Sam Altman

I’m delighted to finally be investing in Asana, which I’ve wanted to do for a long time.

One of the things I’ve learned about companies is that 1) clear tasks and goals, 2) clearly communicated, and 3) with clear and frequent measurement are very important to success.  Most companies fail at all 3 of these, and they become more important as companies get bigger.  Asana is the best way to excel in these 3 areas.

“You make what you measure” is really true, and most companies don’t measure well at all.  I spend a lot of time talking to people who work at startups, and most employees feel like they don’t have a good sense of what specifically the company needs to get done and how all the tasks are going.  Better work tracking leads to better collaboration and better decision-making.

Another thing I’ve learned investing in startups is how important it is to have some users that really love a product (instead of liking it pretty much).  Asana has the level of product love that all great companies have in common.  As a small example, their recurring revenue has been incredibly sticky and more than doubled every year.

Asana is the kind of lever that could someday massively increase the productivity of hundreds of millions of people around the world.  There’s not only an opportunity for Asana to be a huge company, but also for Asana to materially increase the output for the planet—somewhat amazingly, software has not yet eaten this important part of the world. 

Finally, Asana has an incredible team that, as far as I can tell as an outsider, really believes in the mission and loves the work environment (the Glassdoor reviews, something I check before every late-stage investment, are among the best I’ve ever seen). 

These are all the ingredients that go into the development of an incredibly impactful and valuable company.  I’m very happy to be along for the ride.

First of all, congrats to Kyle, Dan, and the rest of the Cruise team.  You all have made amazing progress and we look forward to seeing more in the future. 

A popular criticism of Silicon Valley, usually levied by people not building anything at all themselves, is that no one is working on or funding “hard technology”.  While we disagree with this premise—many of the most important companies start out looking trivial—we want to be clear that we’re actively looking to fund more hard tech companies, and would love to see more get started.

At YC, we started funding these sorts of companies in earnest in 2014, to widespread commentary that this was a silly waste of time.  Cruise, which we funded that winter, is getting acquired by GM.  From the Summer 2014 batch, 3 of the 4 companies who have raised the most money since graduating YC are “hard tech” companies.

We expect many more big wins.  The YC model works much better for these sorts of companies than most people, including ourselves, thought.

So, if you’re thinking about starting one, we’d like to talk.  And we think we can help. (You’ll probably find a lot of other people willing to help too, although unfortunately you’ll still face major fundraising challenges.  But in many ways, it’s easier to start a hard company than an easy company—more people want to join the mission.)

Leave the Medium thought pieces about when the stock market is going to crash and the effect it’s going to have on the fundraising environment to other people—it’s boring, and history will forget those people anyway.   There has never been a better time to take a long-term view and use technology to solve major problems, and we’ve never needed the solutions more than we do right now.

Different YC partners have different interests, but I’m particularly excited about AI (both general AI and narrow AI applied to specific industries, which seems like the most obvious win in all of startups right now), biotech, and energy. 

We hope to hear from you.

We tell startups all the time that they have to grow quickly.  That’s true, and very good advice, but I think the current fashion of Silicon Valley startups has taken this to an unhealthy extreme—startups have a weekly growth goal before they really have any strong idea about what they want to build.

In the first few weeks of a startup’s life, the founders really need to figure out what they’re doing and why.  Then they need to build a product some users really love.  Only after that they should focus on growth above all else.

A startup that prematurely targets a growth goal often ends up making a nebulous product that some users sort of like and papering over this with ‘growth hacking’.  That sort of works—at least, it will fool investors for awhile until they start digging into retention numbers—but eventually the music stops.

I think the right initial metric is “do any users love our product so much they spontaneously tell other people to use it?”  Until that’s a “yes”, founders are generally better off focusing on this instead of a growth target.

The very best technology companies sometimes take awhile to figure out exactly what they’re doing, but when they do, they usually pass that binary test before turning all their energy to growth.  It’s the critical ingredient for companies that do really well [1], and if you don’t figure it out, no amount of growth hacking will make you into a great company. 

As a side note, startups that don’t first figure out a product some users love also seem to rarely develop the sense of mission that the best companies have.

[1] The other thing that these companies have, and that also usually gets figured out early, is some sort of a monopoly.

Maybe instead of a tech bubble, we’re in a tech bust.  No one seems to fervently believe tech valuations are cheap, so it’d be somewhat surprising if we were in a bubble.  In many parts of the market, valuations seem too cheap.  In the part where they seem too high, maybe they aren’t really valuations at all, because the deal structure has changed to become more like debt.

