Summary & Insights
anies a ing $100 million in revenue faster than any SaaS company ever did,
g less s and marketing to get there.
grew 69 -over-year in 2025, generating up to $1 million in revenue per employee.
iciency ok. Demand is so strong, these companies can barely keep up.
every t gets plugged in is maxed out immediately.
s. Debt ering the system.
ng hold k enterprise adoption isn’t the tech itself.
organiz to actually change how they work.
genera er David George about what the data shows and why we’re still early.
ith wha nk the big takeaways are from this piece,
first ‘ve ever done this style piece.
work an ch analysis.
nside o eam.
have so ifferent thoughts and points of view.
em on p d share them out with the world?
esis of
om doin
is cra
rowth q of companies in AI is extremely encouraging from our standpoint.
ing to mselves better.
ou some on that, that there’s been some sort of ex-buzz,
ing, ki ebating what’s going on there.
mpanies ld say, is more impressive than prior crops of companies,
he dema their products is so high.
thy rig
to see igns of things that are stretched a little bit.
t we se hat we’re looking out for.
e to be of a lot of these great companies.
ng acti is happening in the private markets,
ppening private markets.
show so es about that.
big co n,
ted abo e we are now,
e are i product cycle.
e our b .
5-year
he very ing of it right now.
l priva es.
t just ur activity.
ticals.
ide, ha most active in AI and infrared apps
also v ive in our other verticals as well.
om thro e of these.
6Z comm
the ch work with some of the best models
compan viously.
data.
and ton ta as a growth team
ally se ery growth stage company in the market
io comp as a prospect.
eat dat sis team.
alysis.
is just interesting.
lusion is is
accele evenue growth,
lowed i 23, 24 following the rate hikes
some o ech stuff.
hat tre
across nt types of companies
Decile tel,
g the o companies, it really accelerated.
seen u his slide on a page before,
wing AI ies are reaching 100 million bucks of revenue
r than test growing SaaS companies in their era.
y impor ing I want to call out about why that is the case.
end cu demand is so strong and the products are so compelling.
ey spen money on sales and marketing.
pposite
es that owing the fastest are not the ones spending the most amount of money
ing.
g less n sales and marketing than their SaaS counterparts.
wing mu h faster.
showin the growth of the AI companies versus the non-AI companies.
he AI c s are growing two and a half times plus faster than the non-AI companies.
be a hu rise.
compani growing very, very fast.
eck thi when we saw the AI top performers growing 693% year over year.
ur expe and anecdotes that we see from the portfolio companies.
profile e’re seeing in the data set.
e inter a sets that we have of portfolio companies
we look potential investments.
little rse for AI companies.
rd us t ut this before.
el like oss margins for AI companies are sort of a badge of honor
e want if low gross margins are a result of high inference costs.
ple are AI features.
belief ose inference costs over time are going to come down.
f we se pitch and the gross margins are super high,
skeptic use that may mean that the AI features are not actually what is being bought
omers.
about FTE.
hing th e started focusing on.
the thi t got a lot of pickup and discussion on X in the last few days.
of a m of the efficiency of how you run your company in general.
all of sts.
just y es and marketing, which is an efficiency measure that we’ve always kind of looked at
in the but it also captures your overhead.
D.
AI com they’re running at $500,000 to a million dollars per FTE.
mb for s software businesses in the SaaS era was like $400,000 in the last generation.
talk a is a little bit more, but the reason why this is the case is mostly because demand is very, very strong for their products.
less re to go take it to market.
k clari ust before we go to this slide here.
e AI co ?
post-Ch versus historical AI ML companies founded by a certain time period?
post-Ch companies.
re foun e right around that time.
bit of
product ket was an AI native product, then that’s how we divide it.
is a g nt, but where you can punt until later.
tions I a lot of folks are trying to understand is the magnitude of change and expected revenue and growth from companies from the SaaS era to AI era companies.
little out the magnitude of revenue, et cetera.
those e not AI native?
rd time ing against AI native companies?
ng?
llout?
e think ut their historical portfolio?
re appr this with our portfolio is you need to adapt to the AI era or die.
on the nd and the back end.
, you n think about how you can incorporate AI into your product natively.
a chat into your existing workflow, but reimagine what it can mean with AI and be aggressive about disrupting yourself and changing.
k end, d some stats around the efficiency that the companies are running at.
ange to
be full d out with the latest coding models for all of your developers and all of the latest tools across every different function inside your organization.
has bee ding so far, and that’s where we’ve seen the biggest leaps.
or, maj ges, like in the last two months on this, like month and a half in this.
s writt t this.
with o ur sort of pre-AI companies.
er who’ AI, like he’s very AI deep.
g his c
s week, told me that he was frustrated with one of their products.
two en that are very deep in AI and assigned them to build it from scratch with quad code and codex and cursor and just they had unlimited budget on coding tools.
ks it’s somewhere between 10 and 20x faster than progress that they had before.
they ha ciated with that is actually, they’re high enough that it will cause him to rethink what his entire organization will look like.
basica need my entire product and engineering organization working this way, and I think it’s going to happen within the next 12 months.
mean fo the team design actually is?
uct sta where does ing start, and even where does design start in that process?
cember t of a turning point on code.
ths, it er going to hit and take hold in companies, or those companies, I think, are going to be moving much slower than their peers.
the pr mpanies, Adapt, we have another example of a company that is a pre-AI software company, and the CEO has gotten totally AI-pilled, and he’s like, we’re going to become an AI product.
oyees a your AI agents.
you hav
s that lking about.
that w extreme about it, and he said, I now ask the question, for every task that we now need to complete, can I do it with electricity, or do I need to do it with blood?
treme m shift that’s happening, you know, with our companies.
see th pre-AI companies are moving very fast and trying to adapt, but they very much need to adapt to this new era, both front-end product-wise and back-end, how they run their companies.
aybe ta y, almost every portfolio, you have to go line by line on the company to understand where the founder is on that journey and how much they’re implementing from the ground up.
you sa erms of blowing up existing operations, that’s also happening in post-AI companies, too.
eople a looking every six months.
s we bu months ago could be vastly improved based on what is available today.
continu ppening, the pre-AI companies are needing to increasingly 10x catch up to that point.
for th I companies is the business model evolution is still early days.
ive thi can happen to you is a technology and product shift and also a business model shift at the same time.
I thin e business models as like a spectrum.
ut like rise, like B2B, just to keep it simple.
basica enses.
he pre- icense and maintenance business models.
nd subs n, and that was typically seed based.
innovat d it was very disruptive.
re and elivery was disruptive, but the business model change was very disruptive.
t what d to Adobe as they went through that transition.
transit consumption based, so usage based.
clouds , and so many of the sort of volume-based, like task-based type businesses have already adapted that and shifted to that from, you know, seed based to consumption.
teratio be outcome-based.
you do you know, and ideally when you successfully complete a task, you get paid based on the successful completion of that task.
that’s possible today to pull off is probably customer support, customer success, because you can kind of objectively measure the resolution of something.
happens he capabilities of the models to the extent that other functions besides customer support can measure those kinds of outcomes.
e disru orce for incumbents.
to con n might be a big disruption if the composition of companies changes as well.
s the r ig one.
ersus e ity, we should go to ARR over FTE.
hat was on on this one on the next slide was like, oh my gosh, look at the AI efficiency gains that are happening in the market.
le bit in this, like companies running themselves a little bit differently.
take th le that I gave about, you know, the two engineers who are rebuilding the product.
rvation ur companies, even the AI native ones, is they run leaner, partially because they’ve just grown so quickly and the demand is so strong.
we’re a oint where companies have fully reimagined the way they run themselves.
ittle b result of our data set being the best of the best companies and demand signals for those being extremely high.
they ha resources to serve that demand.
ow, eff , general efficiency gains that have happened in the technology market, you know, out of the kind of 2021 most, you know, most bloated era.
o see s ly signs of that efficiency, but the wholesale run your company totally differently.
we’re k early in that journey.
one th seen is in the public markets that anyone can go read about is probably Shopify, where they, you know, Toby’s awesome.
at’s cl
our gro stuff.
job.
e fully ed this a couple of years ago.
staff actually wrote this whole big deep dive on how Shopify AI-ified itself, you know, in terms of, you know, employee direction, process, et cetera.
bably s ng the surface of what’s going to happen over the next five years.
ext sec what are these companies actually doing in our favorite topic, which is lawyers have only increased in this new world of AI’s meeting lawyers.
saw it r this week that a corporate lawyer was quoted saying, LLMs have actually increased my workload because every client thinks they’re a lawyer now.
o Harve h is an excellent.
good.
t.
real t me because, you know, I love talking about our portfolio companies.
go thr is section quickly because, you know, I think people, people know these companies, hopefully.
s one, w, one of the big things that we look for and one of the questions I think that came in was, how do you know that revenue is going to be sustainable?
s, they ew really, really fast.
hat we rselves to do is make sure we go super, super deep on revenue retention, renewals, and product engagement, actually time spent.
e loggi the platform?
platfo t does their activity look like?
this p with the onset of much better product that they’ve built over the last couple of years, plus the improvement of reasoning models, it turns out lawyering and reasoning go hand in hand.
about d he amount in the product as they had before.
t AI is good at lawyering.
fewer l but I think, you know, AI is very, very good at this.
are ge lot more efficient.
thing a lates to Harvey is they’re just spending a lot of time in the product and getting a lot of value out of it, which is great.
e.
to kee ng about lawyer.
g to ma mment.
ars tha known you, I wouldn’t have ever discerned that you’re from Kentucky, other than this moment now, by the way you say lawyer.
those n my vocabulary that I don’t, you know, my wife always jokes.
w, you , you have like one bourbon, and then you talk like you probably did.
ut when e to lawyers.
bourbon wn.
on.
ons.
one th per, super exciting.
e the d rave about getting to have access to a bridge and how much time it saves them and how much, you know, better it makes their lives.
f the c s that we talked to described it like a trusted deputy.
ght sho thing we look for, which is the blue line shows the growth in users and the green line shows the engagement of those users.
massiv wn the number of users, you’d be a little worried if engagement of those incremental users that they were adding was going down.
ave ext high usage among the people who use the product.
ly held and grown a little bit, even as they’ve added tons and tons of more users.
xamples kind of data that we look for to make sure that we feel confident that the revenue these companies are generating is sustainable.
mpanies owing faster than, you know, any of the predecessor companies, but it’s very sustainable.
s high ent, it’s high retention.
ly impo or us.
