1 year in ai is 7 on the internet.

i've been saying it for months now.
1 year in ai is 7yrs on the internet.

and it keeps getting more true.

SoftBank just put $40 billion into OpenAI.
Nvidia acquired Groq for $20 billion.
Meta bought Manus for $1B+.
all within weeks of each other.

i spent the last few months visiting
Stanford, Harvard, CMU,
Waterloo, Toronto, NYC.

talking to hundreds of cs students.
listening to pitches at every campus.
grabbing matcha and cookies with
kids who want to drop out
and founders who already did.

what i kept hearing
was the same question:
"what should i be doing right now?"

so here's my honest answer.
everything i believe, in one place.

the world has changed
and most people haven't caught up.

tools like Replit and Lovable
are making software creation cheap.

you can literally prompt
your way to a SaaS product,
slap a paywall on it, and ship.

what used to be gold is now dirt.

the only things worth building now
are on top of proprietary data.
domain specific. acute. rich.

if your product doesn't have that
it's a commodity.
someone will clone it by lunch.

but here's the thing:
that same shift is creating
massive opportunity.

~20 startups in sf right now
building computer use RL environments.
ecommerce checkouts.
dominos orders. you name it.

Mercor alone is at $900M ARR.

and these companies?
they hire new grads aggressively.
don't need ML PhDs.
betting on raw intelligence and agency.
that's it.

so if you're a cs student
and remotely ambitious,
come to sf.

i know that sounds like
i'm gatekeeping. i'm not.
it's pattern recognition.

the 2x3 mile grid
from north beach to dogpatch
is home to 10-100k
of the best people in ai.

researchers, engineers,
founders, investors.
all in one place.

the serendipity factor is 3x.
coffee shop conversations
turn into companies.
meetups turn into fundraising rounds.

yes we are in a bubble.
carried by openai and nvidia.
no guarantee when it fades.

that's exactly why
you should come now.
the window won't stay open.

i've seen what happens
when people actually make the move.

a year ago Powell St was
~15 people in our backyard.

last month we had 200
of the top founders, builders,
and investors show up.

6 companies found their cofounders
within our house.
9 roommates got into YC.
7 got backed by a16z.

and the thing that makes it work?
it's not the pedigree.

we've got people from
Jane Street, OpenAI, Palantir,
Tesla Autopilot, Meta,
Google, MIT, Waterloo.

but that's not why it works.

most overlooked fact
about our house:
we tolerate each other.

low ego. strong opinions.
debate for hours on ideas
not even startup related.

deep technical skill

highly opinionated
open-minded with each other.

that combination is rare.
and it compounds.

all you need is a year
to change your life.
especially when you have
a community pushing you forward.

now, what actually separates
the cs students who make it?

i've narrowed it down
to two questions:

"what is the most ambitious thing
you have done?"

"what has been the highest pressure
situation you been in?"

that's it.

maybe a third:
some form of grit measure.
sat scores, aps, good school,
or you diy'd it.

but the first two
are the real separators.
not your gpa.
not your club title.
not your linkedin headline.

speaking of clubs,
i need to be blunt.

a student messaged me:
"we are ivy + faang student club
that works w startups for free.
can you connect us?"

no.

startups don't care
about free work from clubs.

at startups, faang experience
is actually negative aura.

all a startup cares about is
full-time boots on the ground.

the greatest asset a startup has
is "leveraged time."

your 2-hour weekly club meetings
are the opposite of that.

takes weeks to understand
the nuances of a codebase
then you need to execute 996.

your little water-gun-pistol of a club
is a distraction.
no matter how smart the team is.

better path:
intern at a YC startup
or any funded startup.
more valuable than faang.

not all faang is bad tho.
core ai teams at google, xai, openai
still have strong signal.

but if you're choosing between offers
think about where the
innovation pressure is highest.
that's where you'll grow fastest.

and once you're in,
the number one skill
companies want right now
is independent thinking.

they're not looking for
hard work; tech stack; smarts.
those are almost given.

what they want is agency.
creative freedom.

they respect you when you debate.
when you ask questions.
when you push back.

3 ways to get better at it:
do projects; be informed; be vocal.

do all three
and you'll stand out
from 90% of applicants.

on top of that,
adopt a commercial mindset.

many ppl have bad linkedins.
others put effort into "zazz" —
the look, the attire.

but lack the depth
of how the world actually clicks.

i think its very powerful
to leverage what you know
about the world
as a way to stand out.

it shows you are tapped in.
it shows you might have
a decent network.

but most importantly
it shows ambition.

that need for more.
that need to learn more
in order to be satisfied.

when you ponder on
why things exist the way they do
good things happen.

i also want to say this:
code for fun.

most people code for roi.
school, work, hackathons.

code for the sake of code.
to tinker, to try, to explore.

see whats new, how it feels
what you can integrate.

expand your artillery.
expand your arsenal.

its easy to feel hesitant.
"i'm a better communicator"
"i'm not the best at coding"

its ok.
still try it.
give it a go.

passion can grow in time.
not every action needs to start
with "because i love it."

three habits for 2026:

adapt.
the way we code has changed overnight
and will keep changing.

Andrej Karpathy said it best:
new abstraction layer forming.
agents, subagents, prompts.

if you can use everything out there
you can be a 10x engineer.

ask.
snoop around the people
who are tapped in
and ask what's actually happening.

read deep technical blogs from
Netflix, LinkedIn, Meta, Google, Nvidia.

try coding without ai sometimes.
think about the "why"
before you reach for Claude.
compare your ideas
with what the ai gives you.

hiring managers are skeptical
of new grads who over-rely
on ai tools and can't explain
the fundamentals.

implement.
not "learn." implement.
get off localhost.
build something on your own
and see what forms from it.

that's how you actually learn
in this era.

now here's the part
most people miss.

i want to win.
and to do so
i have to be selfish
about my time and focus.

but i am also communal.
i want my friends to win.

the same energy i put
into my own pursuits
i put into theirs.

a win for us is a win for all.

the people who build each other up
end up going further
than the ones grinding alone.

that's why every sunday
rain or shine
we run at 10am.

it started as a small thing.
now it's yc founders, bio hackers,
pharmacists, doctors.

all people who share
one thing in common:

we are not from sf
but we came here
to create something special.

sf is known for the grind.
the "what are you building?"
the 996.

but amid all that chaos
sundays ground us.
they remind us we're human
and this is supposed to be fun.

so here's my ask.

if you're a cs student reading this,
make your 2026 plans now.

you have enough time.
just no time to waste.

build something. ship it.
come to sf. find your people.
take a gap year if you have to.

the best engineers run the world.

help me make this
one step closer to reality.

1 year in ai is 7yrs on the internet.
time to cook is now.
shake and bake.

see you at the next meetup.