Many of the small cap public tech companies have taken a beating this year.  Companies like Yelp are trading at less than 4 times trailing revenue.

The tech mega-caps are monopolies and have deservedly high valuations.  But even then, I would not be willing to short a single one of Apple, Google, Amazon, or Facebook against the S&P.  Apple in particular trades at a single-digit ex-cash forward P/E.

2015 has seen the lowest level of tech IPOs as a percentage of all IPOs in seven years.  The S&P Tech P/E is lower than the overall S&P P/E.  Neither of these facts seems suggestive of a tech bubble.

On the private side, people complain all the time about early-stage valuations (and to be fair, they’ve felt high to me for four years).  But if you invested in every single YC company over the past three years at their Demo Day valuation (average Demo Day valuations haven’t moved much in the past three years) you’d be very happy, even though investors complain that YC is the worst example of overpriced companies.

The mid-stages also seem generally reasonable, though of course there are notable exceptions.  These exceptions get all the attention—not the hundreds of companies doing remarkably well, but that handful that have raised money at high valuations and are struggling or dead. 

On the whole, it seems harder than any time in the past four years to raise mid-stage rounds.  This is also not suggestive of a bubble. 

So where is the problem?  Late-stage private valuations.  But perhaps the answer is that these “investments” aren’t really equity—they’re much more like debt. [1] I saw terms recently that had a 2x liquidation preference (i.e. the investors got the first 2x their money out of the company when it exited) and a 3x liquidation cap (i.e. after they made 3x their money, they didn’t get any more of the proceeds).

This is hardly an equity instrument at all. [2] The example here is an extreme case, but not wildly so.  Investors are buying debt but dressing it up close enough to equity to maintain their venture capital fund exemption status.  In a world of 0 percent interest rates, people become pretty focused on finding new sources for fixed income.

There is a massive disconnect in late-stage preferred stock, because if you’re using it to synthesize debt it doesn’t matter what the price is.  The closer the rounds get to common stock (a less-than-1x liquidation preference, for example), the more I think the valuation means something.  Unsurprisingly, the best companies usually have the most common-stock-like terms (and “the best companies” are never the ones that seem overpriced for long anyway).

Some of this debt is poorly underwritten.  Some unicorns will surely die (and those are the ones everyone will talk about).  That doesn’t make it a tech bubble.  It’d be more accurate to say it’s a tech bubble if no unicorns die in the next couple of years. 

To summarize: there does not appear to be a tech bubble in the public markets.  There does not appear to be a bubble in early or mid stages of the private markets.  There does appear to be a bubble in the late-stage private companies, but that’s because people are misunderstanding these financial instruments as equity.  If you reclassify those rounds as debt, then it gets hard to say where exactly the bubble is.

At some point, I expect LPs to realize that buying debt in late-stage tech companies is not what they signed up for, and then prices in late-stage private companies will appear to correct.  And I think that the entire public market is likely to go down—perhaps substantially—when interest rates materially move up, though that may be a long time away.  But I expect public tech companies are likely to trade with the rest of the market and not underperform. 

But no matter what happens in the short- and medium-term, I continue to believe technology is the future, and I still can’t think of an asset I’d rather own and not think about for a decade or two than a basket of public or private tech stocks.

Thanks to Jack Altman, Patrick Collison, Paul Graham, Aaron Levie, Geoff Ralston, and Ali Rowghani for reading draft of this. 

[1] There are real problems with these distorted "valuations".  Employees these companies hire often think of them as real valuations.  It also often makes the company think of itself as much bigger than it is, and do the wrong things for its actual stage.  Finally, too much cheap money lets companies operate with bad unit economics and cover up all sorts of internal problems.  So I think many companies are hurting themselves with access to easy capital.

[2] Even before the shift to debt-like rounds, the disconnect between how much people will pay for 5% of a company in preferred stock vs. 100% of a company in common stock was massive (and for good reason--the downside protection alone with preferred stock makes it much different than common stock).  As this delta has accentuated, the public/private disconnect has gotten worse, and caused a number of problems for companies accustomed to valuations always going up.

Airbnb has recently been attacked by San Francisco politicians for driving up the price of housing in the city.  San Francisco has tried, and will continue to try, to ban Airbnb in various ways.  Last week, this excellent post was published on Prop F—“the Airbnb law”. 