Labs.
piece o ny of the new AI tools.
about c support on the B2B side, but, you know, so much, you know, other personal tools, business tools, you know, start with voice.
the th t I love to look at on this chart.
g.
growin fast.
ample o f these companies that runs extremely efficiently.
y a gre
ne.
this i ferent example.
a good e of what I was describing earlier.
to this now, AI shift.
t of ef king sure that they could take the most of the AI capabilities and make their business better.
way you e it in their business today is in the handling of resolutions.
y have know, agents that have to handle travel bookings or travel changes.
50% of ser interactions.
uff.
bookin
travel.
know, , like tell me the balance of my bank.
ike com rkflow that AI is now able to handle.
t in th ess is a 20 percentage point expansion of gross margins over the last three years.
eptiona t.
ou need pt or die.
tors ar dapting.
hool.
, they’ sitting still and doing things the old way,
rcentag higher gross margins than those incumbents.
Flock, is doing absolutely incredible work.
hem so
lling c value proposition that we see in our portfolio,
ROI is, ving crime.
covered , each year’s Flock is solving 700,000 crimes.
he righ is a data point that just shows per officer that where there’s Flock,
most 10 know, more crimes.
he comm
e a gre know, they have a great business and financial model that goes along with it.
heir pr r from their product is exceptional.
know i ee the chat lighting up of people saying that they’re three bourbons deep.
.
on abou do you think about the benchmark?
o think traditional industries like finance, for example, and using JP Morgan as a benchmark,
brate t une 500 in terms of AI adoption?
overla question that Xavier mentioned as well with, you know, there was that study about enterprise adoption from MIT at the early outset of last year.
ring al of wonky things.
bit mor how and what you’re hearing from Fortune 500 CEOs.
from Fo 00 CEOs, I would say is, and maybe this is the key sort of link between those two points.
from Fo 00 CEOs is we have to adapt.
rstand tools we need.
dy to c
usiness going to fully roll things out.
e ready
me AI c s.
ent tha is actually happening.
gest di t of sort of, you know, that mindset compared to actual change in the businesses is just change management is hard.
enough people to just use an AI assistant to help them do their jobs better.
probab easiest one to get people’s minds wrapped around customer support.
faster er, obvious thing.
ually, w, general management of businesses, changing business processes, change management, it’s extremely hard to do.
rised t re are anecdotes out there that suggest, oh, you know, things are moving slower than expected.
mpanies re fully embracing it and actually know what to do, it has tremendous business impact already.
nk ther ng to be a sort of reckoning over the next five years of who can actually embrace change, push through change management, you know, adopt all the best products and those that don’t.
l be ma ferences in productivity, you know, we have some charts later in the slides, you know, which I can talk to.
expecta round productivity enhancements and, you know, and growth and all that stuff, you know, the expectations are high.
of com will achieve those and the ones that don’t are going to be at a huge disadvantage.
uced th port costs by 60 percent.
d that ved 1.1 million hours in underwriting up to 6x year over year.
lion bu run rate annual savings.
ckets o non-AI businesses.
going really interesting year to watch over the next 12 months.
g to se more anecdotes.
ompanie can figure it out and there are going to be companies that don’t.
these c ions have had to orient their business to be ready for AI as well.
ersion , like, using a chatbot, right, and how much productivity gains that actually gets you.
right?
actuall etely upend your system’s information and back end to be ready for AI, a lot of that is probably latent and being built up now into actually seeing the outcomes associated with it.
ing the markets.
most 80 t of the S&P 500’s return.
the ma ng driving the economy and the stock market.
doing v l.
s are s
going u ou know, there’s some blips like the last couple of days.
ly doin
s are v nd.
eviden roth is minimal.
nce is by UPS growth.
racted y, maybe more than slightly, if you’re a SaaS company over the last few days or a couple weeks.
market ced on, in general, earnings and earnings growth.
ltiples gher than average, but nowhere near the dot-com.
t look charts and see where we are.
gives comfort.
nings o ompanies that are the biggest drivers of the market in general, I feel like are pretty sound.
ood.
ealth o companies, I would say, is pretty good.
are hig n average in the past, but they don’t feel super alarming.
ding te anies that I was just talking about are the best businesses in the history of the world.
er a lo od of time, they have shown margin improvement that suggests that is probably true.
w, that he left side of the page.
aying f its, not loss-making growth.
ntrast , 22 era, sort of 21 era.
ig cont om a dot-com.
s, mult re not that high.
ike, su , you know, five slides worth of materials.
than i een in the past.
ow, the gh expectations for a reason, and we’re optimistic about the impact of AI flowing through to earnings, you know, overall in the public markets in the coming years.
your a n on the right side, which is, you know, if you just took a four box of, like, low growth, high growth, low margin, high margin, and paired up those types of companies, this is a chart that shows how they trade.
or the mpanies.
the tw ns on the right is high growth, high margin companies, and then high growth and low margin companies.
iously wth, low margin.
should rewarded.
ow.
hat are rowth and high margin, and, you know, the high growth and low margin, as long as they have good unit economics and they’re scaling into their margins, they should be rewarded.
is goo
growth, f you’re high margin, it’s tough out there.
rising.
about t the past in many different forms.
wth is gest thing that drives returns over five to 10 years.
r me to gh growth is rewarded more than low growth.
h growt igh margin, you’re one of those great businesses, it’s being very rewarded.
we’re g talk about supply side of the CapEx buildout.
assive.
ncentra the investment is inherently risky, just given how big it is.
ubbly f , the underlying fundamentals, I would say, bear little resemblance to previous bubbles.
inanced ily by historically profitable companies, like very profitable companies that I had talked about.
enter ture.
celerat ch is good.
l, we’r ly monitoring the sort of cost of training and the economics of that whole equation.
pretty
e big m mpanies that spend money on training models is pretty good.
g that .
e think I is going to be, you know, the biggest model buster that I’ve seen in my career, certainly.
model b so I won’t spend too much time on them.
es that aster and longer than anyone would have modeled in any scenario.
classic f this.
ke cons odels from pre-iPhone to five years later, four years later, consensus models were off for Apple’s performance by a factor of 3x over four years.
e most company in the world at the time.
nk that me thing is going to happen in many pockets of AI where the performance just massively exceeds, you know, what any expectations in a spreadsheet would show you.
is itse del buster.
h has d d high margin revenue at unprecedented speed and scale.
xpensiv , but repeatedly surprised to the upside, I would say, and creates value, I would say, far in excess of the capital that’s required to grow.
n to th ll be different, you know, this time around.
dot com is actually supported by cash flows and CapEx as a percentage of revenue is considerably lower.
headlin an zoom to the next slide, but, you know, I feel much better about this CapEx, you know, dynamic than dot com, obviously.
e ones bearing the biggest brunt of the CapEx, and this is a very good thing.
ortfoli nies, this is great.
it.
as much ty in the ground, get as much supply as you possibly can on the ground for training and inference.
thing.
anies t bearing most of the brunt of this are the best businesses of all time that I had talked about before.
e’re st to monitor is the introduction of debt into the equation.
e all o orecast CapEx that’s to come with cash flow, and we’re starting to see some debt.
this cl
invest ily in companies with exposure to debt.
le with h of the companies on the page financing with cash flow, continuing to produce cash flow, and using debt even?
rosoft, VIDIA as counterparties?
reat ab t.
s I fee about.
about a hem.
arties same.
rting t rivate credit get a little bit more involved in the data center build out.
n, the that’s very well covered that is kind of making a bet the company move into becoming a cloud is Oracle.
ow, the en profitable forever and reducing their shares forever.
apital ey are committing is very large.
cash f ative for many years to come.
ou foll of the buzz around it, like the cost of their credit default swaps has gone up, you know, to like 2% over the last three months.
ng stuf this.
general stuff for our portfolio companies, but we want to make sure that the market overall is healthy as well.
lide th s the magnitude of the pace of change of AI.
ld out revenue to what happened with Azure.
s comin relative to the cloud.
years h one year of AI revenue.
rosoft ng data, which I think is a cool way to frame how quickly this has happened.
‘s take y long time.
d out i ning much faster.
s for A venue to surpass their CapEx.
think rt of ratio or equation is going to happen much faster with AI.
ek out h on depreciation, but this is one of the topics that gets a lot of buzz in finance circles.
are yo mptions around depreciation of chips in particular?
cing fo GPUs is very solid.
ith mod it longer, but later users quickly switch to the new thing.
side.
the mo e.
even to year-old TPUs, Google actually disclosed this, seven to eight-year-old TPUs actually have 100% utilization.
monito rice of chips in the secondary market.
nt A100 100s has actually held up very, very well.
s of ch still getting fully utilized.
thing I about yet, but it gets a lot of buzz and, you know, sort of alarmists who like to talk about risk in the system.
itive s
at we t ut all the time is this paradox, right?
cheape umption goes up.
s repor d is well in excess of supply.
they s .
ewed Ga er, a friend of mine at our AI summit, and he was comparing the buildout of the internet and laying all the fiber to the buildout of data centers here.
big lin here is, you know, there is no dark GPU.
PUs.
ber.
r and t u know, it laid there dark and it wasn’t used.
the sy a data center, it gets fully utilized immediately.
y good ou know, in terms of, you know, demand meeting supply immediately.
rlier.
ld come ese companies.
xpectat
then th probably be disrupted if they can’t change.
t, agai he biggest reason why we see things, you know, that haven’t sort of dramatically shifted yet.
