Cursor: IDE of the Tiger...?

Cursor's crazy revenue & 10B valuation - is it sustainable? Lets talk TAM, revenue quality, and other sources of value Cursor might be sitting on (frontier coding model??).

Episode Transcript

Rod Edwards (00:00):
Hey, if you're listening to this, thanks for
joining me. My name is RodEdwards, and welcome to the
minimum viable banter podcastinaugural episode. Today, what
I'd really love to talk aboutfor a few minutes is Cursor, the
10,000,000,000 semi automatedcoding phenomenon that just
finished raising a $900,000,000round. The basic question is, is

(00:21):
this rational? We're gonna talkabout that today on Minimum
Viable Banter.
So, Cursor, the current posterchild of the AI hype train. And

(00:43):
to be clear, I say that with,you know, a bit of sarcasm baked
into it or skepticism perhaps.But for the record, I'm fully on
board of the AI hype train. Igot my ticket a long time ago
and I'm certainly enjoying theride. But Cursor is one of those
standout transactions in theventure capital realm that

(01:03):
raises a lot of eyebrows and notin the least because it's such
an exciting story.
Cursor, the product, which isrun by a bigger umbrella
organization, but Cursor, theproduct, is three years old and
in that time has grown to be a$10,000,000,000 company. A lot
of people think of it as thefastest growing startup in
history, just based on itsrevenue trajectory, which, you

(01:27):
know, is pretty wild. You know,this year, for instance, Cursor
has scaled their revenue from$100,000,000 of ARR to over
$500,000,000 and, you know,that's not nothing. Does that
command a $10,000,000,000valuation? So to qualify that,
to begin with, the valuation isactually $9,900,000,000 post

(01:48):
money, pre money it's$9,000,000,000 I'm just rounding
up to $10,000,000,000 becauseit's way easier to talk about.
It also makes the math easier todo. With $500,000,000 of ARR,
dollars 10,000,000,000 valuationrepresents a 20X multiple on
ARR. That is a good multiple inthe current SaaS valuation
environment. Know, honestly, 20xdoesn't seem crazy for something

(02:12):
that has 5x'd their revenue inthe last six months. You know,
that 20x, I think, bakes in afair amount of skepticism
perhaps.
It's funny to look at, you know,a three year old company with a
$10,000,000,000 valuation thatjust raised $900,000,000 and say

(02:33):
that that's a skepticalvaluation, but it sort of feels
like it is. The big questionthat I'll come back to from the
beginning is, is this rational?Let me ask three questions or
three streams of consciousnessabout the Cursor story. The
question is growth. They'vegrown from $100,000,000 to
$500,000,000 incredibletrajectory.

(02:55):
But how much is left? You know,if you do some Googling or
OpenAI conversating, you'll getvarious estimates as to the
number of software developers inthe market. But that's really
your total addressable marketfor a product like Cursor is
software developers. I've seenanywhere from 20,000,000 to
closer to 50,000,000. For thesake of argument, I'm going to

(03:17):
land on what sort of stuck outas the median estimate for me at
30,000,000 developers and ifevery one of those 30,000,000
developers signed up for Cursorfor $20 a month, plus $20 a
month, $2.40 a year, thatrepresents a total addressable
market of $7,200,000,000.
Cursors obviously got a bigchunk of that already at their

(03:43):
$500,000,000 of ARR but 7.2means there's still a fair
amount out there to grab. Butyou know, the real talk here is
that Cursor is not going to get100% of that market. You know,
let's say that they get 25%. Oneout of every four software
developers on earth ends upusing Cursor on a fully

(04:05):
annualized $20 a monthsubscription. That would be a of
$7,200,000,000, which is1,800,000,000.0, which still
represents more than a X oftheir current revenue.
So, you know, that's a prettygood return. Does that feel
real? Is it likely that one outof every four software

(04:25):
developers is going to end up onCursor? I don't know. There's a
ton of competition.
You've got everything fromReplit and Windsurf, the sort of
peer players in the assistedcoding space, to open source
alternatives like RU Code thathook a language model into
whatever your IDE of choice is.And then there's enterprise

(04:48):
players that are only going tostrike deeper into this space
like Microsoft obviously has avery vested interest in getting
in on this. So, to pull it back,how much growth is left? Total
addressable market of$7,200,000,000. Could Cursor 3x
their current revenue or 4xtheir current revenue and grab

(05:08):
1.5 to $2,000,000,000 of annualrevenue?
Yeah, sure. Possibly. To me,that 25% feels like a big bet.
Know, just so much coming intothis space that it feels like a
big bet. So that's the question,is how much growth is left?
question, that $500,000,000 ofrevenue, what is the quality of