I recently reached out to Brian Chesky, the CEO of Airbnb, to learn more about this. I am decidedly a non-expert on this topic, but here are some thoughts from a layperson.

I met Brian in 2008, when he started Airbedandbreakfast as…an affordable housing company.  He couldn’t afford to pay his rent in 10 days and his credit cards were maxed out.  He looked around and realized that he did have one asset he could monetize—his extra space.  And eventually, Airbnb was born and the sharing economy began.

Unfortunately, a lot of other people have problems paying their rent or mortgage.  75% of Airbnb hosts in San Francisco say that their income from Airbnb helps them stay in their homes, and 60% of the Airbnb income goes to rent/mortgage and other housing expenses. Making it harder to use Airbnb in San Francisco may make it impossible for some of these hosts to afford to stay in their homes and in this city.

In 2014 (the most recent year with available data) there were about 387,000 housing units in SF.  About 38% were owner-occupied, and the remaining 62% or 240,000 were rental units.  About 33,000 of these were vacant, generally as a side effect of rent control laws.  (I don’t honestly know if rent control is a net good or bad thing—I assume more good than bad—but it certainly keeps units off the market.) [1]

In the past year, only about 340 units in SF were rented on Airbnb more than 211 nights, which is what Airbnb has calculated as the break-even point compared to long-term rental.  This is less than one out of every thousand units of housing in SF.  Looking at it another way, it’s just over 1.1% of all unoccupied units.  

There have been about 10,700 SF units that have rented on Airbnb in the last year (obviously a much lower number of units are actively listed at any particular time).  The median number of trips per unit was 5, and mean was 13.3.  The mean revenue per host was about $13,000 per year.  More than 90 percent of Airbnb hosts in SF are listing their primary residence, and making money with an extra room or their entire place when they are out of town.

The whole magic of the sharing economy is better asset utilization and thus lower prices for everyone.  Home sharing makes better utilization out of a fixed asset, and by more optimally filling space it means the same number of people can use less supply.  In fact, Airbnb worked with economist Tom Davidoff of the University of British Columbia and found that Airbnb has affected the price of housing in SF by less than 1% either up or down.

But in the last 5 years, the cost of housing in the city has about doubled.  The reason for this is a lot more people want to live in SF than we have housing for, and the city has been slow to approve new construction.  Who is to blame for this?  The same politicians that are trying to distract you with Airbnb’s 340 “professionally rented” units.

What should the politicians actually be doing about the housing crunch?  The obvious answer would be to support building more housing and fixing the supply side of the equation.  But instead they’re doing the opposite (e.g. a moratorium on new construction in the Mission) and trying to turn Airbnb into a scapegoat.

I love San Francisco. I wish housing here were much cheaper.  This is a special city and more people are going to want to live here, and more are going to want to come visit and do business with people here. Instead of trying to ban the future, we should be making it easier for middle class families to stay in the city.  We can do this by building more units to push the market price of housing down and by making it easier for San Franciscans to share their homes.

Commentators are looking hard for what’s wrong with startups in Silicon Valley.  First they talked about valuations being too high.  Then they talked about valuations not really meaning anything.  Then they talked about companies staying private too long.  Then they talked about burn rates.

But something does feel off, though it’s been hard to precisely identify.

I think the answer is unit economics.  One of the jokes that came out of the 2000 bubble was “we lose a little money on every customer, but we make it up on volume”.  This was then out of fashion for a long time as Google and Facebook hit their stride.

There are now more businesses than I ever remember before that struggle to explain how their unit economics are ever going to make sense.  It usually requires an explanation on the order of infinite retention (“yes, our sales and marketing costs are really high and our annual profit margins per user are thin, but we’re going to keep the customer forever”), a massive reduction in costs (“we’re going to replace all our human labor with robots”), a claim that eventually the company can stop buying users (“we acquire users for more than they’re worth for now just to get the flywheel spinning”), or something even less plausible.

This is particularly common in startups that don’t pass the Peter Thiel monopoly test—these startups seem to have to spend every available dollar on user acquisition, and if they raise prices, customers defect to a similar service.

Most great companies historically have had good unit economics soon after they began monetizing, even if the company as a whole lost money for a long period of time.

Silicon Valley has always been willing to invest in money-losing companies that may eventually make lots of money.  That’s great.  I have never seen Silicon Valley so willing to invest in companies that have well-understood financials showing they will probably always lose money.  Low-margin businesses have never been more fashionable here than they are right now.

Companies that have raised lots of money are at particular risk.  It’s so tempting to paper over a problem with the business by spending more money instead of fixing the product or service.