‘s not diness of the technology itself.
know, p buildout that needs to get built around the technologies and then change management and putting it in production.
as scal staggering clip relative to other categories.
shows ckly generative AI in app revenue has grown from 23, where it was basically, you know, you can barely even see it on the page to now.
ow, you see it on the page to now, you can’t see it on the page to now, but you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page.
the pa ow, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page.
the pa ow, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now.
sume 20 ns and a 22 times PE, that translates into 35 trillion of new market cap. There’s been about 24 trillion of new market cap that’s been pulled forward. Now we can debate if that’s attributable all to AI or otherwise, you know, large tech performance. But there’s still a lot of sort of market cap to go get, where you could have upside if, you know, if those assumptions are right.
sort of few cuts on trying to address this sort of AI payback question. So current estimates put cumulative hyperscaler CapEx at a little less than 5 trillion by 2030. So if you do napkin math on that to achieve a 10% hurdle rate on that 4.8 trillion or almost 5 trillion of investment, annual AI revenue would have to hit about a trillion dollars by 2030. So to put that into context, a trillion dollars, that would be
GDP to e a 10% return. It’s possible that happens. It’s also possible we could fall a little short of that. But I think it’s limiting just to look to 2030. I think the payback of this probably happens, you know, over a longer period of time, like, you know, between 2030 and 2040 as well.
ing it t’s about, you know, 1% GDP to get to, to get to the payback number of a 10% hurdle rate. All right. Heard it on the street. What we’ve started to do is we’ve sort of built software to track what all of the AI or what all of the tech public technology companies discuss in their earnings calls.
how re it is to our business at the early stage and, you know, the growth stage. And we package it all up and we share it out to our CEOs. So, you know, they can kind of have a simple digestible format of like, what do I need to know about AI as it relates to the public technology companies? You know, how does it, how does it impact my business, et cetera. And so we shared a bunch of the, you know, the stuff that we, that we track in here.
one que efore we moved to the private section, which a lot of folks, of course, on this call care about in this transition here. But before we get to that, so where are we calibrating to your trillion dollar in AI revenue, you know, thereabouts in 2030? Where are we today relative to your guesstimate of AI enabled revenue? And how far off are we to that trillion dollar number?
he, I w obably guess in the 50 billion range.
And th o perfect way to do it. I mean, I know, I know some of the big inputs to it. The harder stuff to track is, honestly, the big tech companies, like how much real AI revenue do they have?
ds can , they will from time to time give percentage uplift from AI. But I think depending on how they want to paint the picture, they can play games with that a little bit. So, you know, I think it’s, I think it’s, I think it’s, that’s a, that’s a rough swag. But like, you know, trillion, we’re probably at 50, but it’s growing, you know, way, way, way faster than a hundred percent year over year.
bly tha ue, I mean, ChatGPT launched three years ago, but substantially most of this traction happened in the last year and a half-ish or so, if we’re being really generous too. Is that a fair characterization?
Yeah. k, you know, it’s not just ChatGPT now on the consumer side, you know, Google has a business, XAI has a business. And then, you know, on the B2B side, you know, not only do the big model companies all have large API businesses, but the clouds have it too. And so a lot of the, you know, a lot of the sales that are model sales are also flowing through the clouds.
Yep. O ol. We have some questions on, on the private company side, but I’ll let you get through this section and then I’ll tee you up for it.
appy to o questions if you want on it. I mean, this, a lot of the stuff that we’ve talked about, you know, the big themes for me on the private market side, you know, companies are obviously staying private longer, but this is such a real asset class now.
ars, th r of public companies has been cut in half. Um, you know, the, the vast majority of companies that are a hundred million dollar plus revenue companies are private, something like 86%. Um, so, you know, that’s, that’s a major shift. Um, we can go, you can skip a couple of slides forward. Uh, basically I’ll talk a little bit about power laws. Cause that’s, I think that’s interesting.
f that n’t talked about as much, but value very much concentrates in the outlier companies. So the collective valuation of North American and European unicorns is about five and a half trillion dollars. The 10 largest ones, if you just take those, um, comprise almost 40% of the entire value. So, and that’s actually doubled since 2020. So sort of value, you know, sort of value is being, uh, concentrated in the biggest and best winners.
real t have four, five, six, seven of the 10, our portfolio companies of that 10. Um, so, you know, we, we’ve got a reasonable amount of coverage on that. Power laws are happening in the public markets too. So large cap has tripled since 2019. So what, what constitutes a large cap company has actually tripled since 2019. And I think this, the chart on the right side is super interesting. This was new data analysis that we had done.
you lo he lifespan of an average company on the S and P 500, that’s what that chart shows. That’s what the numbers represent. The light, like once a company is on the S and P 500, how long is it on there?
s actua actually, if you look over the last 50 years, that has declined by 40%. The amount of time it stays as part of the S and P 500. So disruption to companies happens faster and faster and faster, which I think is a very interesting dynamic and, and sort of matches what we’re seeing, you know, just in terms of like speed of change in, in, in the markets driven by technology.
o talk ower laws in our business too. I didn’t choose the title of this slide. Uh, I recognize all of the, you know, questions and concerns about it. Um, so the, the volatility laundering thing is, is a, is a big debate in our circles too. Um, mostly around founders who are trying to debate the merits of the private markets and the public markets. And,
the Co s did an interview where, maybe it was John, uh, did an interview where he talked about, you know, managing your stock price and avoiding volatility. And you can kind of orderly fashion, bring your stock price up over time. And that makes it easier to retain employees, hire employees, manage morale, uh, et cetera, et cetera. Um, and so I get the merits of that. I also think there are really, really strong merits of being a public company as well.
to hav lly, really interesting 18 months where we’re going to have some of the big kind of private for a very long time companies that go public. Um, and that’s a good thing in my opinion too. Um, some of the stuff that we show in this chart is just volatility and the observation that over time volatility has gotten a little bit more extreme in the markets. To me, this is a little bit cycle driven too. I know it’s short, short duration is sort of what we’re measuring. Um, but there’s merits to both.
uch lar the private side. We have embraced that new reality. I think it’s, it’s been a big benefit to our business in terms of getting content, getting to continue to invest in these companies over time. Uh, but obviously, you know, there’s, there’s a path of being a public company and getting liquidity, which we care a lot about too.
um, th e two questions. Uh, I will queue up for you here. Uh, one on Databricks. Can you talk about their transition, uh, from being a pre AI company now to a fully embedded AI company and what that’s been like?
all, I ou need to, you know, I mentioned Toby, like the reason Shopify has embraced it is because Toby has led from the top and he runs the business, you know, with AI at the center and, and he, he sort of performance manages everyone, uh, to, you know, to make sure that they do that. Ali is the same. Ali is this unique blend of, um, sort of commercial kind of terminator. I talk about him. He’s called him the technical terminator.
r. You have a commercial instinct and understand the importance of the value creation opportunity in AI. And then you need to actually be deep enough in the technology to know what to build. And so it just so happens that their, um, their sort of cloud data warehouse, or they call it the data lake, um, is actually a great way to have your data in a place to run AI workloads on top of it. So, you know, that was sort of a good place to be for them.
ry aggr y iterated on new AI products. They have this, uh, new product called agent bricks, which we’re super, super excited about. We think it’s going to be really big and transformative for them. So, um, I would say that’s a piece of it. And then they have the big AI native companies all as customers. And so, you know, they have the technology, they have the low cost technology. Um, and so, you know, a big thing that we look for when we’re making investments in companies is who are their customers.
fer the ers of our portfolio companies to be the modern thinking ones, you know, the door dashes of the world, um, you know, the Instacarts of the world, the Ubers of the world, then the very, very old school stodgy companies. Cause that means that their technology is evaluated by smart technologists and they pick it.
edge AI ies are all building on top of data bricks. Uh, and so, you know, they have the chance to grow with them as they scale. Uh, but it’s also a really good, you know, validator that they have the right technology.
e. Than David, for taking us through that.
ncludin k to our investments, please see a16z.com forward slash disclosures.
lease s .com forward slash disclosures.
lease s .com forward slash disclosures.
lease s .com forward slash disclosures.
lease s .com forward slash disclosures.
lease s .com forward slash disclosures.
lease s .com forward slash disclosures.
lease s .com forward slash disclosures.
g less s and marketing to get there.
grew 69 -over-year in 2025, generating up to $1 million in revenue per employee.
iciency ok. Demand is so strong, these companies can barely keep up.
every t gets plugged in is maxed out immediately.
s. Debt ering the system.
ng hold k enterprise adoption isn’t the tech itself.
organiz to actually change how they work.
genera er David George about what the data shows and why we’re still early.
ith wha nk the big takeaways are from this piece,
first ‘ve ever done this style piece.
work an ch analysis.
nside o eam.
have so ifferent thoughts and points of view.
em on p d share them out with the world?
esis of
om doin
is cra
rowth q of companies in AI is extremely encouraging from our standpoint.
ing to mselves better.
ou some on that, that there’s been some sort of ex-buzz,
ing, ki ebating what’s going on there.
mpanies ld say, is more impressive than prior crops of companies,
he dema their products is so high.
thy rig
to see igns of things that are stretched a little bit.
t we se hat we’re looking out for.
e to be of a lot of these great companies.
ng acti is happening in the private markets,
ppening private markets.
show so es about that.
big co n,
ted abo e we are now,
e are i product cycle.
e our b .
5-year
he very ing of it right now.
l priva es.
t just ur activity.
ticals.
ide, ha most active in AI and infrared apps
also v ive in our other verticals as well.
om thro e of these.
