Discover more from benn.substack
Why are we still surprised that startups are hard?
We think they can be hacked, but the only enduring moat is effort.
I have some follow up questions for Jensen Huang.
Huang, who is sixty, is the founder and CEO of Nvidia, one of five U.S. companies currently worth more than a trillion dollars. A couple weeks ago, in an interview on the Acquired podcast, Huang was asked what company he’d start today if he were thirty again. Without hesitation, he said he wouldn’t start one. The work he put into Nvidia—which, given his net worth of $36 billion, has netted him more than a billion dollars a year, three million dollars a day, or $130,000 an hour—wasn’t worth it:
The reason why I wouldn't do it, and it goes back to why it's so hard, is building a company and building Nvidia turned out to have been a million times harder than I expected it to be—any of us expected it to be. And at that time, if we realized the pain and suffering, and just how vulnerable you're going to feel, and the challenges that you're going to endure, the embarrassment and the shame, and the list of all the things that go wrong, I don't think anyone would start a company. Nobody in their right mind would do it.
That's kind of the superpower of an entrepreneur. They don't know how hard it is, and they only ask themselves, "How hard can it be?"...
You have to get yourself to believe that it's not that hard, because it's way harder than you think. And so, if I go taking all of my knowledge now and I go back, and I said, I'm going to endure that whole journey again? I think it's too much. It is just too much.
The follow ups I wish they’d asked: Is this a statement about how hard it is to build a company, or about the actual utility—or lack thereof, I suppose—of being extraordinarily rich and relatively famous?1 If Nvidia wasn’t personally worth it, was the work worth it, in a kind of self-sacrificial way, for the greater good?2 And most of all, why was it such a surprise that building Nvidia would be so hard?
Though I can’t relate to a man worth more than Delta Airlines in very many ways, I can at least sympathize with that last point, because I’ve often asked myself the same thing. Starting a company—even one that’s worth 0.02 percent of Nvidia’s current market cap—was a lot harder than I ever expected it to be. And our experience seems to be the rule rather than exceptions, because nearly every founder you talk to says the same thing: “This is so much harder than I thought it would be.”
Which is…weird? Everyone is told startups are hard—the hardest thing you’ll ever do. We’re told that we think we understand, but we don’t; that whatever we think, it’s harder than that. Given how pervasive that message is—including from the guy who won the lottery we’re all playing3—why are we still consistently surprised by how hard it actually is?
Contrast Huang’s comments with every post-game interview or Hall of Fame speech from a professional athlete, who’s probably sacrificed more than your average startup founder. They always talk about the work; about putting in endless hours in the gym; about struggles and overcoming adversity and blood and sweat and tears. But in decades of watching athletes reflect on the long climb from their neighborhood gym or Little League field to becoming a world champion, I’ve never once heard one of them say that they were surprised by how hard it was. The injuries were part of the game; the setbacks were what they trained for; the losing seasons were part of the process. Nobody’s surprised by the suffering, because the suffering is what made them great.4
So why was Huang surprised? Why was I surprised? Was we delicate? Delusional? Surprises aren’t supposed to be surprising when you’re told over and over it’s coming. And yet.
Work, for better or for worse
When you tell people you’re starting a company, they will ask you what it does. If the conversation is with a venture capitalist, it will usually go like this:
Investor: So what does getai.tech5 do?
You: It’s like [ some other popular company that you hope that they wish they’d invested in ] but [ different in what you believe is some clever way ].
Investor: Couldn’t [ that popular company that they did invest in, but in their Series E in 2021, and are now just praying to get their money back ] build that?
You: They could but they won’t, because [ what you think is another clever-sounding point about market dynamics, and innovators’ dilemmas, and being “[the latest megatrend]-native” ].
Investor: Interesting, that does sound clever. [ Emails their now-teetering Series E portfolio company to ask them if they’ve ever considered adding [a clever-sounding but useless feature that nobody would ever want unless they just wanted to be a part of the latest fad]. The company’s CEO says thank you, that they’ll look into it, and then forwards the email to their trash can. ]
You: Yes, we’re excited about it.
Investor: It is a very exciting time.
If you have this conversation enough, you’ll stop looking at the popular company that’s been around for a decade for what it is—mostly years and years of very hard work—and start seeing it for what it isn’t—a missing piece; a single blemish; one fatal architectural flaw. Moreover, people will start getting excited about your company because you have that missing piece. They won’t ask you about all the stuff you have to copy to build a great email client, or a new CRM, or the next enduring social network; they’ll ask you about your clever wedge. Tell someone about your cloud-first Alteryx, or your warehouse-native HubSpot, or your dbt-backed Tableau, or your AI-enabled Zendesk, and they’ll want to talk about the cloud-first, warehouse-native, dbt-backed, and AI-enabled parts. And we’ll deceive ourselves into believing those are the important parts. But all the work is in the Alteryx, HubSpot, Tableau, and Zendesk parts.
Because big businesses aren’t built on gimmicks. You can blitz your way through Sand Hill Road on them, and spike up the app store charts on them, but you can’t actually replace Gmail, or Salesforce, or Instagram, or Instagram, or Instagram, or Instagram, with them. Even the seemingly-instant successes can’t become lasting companies without putting in the work—work developing checkbox features, building enterprise sales motions, creating support escalation policies, arguing about sales compensation strategies, defining career ladders, preparing for board meetings, negotiating office leases, hiring lawyers, changing lawyers, dealing with Trinet, helping employees deal with Trinet, responding to investor emails about clever-sounding but useless features that nobody would ever want unless they just wanted to be a part of the latest fad, setting up the office wifi, fixing the office wifi, yelling at Comcast Business about the office wifi, emailing customers about your new pricing model, redoing your all-hands format, introducing OKRs, again, and a list of a hundred other mundane things that we and everyone else entirely overlooks in their pitch decks. And after we got to about a hundred employees, there was a new list.