Burn rates by themselves are not scary.  Burn rates are scary when you scale the business up and the model doesn’t look any better.  Burn rates are also scary when runway is short (i.e., burning $2M a month with $100M in the bank is fine; burning $1M a month with $3M in the bank is really bad) even if the unit economics look great.

The good news is that if you’re aware of this you can avoid the trap.  If there’s no other way to operate in your space, maybe it’s a bad business.  The low-margin, hyper-competitive world is not the only place to be.  Companies always have an explanation about how they’re going to fix unit economics, so you really have to go out of your way not to delude yourself.

If you hold yourself to the standard of making a product that is so good people spontaneously recommend it to their friends, and you have an easy-to-understand business model where you make more than you spend on each user, and it gets better not worse as you get bigger, you may not look like some of hottest companies of today, but you’ll look a lot like Google and Facebook.

First-time startup CEOs make a lot of mistakes, mostly due to ignorance.

One particularly bad one is misunderstanding or misusing basic financial terms.  I started noticing this in Y Combinator applicants a couple of years ago, but see it now in startups at all stages (including some YC companies). 

It is very important to make accurate financial statements to investors, and it is well worth the time it takes to learn the difference between concepts like “revenue” and “GMV” (gross merchandise volume) and revenue from a “contract” or “LOI” (letter of intent).  Most terms have very specific definitions, and it’s well worth a little bit of time learning what these are.  When in doubt, you will never get in trouble for defining the way you’re using a financial term too precisely.

I’ve seen people use GMV for revenue or refer to an LOI as a contract many times in the past year when talking to investors.  This is a felony.

Although investors should be doing more diligence than is currently in fashion, this issue is on the founders to fix.

At the end of a YC batch, the general consensus among the partners is that about 25% of the companies are on a trajectory that could lead to a multi-billion dollar company.  Of course, only a handful of them do.  Most go on to be decent or bad.

These companies have a beautifully exponential growth curve during YC, and then a few months after YC is over, it essentially flatlines.  Because it would be so much better for us if this did not happen, we wonder a lot about why.

The main problem is that companies stop doing what they were doing during YC—instead of relentlessly focusing on building a great product and growing, they focus on everything else.  They also work less hard and less effectively—the peer pressure during YC is a powerful force.

The startups justify this to themselves in all sorts of ways—“We’re doing some longer-term strategic work.  You wouldn’t understand.” “We’re cleaning up our technical debt.” “We’re building out the organization.” “We’re focusing on PR for this month.  I’m going to speak at 6 conferences and writing two thought leadership pieces.” “We are different; growth isn’t our most important thing.” We’ve heard all of these from startups that have gone on to disappoint.

In general, startups get distracted by fake work.  Fake work is both easier and more fun than real work for many founders.  Two particularly bad cases are raising money and getting personal press; we’ve seen many promising founders fall in love with one or (usually) both of these, which nearly always ends badly.  But the list of fake work is long.

I tell founders to consider how directly a task relates to growing.  Obviously, building and selling are the best.  Things like hiring are also very high on the list—you will need to hire to sustain your growth rate at some point.  Interviewing lots of lawyers has got to be near the bottom.

During YC, we are ruthless about reminding startups that fake work does not count and will still get you a failed startup no matter how intensely you do it.  We are also ruthless about asking for your progress, and being honest with you if things aren’t working.  After YC, we have less contact with startups—you can go dark on us if you want.  This, by itself, is almost always a sign that a startup is doing badly.

Momentum is everything in a startup.  If you have momentum, you can survive most other problems.  If you do not have momentum, nothing except getting momentum will solve your problems.  Founders internalize this during YC; many seem to forget in the few years after YC.  Burnout seems to almost always affect founders whose startups are not doing well, and then becomes a downward spiral.  In fact, one of my top few startup commandments is “never let the company lose momentum”.

There are a few other common problems.  One is a feeling of “we made it” that comes after a big financing round and a reduction in intensity.  A related problem is that after you’ve raised a lot of money or become somewhat well-known, it’s harder to admit that things aren’t working and you need to change direction.  Also, very small startups can grow by sheer force of will, even with a bad product.  This stops working after a few months as the numbers get larger, and if you haven’t built something people love, you will not be able to continue growing.