6Z comm
the ch work with some of the best models
compan viously.
data.
and ton ta as a growth team
ally se ery growth stage company in the market
io comp as a prospect.
eat dat sis team.
alysis.
is just interesting.
lusion is is
accele evenue growth,
lowed i 23, 24 following the rate hikes
some o ech stuff.
hat tre
across nt types of companies
Decile tel,
g the o companies, it really accelerated.
seen u his slide on a page before,
wing AI ies are reaching 100 million bucks of revenue
r than test growing SaaS companies in their era.
y impor ing I want to call out about why that is the case.
end cu demand is so strong and the products are so compelling.
ey spen money on sales and marketing.
pposite
es that owing the fastest are not the ones spending the most amount of money
ing.
g less n sales and marketing than their SaaS counterparts.
wing mu h faster.
showin the growth of the AI companies versus the non-AI companies.
he AI c s are growing two and a half times plus faster than the non-AI companies.
be a hu rise.
compani growing very, very fast.
eck thi when we saw the AI top performers growing 693% year over year.
ur expe and anecdotes that we see from the portfolio companies.
profile e’re seeing in the data set.
e inter a sets that we have of portfolio companies
we look potential investments.
little rse for AI companies.
rd us t ut this before.
el like oss margins for AI companies are sort of a badge of honor
e want if low gross margins are a result of high inference costs.
ple are AI features.
belief ose inference costs over time are going to come down.
f we se pitch and the gross margins are super high,
skeptic use that may mean that the AI features are not actually what is being bought
omers.
about FTE.
hing th e started focusing on.
the thi t got a lot of pickup and discussion on X in the last few days.
of a m of the efficiency of how you run your company in general.
all of sts.
just y es and marketing, which is an efficiency measure that we’ve always kind of looked at
in the but it also captures your overhead.
D.
AI com they’re running at $500,000 to a million dollars per FTE.
mb for s software businesses in the SaaS era was like $400,000 in the last generation.
talk a is a little bit more, but the reason why this is the case is mostly because demand is very, very strong for their products.
less re to go take it to market.
k clari ust before we go to this slide here.
e AI co ?
post-Ch versus historical AI ML companies founded by a certain time period?
post-Ch companies.
re foun e right around that time.
bit of
product ket was an AI native product, then that’s how we divide it.
is a g nt, but where you can punt until later.
tions I a lot of folks are trying to understand is the magnitude of change and expected revenue and growth from companies from the SaaS era to AI era companies.
little out the magnitude of revenue, et cetera.
those e not AI native?
rd time ing against AI native companies?
ng?
llout?
e think ut their historical portfolio?
re appr this with our portfolio is you need to adapt to the AI era or die.
on the nd and the back end.
, you n think about how you can incorporate AI into your product natively.
a chat into your existing workflow, but reimagine what it can mean with AI and be aggressive about disrupting yourself and changing.
k end, d some stats around the efficiency that the companies are running at.
ange to
be full d out with the latest coding models for all of your developers and all of the latest tools across every different function inside your organization.
has bee ding so far, and that’s where we’ve seen the biggest leaps.
or, maj ges, like in the last two months on this, like month and a half in this.
s writt t this.
with o ur sort of pre-AI companies.
er who’ AI, like he’s very AI deep.
g his c
s week, told me that he was frustrated with one of their products.
two en that are very deep in AI and assigned them to build it from scratch with quad code and codex and cursor and just they had unlimited budget on coding tools.
ks it’s somewhere between 10 and 20x faster than progress that they had before.
they ha ciated with that is actually, they’re high enough that it will cause him to rethink what his entire organization will look like.
basica need my entire product and engineering organization working this way, and I think it’s going to happen within the next 12 months.
mean fo the team design actually is?
uct sta where does ing start, and even where does design start in that process?
cember t of a turning point on code.
ths, it er going to hit and take hold in companies, or those companies, I think, are going to be moving much slower than their peers.
the pr mpanies, Adapt, we have another example of a company that is a pre-AI software company, and the CEO has gotten totally AI-pilled, and he’s like, we’re going to become an AI product.
oyees a your AI agents.
you hav
s that lking about.
that w extreme about it, and he said, I now ask the question, for every task that we now need to complete, can I do it with electricity, or do I need to do it with blood?
treme m shift that’s happening, you know, with our companies.
see th pre-AI companies are moving very fast and trying to adapt, but they very much need to adapt to this new era, both front-end product-wise and back-end, how they run their companies.
aybe ta y, almost every portfolio, you have to go line by line on the company to understand where the founder is on that journey and how much they’re implementing from the ground up.
you sa erms of blowing up existing operations, that’s also happening in post-AI companies, too.
eople a looking every six months.
s we bu months ago could be vastly improved based on what is available today.
continu ppening, the pre-AI companies are needing to increasingly 10x catch up to that point.
for th I companies is the business model evolution is still early days.
ive thi can happen to you is a technology and product shift and also a business model shift at the same time.
I thin e business models as like a spectrum.
ut like rise, like B2B, just to keep it simple.
basica enses.
he pre- icense and maintenance business models.
nd subs n, and that was typically seed based.
innovat d it was very disruptive.
re and elivery was disruptive, but the business model change was very disruptive.
t what d to Adobe as they went through that transition.
transit consumption based, so usage based.
clouds , and so many of the sort of volume-based, like task-based type businesses have already adapted that and shifted to that from, you know, seed based to consumption.
teratio be outcome-based.
you do you know, and ideally when you successfully complete a task, you get paid based on the successful completion of that task.
that’s possible today to pull off is probably customer support, customer success, because you can kind of objectively measure the resolution of something.
happens he capabilities of the models to the extent that other functions besides customer support can measure those kinds of outcomes.
e disru orce for incumbents.
to con n might be a big disruption if the composition of companies changes as well.
s the r ig one.
ersus e ity, we should go to ARR over FTE.
hat was on on this one on the next slide was like, oh my gosh, look at the AI efficiency gains that are happening in the market.
le bit in this, like companies running themselves a little bit differently.
take th le that I gave about, you know, the two engineers who are rebuilding the product.
rvation ur companies, even the AI native ones, is they run leaner, partially because they’ve just grown so quickly and the demand is so strong.
we’re a oint where companies have fully reimagined the way they run themselves.
ittle b result of our data set being the best of the best companies and demand signals for those being extremely high.
they ha resources to serve that demand.
ow, eff , general efficiency gains that have happened in the technology market, you know, out of the kind of 2021 most, you know, most bloated era.
o see s ly signs of that efficiency, but the wholesale run your company totally differently.
we’re k early in that journey.
one th seen is in the public markets that anyone can go read about is probably Shopify, where they, you know, Toby’s awesome.
at’s cl
our gro stuff.
job.
e fully ed this a couple of years ago.
staff actually wrote this whole big deep dive on how Shopify AI-ified itself, you know, in terms of, you know, employee direction, process, et cetera.
bably s ng the surface of what’s going to happen over the next five years.
ext sec what are these companies actually doing in our favorite topic, which is lawyers have only increased in this new world of AI’s meeting lawyers.
saw it r this week that a corporate lawyer was quoted saying, LLMs have actually increased my workload because every client thinks they’re a lawyer now.
o Harve h is an excellent.
good.
t.
real t me because, you know, I love talking about our portfolio companies.
go thr is section quickly because, you know, I think people, people know these companies, hopefully.
s one, w, one of the big things that we look for and one of the questions I think that came in was, how do you know that revenue is going to be sustainable?
s, they ew really, really fast.
hat we rselves to do is make sure we go super, super deep on revenue retention, renewals, and product engagement, actually time spent.
e loggi the platform?
platfo t does their activity look like?
this p with the onset of much better product that they’ve built over the last couple of years, plus the improvement of reasoning models, it turns out lawyering and reasoning go hand in hand.
about d he amount in the product as they had before.
t AI is good at lawyering.
fewer l but I think, you know, AI is very, very good at this.
are ge lot more efficient.
thing a lates to Harvey is they’re just spending a lot of time in the product and getting a lot of value out of it, which is great.
e.
to kee ng about lawyer.
g to ma mment.
ars tha known you, I wouldn’t have ever discerned that you’re from Kentucky, other than this moment now, by the way you say lawyer.
those n my vocabulary that I don’t, you know, my wife always jokes.
w, you , you have like one bourbon, and then you talk like you probably did.
ut when e to lawyers.
bourbon wn.
on.
ons.
one th per, super exciting.
e the d rave about getting to have access to a bridge and how much time it saves them and how much, you know, better it makes their lives.
f the c s that we talked to described it like a trusted deputy.
ght sho thing we look for, which is the blue line shows the growth in users and the green line shows the engagement of those users.
massiv wn the number of users, you’d be a little worried if engagement of those incremental users that they were adding was going down.
ave ext high usage among the people who use the product.
ly held and grown a little bit, even as they’ve added tons and tons of more users.
xamples kind of data that we look for to make sure that we feel confident that the revenue these companies are generating is sustainable.
mpanies owing faster than, you know, any of the predecessor companies, but it’s very sustainable.
s high ent, it’s high retention.
ly impo or us.
Labs.
piece o ny of the new AI tools.
about c support on the B2B side, but, you know, so much, you know, other personal tools, business tools, you know, start with voice.
the th t I love to look at on this chart.
g.
growin fast.
ample o f these companies that runs extremely efficiently.
y a gre
ne.
this i ferent example.
a good e of what I was describing earlier.
to this now, AI shift.
t of ef king sure that they could take the most of the AI capabilities and make their business better.
way you e it in their business today is in the handling of resolutions.
y have know, agents that have to handle travel bookings or travel changes.
50% of ser interactions.
uff.
bookin
travel.
know, , like tell me the balance of my bank.
ike com rkflow that AI is now able to handle.
t in th ess is a 20 percentage point expansion of gross margins over the last three years.
eptiona t.
ou need pt or die.
tors ar dapting.
hool.
, they’ sitting still and doing things the old way,
rcentag higher gross margins than those incumbents.