In other words, building companies and products are, to borrow Jeff Bezos’ famous analogy, like learning how to do a handstand. Most people “think that if they work hard, they should be able to master a handstand in about two weeks.” We get excited about our smart idea; we race it to market; we expect the cavalry to arrive and make this easier. But the reality is that doing a handstand “takes about six months of daily practice. If you think you should be able to do it in two weeks, you’re just going to end up quitting”—or, it seems, end up surprised by how hard you have to work.6
The downside of this is that we can’t hack your way to startup success. We can hack your way into getting attention, or term sheets, or even a few years of vertical revenue growth. But we can’t outrun the work; not forever. Eventually—as the company grows, as the market figures out our hacks, as more people copy what we’ve done—the work we’ve put in becomes our only competitive advantage. We can’t outwit it any more than an athlete can outwit their training. The only true shortcuts are lots of steroids or lots of crimes.7
I’d argue that this isn’t just true for companies; it’s also true for products. We talk a lot around here about how there’s no silver bullet in BI. Lots of us have looked for one; there’s a new startup every week a new one says they’re going to replace PowerBI or Tableau because what we think is a clever-sounding point about market dynamics, and innovators’ dilemmas, and being “[the latest megatrend]-native.” But we’ve all found out that these companies’ durable advantage isn’t that they once had a good idea; it’s that they’ve built hundreds of obvious features around a couple clever ones.
Go in the front door
The upside of all of this, however, is that we don’t actually need a clever idea to build a great company. When people ask us how getai.tech plans on beating Mailchimp, it’s tempting to say how we’ll start with a freemium product in some vertical they’re missing, build a community of enthusiasts, use data from those users to collect training data for our proprietary LLM, and so on. We’ll win by outflanking and outsmarting the competition. We can’t take them on directly, because they have such a big head start.
In hindsight, I think that’s a mistake in two ways. First, we can’t actually avoid taking them on directly. The clever plan doesn’t remove the need to compete head on; it just distracts us from it. And second, a lot of companies’ head starts aren’t actually that big. Handstands don’t take two weeks to learn, but they don’t take a lifetime either. Over years of work, companies will have spent a lot of time running in the wrong direction. There will be quirks and rough edges that come with constantly building new product additions.8
The better answer for how we’ll beat Mailchimp, then, is that we’ll line up with them, toe to toe, and build a better product and business. We’ll make Mailchimp, but faster; but better designed; but with more dedicated and relentless customer support. We’ll make Mailchimp without the rust.9
I don’t think it’s a coincidence that the companies that have a reputation for building the best products of this generation—places like Slack, Figma, Superhuman, Notion, and Linear—are basically new versions of a successful predecessor. These companies didn’t distract themselves with four-step strategic maneuvers or clever wedges into a new category that they claimed to be creating. They instead took on the challenge directly—build a better Hipchat, or Illustrator, or Gmail, or Google Docs, or Asana. They knew they’d have to do it eventually, and rather than pretend that they could avoid it, they started with that plan. The strategy was to do the work.10
Which all startups have to do eventually. The only question is if we’ll be surprised by it.
Jensen Huang, crushed by Brooke Monk (?), Jacksepticeye (?), and Bailey Sarian (?), but slightly more famous than TooTurntTony (?).
After winning the World Series Wednesday night, 11-year veteran Marcus Semien: “Everything I’ve ever worked for is for this moment.” Catcher Jonah Heim: “This is what you work your whole life for…We fought through adversity, injuries, and we came out on top.”
There’s another, less generous reason why some people are surprised: We’re used to being successful. We got good grades, or went to good colleges, or were told we were a high performer in another job. We did everything right so far; in a just and just world, shouldn’t our company be a winner too?
But that’s not how startups work. The market doesn’t care about how hard you worked before (or didn’t), or if you were successful (or weren’t). The market will erase that history. It gives you no credit (or blame) for it (or the lack of it). For people who are used to being respected for their resumes—and especially for repeat or celebrity founders who’ve logged a lot of hours or done something notable, and feel like they should be rewarded for that—that can be a devastating adjustment.
I’d argue that FTX is actually a really good example of how you can’t escape the work. My general sense of the whole situation—not that I’ve been mainlining a half-dozen daily updates about the trial or anything—is that FTX was an extremely lucrative business as a standalone exchange, Alameda was an extremely bad business that made lots of terrible trades, and SBF was a guy who was extremely uninterested in building anything resembling a proper company. So he blew up all three, by, one, committing lots of crimes, and two—and maybe more importantly, because I think it’s a big part of why he committed all the crimes—by not making any effort whatsoever to build the business around the very good exchange that would’ve protected it from the very bad hedge fund. It was all exciting hacks, and no boring work. And fraud. There were also seven counts of fraud.
I’m skeptical of a lot of today’s hot AI companies for this reason. It’s not that they’re bad companies; it’s that nearly all of these companies’ competitive advantage is that they have something like a year-long head start over companies being started today. That gap can be closed pretty quickly, especially when the market landscape is constantly shifting and everyone may be running in the wrong direction.
Mailchimp is just illustrative here; I have no idea if Mailchimp is good, bad, crusty, musty, dusty, or rusty.
Importantly, this is different than some rise-and-grind hustle culture cult. If a startup forces its employees to work 16-hour days building some supposedly differentiating hack, they’ll lose; if a startup asks its employees to work 8-hour days polishing core interfaces and providing great customer service and doing the boring things that lots of other people try to shortcut with supposedly differentiating hacks, they’ll win.