So how can startups avoid this slump?  Work on real work.  Stay focused on building a product your users love and hitting your growth targets.  Try to have a board and peers who will make you hold yourself accountable—don’t lose the urgency that you developed during YC.  Keep sending updates on your traction to your investors and anyone else who will read them (in fact, we’re building some new software at YC to automate this for our startups in the hope that it prevent some of them from going off the rails).  Make the mistake of focusing too much on what matters most, not too little, and relentlessly protect your time from everything else.  Don’t ever let yourself feel like you’ve won before you have.  I still don’t think the Airbnb founders feel like they’ve won.  You have to keep up a high level of intensity for many, many years.

Many YC startups learns these lessons after a year or two in the wilderness, but for some it’s too late and for all it’s a waste of time.

The best startups we fund keep on doing exactly what they did during YC.  This sounds so simple and so obvious, but in practice so few founders do it.

The good news is it’s doable with deliberate effort.  If every founder (YC and otherwise) did it, the number of successful startups would probably double.

A lot of us complain about how the government is not very good at technology.  The U.S. Digital Service is actually trying to do something about it, by applying the way startups build products to make government services work better for veterans, immigrants, students, seniors, and the American public as a whole.

This is clearly a good idea.  (See U.S. Digital Service Playbook for more details.)

Inspired by the successful rescue of HealthCare.gov, small teams get deployed inside government agencies to improve critical government software. 

It seems to be working.  To use HealthCare.gov again as an example, the Digital Service effort helped replace a $200 million login system that cost $70 million per year to operate (I know…) with one that cost $4 million to build and less than $4 million per year to operate, and worked better in every way.  In another example, at U.S. Citizenship and Immigration Services, a Digital Service team has been instrumental in enabling green cards to be renewed online for the first time and a growing number of other improvements to the immigrant experience.

The Digital Service attracted talent on par with the best Silicon Valley startups, including talented veterans from Amazon, Google, Facebook, Twitter, Twilio, YC, and more – engineers, designers, and product managers who have committed to do tours of duty serving the country.

As an American, I am grateful to these men and women for doing this.  Because of their work, the government will work better.

I often get asked about what people can do for a year or two to make a big impact between projects.  Here is a good answer.  Consider joining the ranks.  I think it’d be great if it became a new tradition that people from the tech world do a tour of duty serving our country at some point in their careers.  We need better technology in government.

In the early days of my startup, I used to get slightly offended when people would refer to it as a “project”.  “How’s your project going?” seemed like the asker didn't take us seriously, even though everything felt serious to us.  I remember assuming this would stop after we announced a $5 million Series A; it didn’t.  I kept feeling like we’d know we made it when people started referring to us a company.

I now have the opposite belief.  It’s far better to be thought of—and to think of yourself—as a project than a company for as long as possible.

Companies sound serious.  When you start thinking of yourself as a company, you start acting like one.  You worry more about pretend work involving things like lawyers, conferences, and finance stuff, and less about building product, because that’s what people who run companies are supposed to do.  This is, of course, the kiss of death for promising ideas.

Projects have very low expectations, which is great.  Projects also usually mean less people and less money, so you get the good parts of both flexibility and focus.  Companies have high expectations—and the more money out of the gate and the more press, the worse off they are (think Color and Clinkle, for example).

Worst of all, you won’t work on slightly crazy ideas—this is a company, not a hobby, and you need to do something that sounds like a good, respectable idea.  There is a limit to what most people are willing to work on for something called a company that does not exist if it’s just a project.  The risk of seeming stupid when something is just a project is almost zero, and no one cares if you fail.  So you’re much more likely to work on something good, instead of derivative but plausible-sounding crap.

When you’re working on a project, you can experiment with ideas for a long time.  When you have a company, the clock is ticking and people expect results.  This gets to the danger with projects—a lot of people use them as an excuse to not work very hard.  If you don’t have the self-discipline to work hard without external pressure, projects can be a license to slack off.

The best companies start out with ideas that don’t sound very good.  They start out as projects, and in fact sometimes they sound so inconsequential the founders wouldn't let themselves work on them if they had to defend them as a company.  Google and Yahoo started as grad students’ projects.  Facebook was a project Zuckerberg built while he was a sophomore in college.  Twitter was a side project that started with a single engineer inside a company doing something totally different.  Airbnb was a side project to make some money to afford rent.  They all became companies later.

All of these were ideas that seemed bad but turned out to be good, and this is the magic formula for major success.  But in the rush to claim a company, they could have been lost.  The pressure  from external (and internal) expectations is constant and subtle, and it often kills the magic ideas.  Great companies often start as projects.