Flock, is doing absolutely incredible work.
hem so
lling c value proposition that we see in our portfolio,
ROI is, ving crime.
covered , each year’s Flock is solving 700,000 crimes.
he righ is a data point that just shows per officer that where there’s Flock,
most 10 know, more crimes.
he comm
e a gre know, they have a great business and financial model that goes along with it.
heir pr r from their product is exceptional.
know i ee the chat lighting up of people saying that they’re three bourbons deep.
.
on abou do you think about the benchmark?
o think traditional industries like finance, for example, and using JP Morgan as a benchmark,
brate t une 500 in terms of AI adoption?
overla question that Xavier mentioned as well with, you know, there was that study about enterprise adoption from MIT at the early outset of last year.
ring al of wonky things.
bit mor how and what you’re hearing from Fortune 500 CEOs.
from Fo 00 CEOs, I would say is, and maybe this is the key sort of link between those two points.
from Fo 00 CEOs is we have to adapt.
rstand tools we need.
dy to c
usiness going to fully roll things out.
e ready
me AI c s.
ent tha is actually happening.
gest di t of sort of, you know, that mindset compared to actual change in the businesses is just change management is hard.
enough people to just use an AI assistant to help them do their jobs better.
probab easiest one to get people’s minds wrapped around customer support.
faster er, obvious thing.
ually, w, general management of businesses, changing business processes, change management, it’s extremely hard to do.
rised t re are anecdotes out there that suggest, oh, you know, things are moving slower than expected.
mpanies re fully embracing it and actually know what to do, it has tremendous business impact already.
nk ther ng to be a sort of reckoning over the next five years of who can actually embrace change, push through change management, you know, adopt all the best products and those that don’t.
l be ma ferences in productivity, you know, we have some charts later in the slides, you know, which I can talk to.
expecta round productivity enhancements and, you know, and growth and all that stuff, you know, the expectations are high.
of com will achieve those and the ones that don’t are going to be at a huge disadvantage.
uced th port costs by 60 percent.
d that ved 1.1 million hours in underwriting up to 6x year over year.
lion bu run rate annual savings.
ckets o non-AI businesses.
going really interesting year to watch over the next 12 months.
g to se more anecdotes.
ompanie can figure it out and there are going to be companies that don’t.
these c ions have had to orient their business to be ready for AI as well.
ersion , like, using a chatbot, right, and how much productivity gains that actually gets you.
right?
actuall etely upend your system’s information and back end to be ready for AI, a lot of that is probably latent and being built up now into actually seeing the outcomes associated with it.
ing the markets.
most 80 t of the S&P 500’s return.
the ma ng driving the economy and the stock market.
doing v l.
s are s
going u ou know, there’s some blips like the last couple of days.
ly doin
s are v nd.
eviden roth is minimal.
nce is by UPS growth.
racted y, maybe more than slightly, if you’re a SaaS company over the last few days or a couple weeks.
market ced on, in general, earnings and earnings growth.
ltiples gher than average, but nowhere near the dot-com.
t look charts and see where we are.
gives comfort.
nings o ompanies that are the biggest drivers of the market in general, I feel like are pretty sound.
ood.
ealth o companies, I would say, is pretty good.
are hig n average in the past, but they don’t feel super alarming.
ding te anies that I was just talking about are the best businesses in the history of the world.
er a lo od of time, they have shown margin improvement that suggests that is probably true.
w, that he left side of the page.
aying f its, not loss-making growth.
ntrast , 22 era, sort of 21 era.
ig cont om a dot-com.
s, mult re not that high.
ike, su , you know, five slides worth of materials.
than i een in the past.
ow, the gh expectations for a reason, and we’re optimistic about the impact of AI flowing through to earnings, you know, overall in the public markets in the coming years.
your a n on the right side, which is, you know, if you just took a four box of, like, low growth, high growth, low margin, high margin, and paired up those types of companies, this is a chart that shows how they trade.
or the mpanies.
the tw ns on the right is high growth, high margin companies, and then high growth and low margin companies.
iously wth, low margin.
should rewarded.
ow.
hat are rowth and high margin, and, you know, the high growth and low margin, as long as they have good unit economics and they’re scaling into their margins, they should be rewarded.
is goo
growth, f you’re high margin, it’s tough out there.
rising.
about t the past in many different forms.
wth is gest thing that drives returns over five to 10 years.
r me to gh growth is rewarded more than low growth.
h growt igh margin, you’re one of those great businesses, it’s being very rewarded.
we’re g talk about supply side of the CapEx buildout.
assive.
ncentra the investment is inherently risky, just given how big it is.
ubbly f , the underlying fundamentals, I would say, bear little resemblance to previous bubbles.
inanced ily by historically profitable companies, like very profitable companies that I had talked about.
enter ture.
celerat ch is good.
l, we’r ly monitoring the sort of cost of training and the economics of that whole equation.
pretty
e big m mpanies that spend money on training models is pretty good.
g that .
e think I is going to be, you know, the biggest model buster that I’ve seen in my career, certainly.
model b so I won’t spend too much time on them.
es that aster and longer than anyone would have modeled in any scenario.
classic f this.
ke cons odels from pre-iPhone to five years later, four years later, consensus models were off for Apple’s performance by a factor of 3x over four years.
e most company in the world at the time.
nk that me thing is going to happen in many pockets of AI where the performance just massively exceeds, you know, what any expectations in a spreadsheet would show you.
is itse del buster.
h has d d high margin revenue at unprecedented speed and scale.
xpensiv , but repeatedly surprised to the upside, I would say, and creates value, I would say, far in excess of the capital that’s required to grow.
n to th ll be different, you know, this time around.
dot com is actually supported by cash flows and CapEx as a percentage of revenue is considerably lower.
headlin an zoom to the next slide, but, you know, I feel much better about this CapEx, you know, dynamic than dot com, obviously.
e ones bearing the biggest brunt of the CapEx, and this is a very good thing.
ortfoli nies, this is great.
it.
as much ty in the ground, get as much supply as you possibly can on the ground for training and inference.
thing.
anies t bearing most of the brunt of this are the best businesses of all time that I had talked about before.
e’re st to monitor is the introduction of debt into the equation.
e all o orecast CapEx that’s to come with cash flow, and we’re starting to see some debt.
this cl
invest ily in companies with exposure to debt.
le with h of the companies on the page financing with cash flow, continuing to produce cash flow, and using debt even?
rosoft, VIDIA as counterparties?
reat ab t.
s I fee about.
about a hem.
arties same.
rting t rivate credit get a little bit more involved in the data center build out.
n, the that’s very well covered that is kind of making a bet the company move into becoming a cloud is Oracle.
ow, the en profitable forever and reducing their shares forever.
apital ey are committing is very large.
cash f ative for many years to come.
ou foll of the buzz around it, like the cost of their credit default swaps has gone up, you know, to like 2% over the last three months.
ng stuf this.
general stuff for our portfolio companies, but we want to make sure that the market overall is healthy as well.
lide th s the magnitude of the pace of change of AI.
ld out revenue to what happened with Azure.
s comin relative to the cloud.
years h one year of AI revenue.
rosoft ng data, which I think is a cool way to frame how quickly this has happened.
‘s take y long time.
d out i ning much faster.
s for A venue to surpass their CapEx.
think rt of ratio or equation is going to happen much faster with AI.
ek out h on depreciation, but this is one of the topics that gets a lot of buzz in finance circles.
are yo mptions around depreciation of chips in particular?
cing fo GPUs is very solid.
ith mod it longer, but later users quickly switch to the new thing.
side.
the mo e.
even to year-old TPUs, Google actually disclosed this, seven to eight-year-old TPUs actually have 100% utilization.
monito rice of chips in the secondary market.
nt A100 100s has actually held up very, very well.
s of ch still getting fully utilized.
thing I about yet, but it gets a lot of buzz and, you know, sort of alarmists who like to talk about risk in the system.
itive s
at we t ut all the time is this paradox, right?
cheape umption goes up.
s repor d is well in excess of supply.
they s .
ewed Ga er, a friend of mine at our AI summit, and he was comparing the buildout of the internet and laying all the fiber to the buildout of data centers here.
big lin here is, you know, there is no dark GPU.
PUs.
ber.
r and t u know, it laid there dark and it wasn’t used.
the sy a data center, it gets fully utilized immediately.
y good ou know, in terms of, you know, demand meeting supply immediately.
rlier.
ld come ese companies.
xpectat
then th probably be disrupted if they can’t change.
t, agai he biggest reason why we see things, you know, that haven’t sort of dramatically shifted yet.
‘s not diness of the technology itself.
know, p buildout that needs to get built around the technologies and then change management and putting it in production.
as scal staggering clip relative to other categories.
shows ckly generative AI in app revenue has grown from 23, where it was basically, you know, you can barely even see it on the page to now.
ow, you see it on the page to now, you can’t see it on the page to now, but you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page.
the pa ow, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page.
the pa ow, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now, you can’t see it on the page to now.
sume 20 ns and a 22 times PE, that translates into 35 trillion of new market cap. There’s been about 24 trillion of new market cap that’s been pulled forward. Now we can debate if that’s attributable all to AI or otherwise, you know, large tech performance. But there’s still a lot of sort of market cap to go get, where you could have upside if, you know, if those assumptions are right.
sort of few cuts on trying to address this sort of AI payback question. So current estimates put cumulative hyperscaler CapEx at a little less than 5 trillion by 2030. So if you do napkin math on that to achieve a 10% hurdle rate on that 4.8 trillion or almost 5 trillion of investment, annual AI revenue would have to hit about a trillion dollars by 2030. So to put that into context, a trillion dollars, that would be
GDP to e a 10% return. It’s possible that happens. It’s also possible we could fall a little short of that. But I think it’s limiting just to look to 2030. I think the payback of this probably happens, you know, over a longer period of time, like, you know, between 2030 and 2040 as well.
ing it t’s about, you know, 1% GDP to get to, to get to the payback number of a 10% hurdle rate. All right. Heard it on the street. What we’ve started to do is we’ve sort of built software to track what all of the AI or what all of the tech public technology companies discuss in their earnings calls.
how re it is to our business at the early stage and, you know, the growth stage. And we package it all up and we share it out to our CEOs. So, you know, they can kind of have a simple digestible format of like, what do I need to know about AI as it relates to the public technology companies? You know, how does it, how does it impact my business, et cetera. And so we shared a bunch of the, you know, the stuff that we, that we track in here.
one que efore we moved to the private section, which a lot of folks, of course, on this call care about in this transition here. But before we get to that, so where are we calibrating to your trillion dollar in AI revenue, you know, thereabouts in 2030? Where are we today relative to your guesstimate of AI enabled revenue? And how far off are we to that trillion dollar number?
he, I w obably guess in the 50 billion range.
