your moat is a lie

there was a great article that i read a while back by benn stancil that talked about how clever sounding moats don’t actually do anything in real life.

it reminded me of when ben horowitz talked about his fight w/ bladelogic in hard thing about hard things:

if both ben and benn agree - it must be true right?

winning isn’t the result of being clever - it’s the result of outworking the opposition head on

moats are silver bullets. they’re clever insights that make you feel really good because philosophically it is a simple pattern that you think clinging to ensures that you can convince anyone that you’ll win. it’s the favorite tool of the armchair strategist or monday morning quarterback.

there are people we’ve interviewed at textql who have asked me to talk about why we’re going to win super long term against players like palantir or c3.ai. the question boils down to “what is your moat“. it also isn’t how companies win.

companies win by rolling up their shirtsleeves, putting on the brass knuckles, lining up by the starting gun alongside your competitor and beating the shit out of them in the streets. it means outworking them on every deal you get your hands on - you care more about the customer, you build solutions faster, you are more eloquent, you understand more. none of these things can be summarized by a quippy one-liner like “oh this is how we win”… unless the answer “we’re going to be better than them in every way” is a good answer

now i’m sure a dozen people who fancy themselves strategic thinkers are reading this and recounting “oh what about the time this company x did y thing that caused them to win”.

this is obviously a just so story.

• mpany builds a better product along every margin, and rides the engineering team to talk to customers, works 16 hours a day

• during the fundraise, since talking about how hard you work doesn’t tell a convincing narrative to investors about how your margins are going to improve, you try to come up with a quippy one-liner that’s easy to digest and convinces your investors gross margins are going to go to 100%

• afterwards - everyone who hears the story attributes that success to the founder’s ability to find that quippy insight

• in 10 years, people say things like “wow zuckerberg’s idea to do social media was so original and was the reason they won“

also - vcs are aware it’s a stupid question. if it was true that X company succeeded bc of their moat or insight, and wasn’t just the case that the people were just categorically better and outclassed the competition - then they wouldn’t be so willing to invest in early ex-stripe, uber, palantir operators-turned-founders.

they’d think “oh i mean those companies won bc of their moat and strategic thinking - not bc their people were just categorically superior.” yet we all know that if you have the best people - you can pretty much start anything, and you’ll get to success

^which does raise a kinda scary thought - if you don’t succeed, you probably just suck. and it wasn’t related luck, and the base rate has nothing to do with it.

so what makes a company valuable?

how much would you value snowflake at if all of snowflakes traction (brand & revenue & customers) disappeared tomorrow? ignore the market cap & just answer in percentage mark down

the value of a company is distributed across 3 parts:

• leadership & people | the brains, the strategy, the dna

  • consider what if the company got acquired tmr by KKR, brought in generic leadership, etc

  • how much of the company’s value is tied up in their ability to be a “live-player”

    • companies w/ very high index on this are palantir, ramp, scale

• brand & traction | current market momentum, brand, customers

  • consider what would happen if the customer lost all of its contracts & brand tomorrow?

  • how much of the company’s value is tied up in their snowballing “you buy this bc that’s what companies at your stage do“

    • companies w/ very high index on this are snowflake, confluent… really most hot software incumbents

• product & technology

  • consider what would happen if someone did a full wipe of a company’s codebase + physical assets

  • how much of a company’s success is tied up w/ their physical or digital goods they own

    • companies w/ very high index on this are companies w/ very large projects that are complex to stand u, like SAP, Cerner, or TSMC

if you replaced snowflake’s leadership w/ KKR clones - i honestly wouldn’t expect to mark it down by more than 20%. i don’t think a ton of snowflake’s future success is tied up in their ability to make strategic decisions - they’re a massive market incumbent with good-enough technology and very strong unit economics that carry itself forward

if you deleted all of snowflake’s software tomorrow - i’d barely mark down snowflake’s valuation at all (provided it got a few weeks of grace period to build back up on existing contracts). the reality is - snowflake isn’t very hard to build. firebolt, clickhouse, oracle heatwave are all architecturally identical products on speed & performance - building it from scratch again might actually be a good exercise to clear out a bunch of existing bugs. w/ snowflake’s people + reputation, i think snowflake would be back to where it is inside if 6 months.

if you removed snowflake’s momentum, but retained the leadership & software, i think it’d be over. i’d mark down snowflake by 80%. snowflake is a classic “buy it because it’s the incumbent” solution - where most data engineers i’ve met can’t even articulate why it’s better than redshift or clickhouse. it’s just become “the thing you buy” because it’s what companies do.

this might contrast to a company like scale - where the entire company seems to pivot around alexandr wang and the humans at scale.

every individual i’ve met there is a monster and seems to be ready to grind 80 hours out on a whim on a totally new contract w/ a new use case. from a distance - it looks like their contracts haven’t historically been designed to be recurring (data labeling needs vary by year) and their account managers need to find new dynamic uses cases to drive them up each renewal cycle.

their software is often also built out very quickly for specific use cases that pop up.

all of this creates a company where low retention high net-new-revenue contracts + short software lifecycles mean that the people at scale make up the bulk of their valuation.

certain companies can of course index more heavily on two - and have one that’s obviously unimportant.

the most obvious case is TSMC or NVIDIA - if they lost all of their reputation, brand and customers, their value probably wouldn’t move at all. the combination of their people and technology would immediately carry their reputation back to where it is.

in other words, reputation is highly liquid in the space of semiconductors and GPUs - the benchmarks are clear and undeniable.

which raises an interesting question - why is reputation in databases not nearly as liquid, w/ CIOs willing to invest in snowflake when they only have 100gbs of data, despite databases having clear quantifiable benchmarks…