unit of incentive can vary greatly. engineers could be compensated with royalties on future revenue of a piece of software, future profit, exposure to the P&L, a liquid internal equity-like market that benchmarks the value of a subsidiary companies stock against the parents. it’s already happening. the most efficient companies in tech are already structured like this - rippling hires ex-founders to be general managers who own a horizontal slice, and have direct competitors outside of rippling to power the compound startup model - scale has general managers that run P&Ls on verticals, scaling up and down their budgets based on the amount they’ve grown in the past quarter, with the ceo allocating budgets like a VC investing rounds - palantir expects every customer to grow to $10m/yr, causing them to deploy forward deployed engineers as “CTOs of the specific customer” - and draw resources as demand grows - netflix gives wide deference to team leads to set their own policies and hires only the higher leverage engineers w/ top of market pay. the 10x leveraged exercised by the best is explicitly referred to in their culture docs

TL;DR

AI startup that attracts and retains generational talent for the long-haul via KKR / McKinsey / Cravath / Benchmark / Millennium-style partnership

Companies

Relevant Models: KKR, Cravath, McKinsey, Benchmark, Millennium

Companies Experimenting with the idea: Rippling, Scale, Palantir, Netflix

Overview

Top 0.01% engineers are orders of magnitude more productive than those elsewhere on the bell curve See the resumes of Jeff Dean, Dennis Ritchie, Ken Thompson, John Carmack, Bret Taylor, Linus Torvalds, Michael Stonebraker, Geoffrey Hinton, Noam Shazeer, Diederik P. Kingma, John Schulman, etc.
Hiring top 0.01% engineers’ has gotten exponentially more expensive for decades:
From science being only for the pay-less hobbies of landed aristocracy → fairchild offering early employees intellectual autonomy → Google offering sleep pods & 20% Time → Facebook inventing RSUs → Ramp offering college dropouts $700k → early Ramp employees being offering offered $2M checks from VCs to drop out of Ramp → acqui-hires Meta offering individual engineers $1B comp and concentrating computing resources of entire countries in the top 20 people
Combine those two and it’s possible that the way to create a new AI startup with generational talent that remains committed for the long-haul instead of leaving after their equity vests to start their own thing is a partnership modeled after industries with a few extremely high leverage individuals like PE, hedge funds, IB, law, and consulting where partners get a direct share of the 10-100s of millions in revenue they bring in a year:

if you were andy jassy at Amazon in 2006 in today’s funding environment, what incentive would you have to create AWS, making Jeff Bezos 1000x more wealth than you, when you have so little exposure… if you can just get any VC to write you 20 at 200 post to start aws yourself? btw jassy’s not even a billionaire while software engineering is traditionally thought of as “high leverage”, the revenue commanded per employee at google or facebook is only around ~$1.5M. this pales in comparison to the $20M every mckinsey partner is expected to bring into the firm. today, the best engineers are JUST starting to rival the leverage commanded by a general partner at KKR or cravath partner. of course, there are going to be challenges in structuring R&D as an partner w/ revenue exposure - when it’s not directly revenue generating. there will need to be more modular ways of valuing engineering output. it’s not obvious why software assets couldn’t be treated like other assets like real estate, just make the asset amortization faster and treat it like super-high maintenance costs. these partnerships could be more decentralized or more centralized, and the unit of incentive can vary greatly. engineers could be compensated with royalties on future revenue of a piece of software, future profit, exposure to the P&L, a liquid internal equity-like market that benchmarks the value of a subsidiary companies stock against the parents. it’s already happening. the most efficient companies in tech are already structured like this - rippling hires ex-founders to be general managers who own a horizontal slice, and have direct competitors outside of rippling to power the compound startup model - scale has general managers that run P&Ls on verticals, scaling up and down their budgets based on the amount they’ve grown in the past quarter, with the ceo allocating budgets like a VC investing rounds - palantir expects every customer to grow to $10m/yr, causing them to deploy forward deployed engineers as “CTOs of the specific customer” - and draw resources as demand grows - netflix gives wide deference to team leads to set their own policies and hires only the higher leverage engineers w/ top of market pay. the 10x leveraged exercised by the best is explicitly referred to in their culture docs

Partnership

TL;DR

Companies

Overview

Further Reading

Partnership

TL;DR

Companies

Overview

Further Reading