And th o perfect way to do it. I mean, I know, I know some of the big inputs to it. The harder stuff to track is, honestly, the big tech companies, like how much real AI revenue do they have?
ds can , they will from time to time give percentage uplift from AI. But I think depending on how they want to paint the picture, they can play games with that a little bit. So, you know, I think it’s, I think it’s, I think it’s, that’s a, that’s a rough swag. But like, you know, trillion, we’re probably at 50, but it’s growing, you know, way, way, way faster than a hundred percent year over year.
bly tha ue, I mean, ChatGPT launched three years ago, but substantially most of this traction happened in the last year and a half-ish or so, if we’re being really generous too. Is that a fair characterization?
Yeah. k, you know, it’s not just ChatGPT now on the consumer side, you know, Google has a business, XAI has a business. And then, you know, on the B2B side, you know, not only do the big model companies all have large API businesses, but the clouds have it too. And so a lot of the, you know, a lot of the sales that are model sales are also flowing through the clouds.
Yep. O ol. We have some questions on, on the private company side, but I’ll let you get through this section and then I’ll tee you up for it.
appy to o questions if you want on it. I mean, this, a lot of the stuff that we’ve talked about, you know, the big themes for me on the private market side, you know, companies are obviously staying private longer, but this is such a real asset class now.
ars, th r of public companies has been cut in half. Um, you know, the, the vast majority of companies that are a hundred million dollar plus revenue companies are private, something like 86%. Um, so, you know, that’s, that’s a major shift. Um, we can go, you can skip a couple of slides forward. Uh, basically I’ll talk a little bit about power laws. Cause that’s, I think that’s interesting.
f that n’t talked about as much, but value very much concentrates in the outlier companies. So the collective valuation of North American and European unicorns is about five and a half trillion dollars. The 10 largest ones, if you just take those, um, comprise almost 40% of the entire value. So, and that’s actually doubled since 2020. So sort of value, you know, sort of value is being, uh, concentrated in the biggest and best winners.
real t have four, five, six, seven of the 10, our portfolio companies of that 10. Um, so, you know, we, we’ve got a reasonable amount of coverage on that. Power laws are happening in the public markets too. So large cap has tripled since 2019. So what, what constitutes a large cap company has actually tripled since 2019. And I think this, the chart on the right side is super interesting. This was new data analysis that we had done.
you lo he lifespan of an average company on the S and P 500, that’s what that chart shows. That’s what the numbers represent. The light, like once a company is on the S and P 500, how long is it on there?
s actua actually, if you look over the last 50 years, that has declined by 40%. The amount of time it stays as part of the S and P 500. So disruption to companies happens faster and faster and faster, which I think is a very interesting dynamic and, and sort of matches what we’re seeing, you know, just in terms of like speed of change in, in, in the markets driven by technology.
o talk ower laws in our business too. I didn’t choose the title of this slide. Uh, I recognize all of the, you know, questions and concerns about it. Um, so the, the volatility laundering thing is, is a, is a big debate in our circles too. Um, mostly around founders who are trying to debate the merits of the private markets and the public markets. And,
the Co s did an interview where, maybe it was John, uh, did an interview where he talked about, you know, managing your stock price and avoiding volatility. And you can kind of orderly fashion, bring your stock price up over time. And that makes it easier to retain employees, hire employees, manage morale, uh, et cetera, et cetera. Um, and so I get the merits of that. I also think there are really, really strong merits of being a public company as well.
to hav lly, really interesting 18 months where we’re going to have some of the big kind of private for a very long time companies that go public. Um, and that’s a good thing in my opinion too. Um, some of the stuff that we show in this chart is just volatility and the observation that over time volatility has gotten a little bit more extreme in the markets. To me, this is a little bit cycle driven too. I know it’s short, short duration is sort of what we’re measuring. Um, but there’s merits to both.
uch lar the private side. We have embraced that new reality. I think it’s, it’s been a big benefit to our business in terms of getting content, getting to continue to invest in these companies over time. Uh, but obviously, you know, there’s, there’s a path of being a public company and getting liquidity, which we care a lot about too.
um, th e two questions. Uh, I will queue up for you here. Uh, one on Databricks. Can you talk about their transition, uh, from being a pre AI company now to a fully embedded AI company and what that’s been like?
all, I ou need to, you know, I mentioned Toby, like the reason Shopify has embraced it is because Toby has led from the top and he runs the business, you know, with AI at the center and, and he, he sort of performance manages everyone, uh, to, you know, to make sure that they do that. Ali is the same. Ali is this unique blend of, um, sort of commercial kind of terminator. I talk about him. He’s called him the technical terminator.
r. You have a commercial instinct and understand the importance of the value creation opportunity in AI. And then you need to actually be deep enough in the technology to know what to build. And so it just so happens that their, um, their sort of cloud data warehouse, or they call it the data lake, um, is actually a great way to have your data in a place to run AI workloads on top of it. So, you know, that was sort of a good place to be for them.
ry aggr y iterated on new AI products. They have this, uh, new product called agent bricks, which we’re super, super excited about. We think it’s going to be really big and transformative for them. So, um, I would say that’s a piece of it. And then they have the big AI native companies all as customers. And so, you know, they have the technology, they have the low cost technology. Um, and so, you know, a big thing that we look for when we’re making investments in companies is who are their customers.
fer the ers of our portfolio companies to be the modern thinking ones, you know, the door dashes of the world, um, you know, the Instacarts of the world, the Ubers of the world, then the very, very old school stodgy companies. Cause that means that their technology is evaluated by smart technologists and they pick it.
edge AI ies are all building on top of data bricks. Uh, and so, you know, they have the chance to grow with them as they scale. Uh, but it’s also a really good, you know, validator that they have the right technology.
e. Than David, for taking us through that.
ncludin k to our investments, please see a16z.com forward slash disclosures.
lease s .com forward slash disclosures.
lease s .com forward slash disclosures.
lease s .com forward slash disclosures.
lease s .com forward slash disclosures.
lease s .com forward slash disclosures.
lease s .com forward slash disclosures.
lease s .com forward slash disclosures.
Các công ty AI thành công nhất hiện đang đạt doanh thu 100 triệu USD với tốc độ vượt xa mức tăng trưởng lịch sử của cả những công ty SaaS nhanh nhất, đồng thời chi tiêu ít hơn cho hoạt động bán hàng và tiếp thị. Tốc độ đáng kinh ngạc này, được thúc đẩy bởi nhu cầu khổng lồ từ người dùng cuối, đang định hình sự bùng nổ AI hiện tại. Theo nhà đầu tư mạo hiểm David George, những công ty dẫn đầu đã tăng trưởng gần 700% so với cùng kỳ năm trước, tạo ra tới 1 triệu USD trên mỗi nhân viên. Đây không chỉ là câu chuyện về hiệu quả, mà còn là minh chứng cho sự tăng trưởng dẫn dắt bởi sản phẩm, khi cầu vượt cung đến mức mọi GPU mới đều được tận dụng tức thì.
Tuy nhiên, đằng sau mức tăng trưởng chóng mặt này là những thay đổi quan trọng trong cách vận hành của doanh nghiệp. Các công ty truyền thống đối mặt với tình thế “thích ứng hoặc diệt vong”, buộc phải tái tưởng tượng sản phẩm và quy trình nội bộ xoay quanh AI một cách bản địa – không chỉ đơn thuần gắn thêm chatbot. Điểm nghẽn thực sự cho việc áp dụng AI trong doanh nghiệp không phải là sự sẵn sàng của công nghệ mà là thách thức khổng lồ trong quản lý thay đổi tổ chức. Trong khi đó, chính mô hình kinh doanh cũng đang tiến hóa, có khả năng chuyển từ định giá theo đăng ký và mức tiêu thụ sang các mô hình dựa trên kết quả, nơi công ty chỉ được thanh toán khi AI hoàn thành thành công một nhiệm vụ cụ thể.
Bức tranh tài chính mang nhiều sắc thái phức tạp. Dù định giá thị trường công khai của các công ty công nghệ hàng đầu ở mức cao, chúng chủ yếu được hỗ trợ bởi tăng trưởng thu nhập thay vì bong bóng đầu cơ, với AI đóng góp phần lớn vào lợi nhuận thị trường. Về phía cung, một chu kỳ chi tiêu vốn lịch sử đang diễn ra, chủ yếu được tài trợ bởi dòng tiền mạnh mẽ từ các gã khổng lồ như Microsoft và Meta, dù nợ cũng bắt đầu len lỏi vào hệ thống. Bất chấp quy mô đầu tư, việc sử dụng ngay lập tức và toàn bộ năng lực tính toán mới cho thấy nhu cầu cơ bản là rất thực tế. Kết luận là chúng ta vẫn đang ở giai đoạn rất sớm của chu kỳ sản phẩm 10-15 năm, với toàn bộ tác động đột phá và giá trị sáng tạo vẫn chưa bộc lộ đầy đủ.