(05:31):
that revenue? Cursor calls itannual recurring revenue, but
their subscriptions are monthly.You can sign up for an annual,
but I'm assuming that that$500,000,000 is literally taking
their monthly recurring revenue,whatever that is, and
multiplying it by 12. I could dothe math and it's, you know,

(05:51):
$40,000,000 or something likethat. But does that ring true?
You know, how what percentage ofCursor's $500,000,000 of revenue
or, you know, their theirmonthly revenue, what percentage
is coming from drive bys? Howmany of those people are going
to stay on? Essentially whatwe're asking is what's the churn

(06:13):
like at Cursor? And I don'tthink we know, I don't know that
I've seen a statistic, know,they've published something
about that. But I think that'san important question because
the quality of that revenuematters.
I think that goes withoutsaying. And it strikes me that
there isn't a ton of barrier toentry to switching. You're on

(06:36):
Cursor today, tomorrow you pickup shop and point Windsurf at
your GitHub repository and youuse Windsurf or whatever the
next flavor du jour is. What isthe sticking power of Cursor? I
don't think anyone knows thatyet.
So that's a question as well. Sohow much growth is left? What's
the quality of the currentrevenue? The last sort of flag

(06:59):
or question that I would ask iswhat are the other pieces of the
puzzle? You know, what othersources of value does an
organization or, you know,product like Cursor create?
So what do we know? We know thatCursor is built on Versus Code,
an open source IDE. So, youknow, it's built on open source.

(07:20):
There's value to be found there,but obviously not the same type
of value lock in that you canget from rolling your own
platform. The other thing isCursor isn't a frontier or a
foundational model provider.
Provider so they're back endedby other companies' models. The
documentation seems to indicatethat it's part Claude or Clode

(07:43):
depending on how you want topronounce it and part OpenAI.
You know, But the model on theback end isn't a source of
unique value for Cursor either.So they don't own the software
on the endpoint. They don't ownthe model on the back end.
What do they own? Arguably, themost important piece, which is
the user experience. So adaptingUX code, sorry, Versus code to

(08:08):
be on AI enabled platform andmaking it feel totally
frictionless and natural forboth a developer and the
ecosystem surrounding thedevelopers, presumably QAs,
enterprise software managers,you know, all the other players
in the ecosystem that would beimpacted by the presence of
something like Cursor. I thinkwhat we've seen so far is that

(08:28):
Cursor is executing really well.They're doing the best job of
anyone in market in terms ofmaking the experience of
becoming an AI equipped supercoder really frictionless and
seamless.
So that's very powerful. Youknow, another proposition out
there is that, you know, maybethings like Cursor and Windsurf
are actually a foundationalmodel play or a frontier model

(08:52):
play. So what these do is, intheory, let Winsorf, Recursor,
whoever, see a ton of code andsee a ton of human created code
so you can sort of tap dancearound the risk of training
models on output from models andavoiding model collapse. So in
theory, these tools are anincredible resource for

(09:15):
gathering all that and in theorysomeone like Cursor could be
planning to create their ownfrontier coding model or, you
know, tuning, you know,something like Clode from
Anthropic to be a superiorcoding model. But the truth of
it is obviously a large amount,if not all of the code that goes
into something like Cursor isgoing to be bound by

(09:38):
intellectual property andprivacy laws and so I think
unwinding all of that and havinguser agreements that are
permissive enough to let youtake proprietary human created
code and bake it into your ownfrontier model, that feels like
a stretch to me as well.
It feels like what Cursor's gotis an incredible user

(10:00):
experience, which carries a lotof value that they've leveraged
into incredible adoption inmarket, which shows up in their
revenue story. There we go. Thebottom line is that growth is
phenomenal, but there's aceiling set by the total
addressable market. That is tosay the number of developers,
the number of developers thatCursor can get vis a vis the

(10:24):
competition, the amount thatthey're willing to pay because
there will be downward pressureon price over time. There's a
ceiling set on the tooladdressable market and the
revenue quality that Cursor isgetting is to be determined as
well.
So when you take all that intoaccount with the fact that
beyond their user experience anddistribution, Cursor doesn't
necessarily have a unique sourceof value that they're creating,

(10:46):
it feels like the participantsin this recent raise are going
to be pushing hard for anotherseries or an exit in the next
twelve to twenty four months towatch that revenue 3x and then
get their value out. So that'sthe bottom line on Cursor. Thank
you for listening to minimumviable banter today and I will
catch you next time. Cheers.