Những hiểu biết bất ngờ
- Biên lợi nhuận gộp thấp có thể là tín hiệu tích cực: Đối với các công ty bản địa AI, biên lợi nhuận gộp thấp – thường do chi phí suy luận cao – có thể cho thấy khách hàng đang tích cực sử dụng tính năng AI, đây là một huy hiệu danh dự. Giả định là những chi phí này sẽ giảm dần theo thời gian, trong khi biên lợi nhuận ban đầu cao có thể ám chỉ AI không phải là cốt lõi trong giá trị sản phẩm.
- Tư duy “điện vs máu”: Một triết lý do nhà sáng lập dẫn dắt đang xuất hiện, nơi mọi nhiệm vụ mới được đánh giá bằng câu hỏi: “Tôi có thể làm việc này bằng điện (AI) hay cần làm bằng máu (sức lao động con người)?”. Điều này thể hiện sự thay đổi căn bản trong tư duy vận hành.
- AI đang làm tăng, không giảm, một số khối lượng công việc chuyên môn: Trái ngược với quan điểm tự động hóa, một ví dụ thực tế cho thấy luật sư doanh nghiệp đang có khối lượng công việc tăng lên bởi LLM trao quyền cho mọi khách hàng soạn thảo tài liệu pháp lý, đòi hỏi nhiều thời gian hơn của luật sư để xem xét và chỉnh sửa.
- Phần cứng máy tính cũ vẫn giữ giá trị cực cao: Chip chuyên dụng (TPU) 7-8 năm tuổi vẫn duy trì mức sử dụng 100%, và giá GPU thế hệ trước như A100 vẫn mạnh trên thị trường thứ cấp, cho thấy nhu cầu tính toán liên tục và không thể thỏa mãn.
- Đột phá đang tăng tốc: Thời gian trung bình một công ty nằm trong S&P 500 đã giảm 40% trong 50 năm qua, nổi bật cách mà sự đột phá công nghệ đang diễn ra với tốc độ ngày càng nhanh.
Điều đúc kết thực tiễn
- Với bất kỳ công ty nào, thích ứng AI phải mang tính nền tảng, không phải tăng dần: Thành công đòi hỏi tái tưởng tượng quy trình làm việc và sản phẩm với AI làm trung tâm, không chỉ thêm vào những phần hời hợt. Lãnh đạo phải thúc đẩy mạnh mẽ sự thay đổi này cả trong phát triển sản phẩm lẫn hoạt động nội bộ.
- Ưu tiên và đo lường mức độ tương tác sâu với sản phẩm: Để tăng trưởng bền vững, tập trung vào các chỉ số như thời gian người dùng trong sản phẩm và tỷ lệ hoàn thành nhiệm vụ, không chỉ doanh thu tổng. Tương tác cao cho thấy giá trị thực và dự báo khả năng giữ chân mạnh mẽ.
- Áp dụng công cụ phát triển hỗ trợ AI ngay bây giờ: Bước nhảy vọt về năng suất từ các công cụ như Cursor và mô hình mã hóa tiên tiến rất đáng kể (nhanh hơn 10-20 lần trong một số trường hợp) đến mức các tổ chức sản phẩm và kỹ thuật không áp dụng hoàn toàn chúng có nguy cơ tụt hậu thảm khốc trong vòng 12 tháng.
- Đánh giá nhiệm vụ mới qua lăng kính “ưu tiên AI”: Thiết lập thói quen đặt câu hỏi liệu bất kỳ quy trình hoặc khả năng mới nào có thể thực hiện được bằng tác nhân AI (“điện”) trước khi mặc định dùng sức lao động con người (“máu”).
- Công ty truyền thống phải nghiên cứu các chuẩn mực mới: Hiệu quả của các công ty AI hàng đầu, đo bằng doanh thu trên mỗi nhân viên, đặt ra một tiêu chuẩn mới. Các công ty hiện hữu cần hiểu rằng để cạnh tranh sẽ đòi hỏi đạt được đòn bẩy vận hành tương tự thông qua tích hợp AI.
最成功的人工智能企業正以驚人的速度實現1億美元營收，其增長步伐甚至超越了史上成長最快的SaaS公司，且它們在銷售與行銷上的投入更少。這種由終端用戶強勁需求驅動的驚人增速，定義了當前的人工智能熱潮。創投大衛·喬治指出，頂尖AI企業的年增長率接近700%，人均創造營收高達100萬美元。這不僅是效率的體現，更彰顯了「產品驅動增長」的威力——需求遠超供給，以致每顆新GPU上線即被全力佔用。
然而在這瘋狂增長背後，企業營運模式正面臨根本性變革。傳統企業面臨「不進化即淘汰」的壓力，必須以AI為核心重塑產品與內部流程，而非簡單嫁接聊天機器人。企業導入AI的真正瓶頸不在技術成熟度，而在艱鉅的組織變革管理。與此同時，商業模式正在演變，可能從訂閱制與用量計費轉向成果導向模式——唯有AI成功完成特定任務，企業才能獲得報酬。
資本市場呈現複雜圖景。儘管龍頭科技股估值高企，但主要受盈利增長支撐而非投機泡沫，AI貢獻了市場回報的重要部分。供給端正經歷歷史性資本支出循環，主要由微軟、Meta等超大規模企業的強勁現金流推動，債務融資也開始進入體系。儘管投資規模龐大，新增算力即時且徹底的利用率表明底層需求真實存在。結論是：我們正處於10-15年產品周期的起始階段，完整的顛覆性影響與價值創造尚未完全展開。
顛覆性洞見
- 低毛利率可能是積極信號：對原生AI企業而言，因高昂推論成本導致的較低毛利率，反而標誌著客戶正積極使用AI功能，這是一種榮譽勳章。業界預期這些成本將隨時間下降，而初始毛利率過高可能意味著AI並非產品價值核心。
- 「電力vs血液」思維模式：創業家群體正形成一種哲學——評估每項新任務時皆須自問：「我能用電力（AI）完成，還是必須用血液（人力）完成？」這代表營運思維的根本轉變。
- AI正在增加而非減少某些專業工作量：與自動化敘事相反，有案例顯示企業律師工作量不減反增——因為大型語言模型使每位客戶都能起草法律文件，反而需要更多律師時間進行審查修正。
- 老舊運算硬體保有極高價值：已有七八年歷史的專用晶片（TPU）仍維持100%利用率，前代GPU（如A100）在二手市場價格堅挺，顯示運算需求持續且難以滿足。
- 顛覆正在加速：標普500指數成分股的平均存續時間過去50年縮短40%，凸顯科技驅動的顛覆正以越來越快的節奏發生。
實戰啟示
- 對任何企業而言，AI適配必須是根本性而非漸進式的：成功需要以AI為核心重塑工作流與產品，而非表面添加。領導者必須在產品開發與內部營運兩方面強力推動這種變革。
- 優先關注並衡量深層產品參與度：追求永續增長時，應聚焦用戶產品使用時長、任務完成率等指標，而非僅看總營收。高參與度意味著真實價值，並預示著強勁的用戶留存。
- 立即擁抱AI驅動的開發工具：Cursor等高級編程工具帶來的生產力飛躍（某些場景效率提升10-20倍）如此顯著，產品與工程團隊若未全面採用，恐在12個月內災難性落後。
- 以「AI優先」視角評估新任務：建立制度性實踐——在默認採用人力（「血液」）前，優先思考任何新流程或功能能否由AI智能體（「電力」）完成。
- 傳統企業必須研究新標竿：頂尖AI企業以人均營收衡量的效率樹立了新標準。既有企業需意識到，未來競爭必須透過AI整合達到類似的營運槓桿效應。
Las empresas de IA más exitosas están alcanzando los 100 millones de dólares en ingresos a un ritmo que deja en ridículo el crecimiento histórico incluso de las empresas SaaS más rápidas, y lo hacen gastando menos en ventas y marketing. Esta velocidad asombrosa, impulsada por una demanda abrumadora de los usuarios finales, define el actual boom de la IA. Según el capitalista de riesgo David George, los mejores desempeños crecieron casi un 700% interanual, generando hasta 1 millón de dólares por empleado. Esto no es solo una historia de eficiencia; es un testimonio del crecimiento impulsado por el producto, donde la demanda supera tan drásticamente a la oferta que cada nueva GPU se satura al instante.
Sin embargo, detrás de este crecimiento vertiginoso hay cambios significativos en la forma en que las empresas deben operar. Las empresas tradicionales se enfrentan a un imperativo de “adaptarse o morir”, necesitando reinventar sus productos y procesos internos alrededor de la IA de forma nativa, no simplemente añadiendo un chatbot. El verdadero cuello de botella para la adopción empresarial no es la preparación de la tecnología, sino el inmenso desafío de la gestión del cambio organizacional. Mientras tanto, el modelo de negocio en sí está evolucionando, pasando potencialmente de precios basados en suscripción y consumo hacia modelos basados en resultados, donde las empresas solo cobran cuando una IA completa con éxito una tarea específica.
El panorama financiero presenta una imagen compleja. Si bien las valoraciones en el mercado público de las principales empresas tecnológicas son altas, están mayormente respaldadas por el crecimiento de los beneficios y no por la especulación, siendo la IA la que impulsa una parte significativa de los rendimientos del mercado. En el lado de la oferta, está en marcha un ciclo histórico de gasto de capital, financiado principalmente por los robustos flujos de caja de los hiperscaladores como Microsoft y Meta, aunque la deuda comienza a entrar en el sistema. A pesar de la escala de la inversión, la utilización inmediata y total de la nueva capacidad de computación sugiere que la demanda subyacente es muy real. La conclusión es que todavía estamos al principio de un ciclo de producto de 10 a 15 años, cuyo pleno impacto disruptivo y creación de valor aún está por desarrollarse.
Perspectivas Sorprendentes
- Los bajos márgenes brutos pueden ser una señal positiva: Para las empresas nativas de IA, los márgenes brutos más bajos, a menudo debidos a los altos costes de inferencia, pueden indicar que los clientes están utilizando activamente las funciones de IA, lo cual es una insignia de honor. Se asume que estos costes caerán con el tiempo, mientras que unos márgenes iniciales altos podrían sugerir que la IA no es fundamental para el valor del producto.
- La mentalidad “electricidad vs. sangre”: Está surgiendo una filosofía impulsada por los fundadores donde cada nueva tarea se evalúa preguntando: “¿Puedo hacer esto con electricidad (IA), o necesito hacerlo con sangre (trabajo humano)?” Esto representa un cambio fundamental en el pensamiento operativo.
- La IA está aumentando, no disminuyendo, algunas cargas de trabajo profesionales: Contrariamente a la narrativa de la automatización, una anécdota señaló que los abogados corporativos están viendo una carga de trabajo mayor porque los LLM capacitan a cada cliente para redactar documentos legales, requiriendo más tiempo de los abogados para revisión y corrección.
- El hardware de computación antiguo retiene un valor extremo: Los chips especializados (TPU) de siete u ocho años mantienen una utilización del 100%, y los precios de las GPU de la generación anterior como la A100 se mantienen firmes en el mercado secundario, lo que indica una demanda de capacidad de cálculo sostenida e insaciable.
- La disrupción se está acelerando: La permanencia promedio de una empresa en el S&P 500 ha disminuido un 40% en los últimos 50 años, destacando cómo la disrupción impulsada por la tecnología ocurre a un ritmo cada vez más rápido.
Conclusiones Prácticas
- Para cualquier empresa, la adaptación a la IA debe ser fundamental, no incremental: El éxito requiere reinventar los flujos de trabajo y los productos con la IA en el centro, no hacer adiciones superficiales. Los líderes deben impulsar agresivamente este cambio tanto en el desarrollo de productos como en las operaciones internas.
- Priorice y mida el compromiso profundo con el producto: Para un crecimiento sostenible, concéntrese en métricas como el tiempo del usuario en el producto y las tasas de finalización de tareas, no solo en los ingresos brutos. Un alto compromiso indica un valor real y predice una fuerte retención.
- Adopte ahora las herramientas de desarrollo potenciadas por IA: El salto en productividad de herramientas como Cursor y modelos de codificación avanzados es tan significativo (de 10 a 20 veces más rápido en algunos casos) que las organizaciones de producto e ingeniería que no los adopten por completo corren el riesgo de quedarse catastróficamente atrás en 12 meses.
- Evalúe nuevas tareas con una perspectiva “primero la IA”: Instituya una práctica de cuestionar si cualquier nuevo proceso o capacidad puede ser logrado con agentes de IA (“electricidad”) antes de recurrir por defecto al trabajo humano (“sangre”).
- Las empresas tradicionales deben estudiar los nuevos puntos de referencia: La eficiencia de las principales empresas de IA, medida por los ingresos por empleado, establece un nuevo estándar. Las empresas establecidas necesitan entender que competir requerirá lograr un apalancamiento operativo similar mediante la integración de la IA.
As empresas de IA mais bem-sucedidas estão atingindo 100 milhões de dólares em receita a um ritmo que supera em muito o crescimento histórico até das empresas de SaaS mais rápidas, e fazem isso gastando menos com vendas e marketing. Esta velocidade vertiginosa, alimentada por uma demanda avassaladora dos utilizadores finais, define o atual boom da IA. Segundo o capitalista de risco David George, os melhores desempenhos cresceram quase 700% em termos homólogos, gerando até 1 milhão de dólares por funcionário. Esta não é apenas uma história de eficiência; é uma prova do crescimento liderado pelo produto, onde a demanda supera tão drasticamente a oferta que cada nova GPU é instantaneamente utilizada ao máximo.
Porém, por detrás deste crescimento acelerado, estão mudanças significativas na forma como as empresas devem operar. Os negócios tradicionais enfrentam um imperativo de “adaptar-se ou morrer”, necessitando de repensar os seus produtos e processos internos de forma nativa em torno da IA — não apenas adicionando superficialmente um chatbot. O verdadeiro estrangulamento para a adoção empresarial não é a maturidade da tecnologia, mas o imenso desafio da gestão da mudança organizacional. Entretanto, o próprio modelo de negócio está a evoluir, passando potencialmente de preços baseados em subscrição e consumo para modelos baseados em resultados, onde as empresas são pagas apenas quando uma IA completa com sucesso uma tarefa específica.
O cenário financeiro apresenta uma imagem complexa. Embora as avaliações no mercado público das principais empresas tecnológicas sejam elevadas, elas são largamente sustentadas pelo crescimento dos lucros e não por especulação, sendo que a IA impulsiona uma parte significativa dos retornos do mercado. No lado da oferta, está em curso um ciclo histórico de despesas de capital, financiado principalmente pelos fluxos de caixa robustos dos hiperscalers como a Microsoft e a Meta, embora a dívida esteja a começar a entrar no sistema. Apesar da escala do investimento, a utilização imediata e total da nova capacidade de computação sugere que a demanda subjacente é muito real. A conclusão é que ainda estamos no início de um ciclo de produto de 10 a 15 anos, com o pleno impacto disruptivo e a criação de valor ainda por se desenrolar.
Visões Surpreendentes
- Margens brutas baixas podem ser um sinal positivo: Para empresas nativas em IA, margens brutas mais baixas — muitas vezes devido a altos custos de inferência — podem indicar que os clientes estão a utilizar ativamente as funcionalidades de IA, o que é uma medalha de honra. Pressupõe-se que estes custos diminuirão com o tempo, enquanto margens iniciais elevadas podem sugerir que a IA não é central para o valor do produto.
- A mentalidade “eletricidade vs. sangue”: Está a emergir uma filosofia impulsionada pelos fundadores, onde cada nova tarefa é avaliada perguntando: “Posso fazer isto com eletricidade (IA), ou preciso de fazê-lo com sangue (trabalho humano)?” Isto representa uma mudança fundamental no pensamento operacional.
- A IA está a aumentar, e não a diminuir, algumas cargas de trabalho profissionais: Contrariamente à narrativa da automação, um exemplo anedótico notou que os advogados empresariais estão a ver um aumento da carga de trabalho porque os LLMs capacitam cada cliente a redigir documentos legais, exigindo mais tempo dos advogados para revisão e correção.
- O hardware de computação mais antigo mantém um valor extremo: Chips especializados (TPUs) com sete a oito anos ainda mantêm uma utilização de 100%, e os preços de GPUs da geração anterior, como a A100, permanecem elevados no mercado secundário, indicando uma demanda sustentada e insaciável por capacidade de computação.
- A disrupção está a acelerar: O tempo médio de permanência de uma empresa no S&P 500 diminuiu 40% nos últimos 50 anos, destacando como a disrupção impulsionada pela tecnologia está a acontecer a um ritmo cada vez mais rápido.
Lições Práticas
- Para qualquer empresa, a adaptação à IA deve ser fundamental, não incremental: O sucesso requer repensar fluxos de trabalho e produtos com a IA no centro, não fazer adições superficiais. Os líderes devem impulsionar agressivamente esta mudança tanto no desenvolvimento de produtos como nas operações internas.
- Priorize e meça o envolvimento profundo com o produto: Para um crescimento sustentável, concentre-se em métricas como o tempo dos utilizadores no produto e as taxas de conclusão de tarefas, não apenas na receita global. Um alto envolvimento indica valor real e prediz uma forte retenção.
- Adote agora ferramentas de desenvolvimento potenciadas por IA: O salto na produtividade com ferramentas como o Cursor e modelos avançados de programação é tão significativo (10 a 20 vezes mais rápido em alguns casos) que as organizações de produto e engenharia que não os adotarem totalmente correm o risco de ficar catastropicamente atrasadas em 12 meses.
- Avalie novas tarefas através de uma lente “IA-primeiro”: Institua uma prática de questionar se qualquer novo processo ou capacidade pode ser realizado com agentes de IA (“eletricidade”) antes de recorrer ao trabalho humano (“sangue”).
- As empresas tradicionais devem estudar os novos benchmarks: A eficiência das principais empresas de IA, medida pela receita por funcionário, estabelece um novo padrão. As empresas estabelecidas precisam de entender que competir exigirá alcançar uma alavancagem operacional semelhante através da integração da IA.
a16z Head of Investor Relations Jen Kha speaks with general partner David George about the state of AI and private technology markets. David shares data on why AI companies are growing 2.5x faster than traditional software while spending significantly less on sales and marketing, driven by massive market pull and record-breaking ARR per employee. They discuss the rise of Model Busters, which are companies that grow faster and longer than anyone would have modeled, like the iPhone. They also highlight real-world adoption at Chime and Rocket Mortgage alongside portfolio breakouts like Harvey, Abridge, and ElevenLabs.
Resources:
Follow David on X: https://x.com/DavidGeorge83
Follow Jen on X: https://x.com/jkhamehl
Read The State of Markets – https://a16z.com/state-of-markets/
Stay Updated:
If you enjoyed this episode, be sure to like, subscribe, and share with your friends!
Find a16z on X: https://twitter.com/a16z
Find a16z on LinkedIn: https://www.linkedin.com/company/a16z
Listen to the a16z Podcast on Spotify: https://open.spotify.com/show/5bC65RDvs3oxnLyqqvkUYX
Listen to the a16z Podcast on Apple Podcasts: https://podcasts.apple.com/us/podcast/a16z-podcast/id842818711
Follow our host: https://twitter.com/eriktorenberg](https://x.com/eriktorenberg
Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures.
Stay Updated:
Find a16z on X
Find a16z on LinkedIn
Listen to the a16z Show on Spotify
Listen to the a16z Show on Apple Podcasts
Follow our host: https://twitter.com/eriktorenberg
Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures.
Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
Leave a Reply
You must be logged in to post a comment.