Time to Profit and Why Business Sustainability Matters

Considering the most prominent AI startups it feels like startups have more than a decade to get to profitability. It's a misleading picture.

Disclaimer: Examining the time to profit or funding of the most successful products has limited relevance to the overwhelming majority of early-stage startups. My goals here are twofold:

• Demonstrate how absurdly off we got with assessing the largest AI startups.

• Show how they still are ultimately subjects of the same economic realities as any other startup, even if they get way more leeway.

So if there are any lessons here—and there are—they are as applicable for your average Joe’s startup. Except even more urgent.

Otherwise, feel free to treat this one as a random musing.

---

One of the repeated themes here at The Pre-Pre-Seed is business sustainability. In pure financial terms, it means getting to a break-even point relatively fast.

By the way, I considered saying “as fast as possible,” but I don’t think it would always be a good aspiration. “Relatively fast” still means “way faster than VCs would suggest you to.”

My basic argument is that once the startup is profitable, the runway becomes infinite (OK, semi-infinite; we don’t operate in a perfectly stable business environment). You can experiment away till you find the vehicle that will give the product its ride for the next 5 to 10 years.

What’s more, save for tech startups, just about any other firm optimizes for fast break-even.

Do you know of a restaurant that’s been losing money for an entire first decade of its operations, and yet remained open? Or any small business in such a situation?

This passage from my look at the long-term viability of the current traction of the AI industry neatly describes what I’m trying to say here. It also inevitably triggers some reactions from the startup ecosystem.

Startups Are Different

We got used to the idea that tech startups somehow operate against a different set of economic rules. When you challenge that assumption, the riposte is at the ready.

You can always count on HackerNews for “balanced” comments

It boils down to: “We do have startups that were losing money for 10+ years, and they’re great! The corner restaurant metaphor is clearly wrong.”

I agree, the metaphor doesn’t translate 1:1, since no one invests billions in a small business based on a chef’s charisma (and yes, we do that with AI startups). Also, with VCs, once someone has already sunk hundreds of millions into a tech startup, they have an interest in keeping it going. It’s the classic sunk cost fallacy.

Yet, I would argue that even for startups, waiting for 10-15 years for the first sign of profitability is reckless. At best.

My classic argument:

• Google was profitable in year 3.

• Facebook was so in year 6. And that given that it had no monetization strategy whatsoever when they started.

OpenAI? For those who believe Sam Altman, it’s year 15.

Is Facebook an Outlier?

I heard a riposte that Facebook, with its 6 years, is an outlier. It took Uber 14 years to become profitable, so why not give Altman one more to break even with OpenAI?

OK, I’m game. Let’s dig deeper. I could find just one significant startup pre-2015 that took more than 10 years to get into the black. It’s SpaceX. In the class in which Facebook (let alone Google) played, a few years to profitability wasn’t an outlier. It was the norm.

However, as the argument goes, the startup timelines have extended over time. And they did. In the last decade, we’ve seen way more companies that didn’t care to, or couldn’t, show profit for long years, and yet they ultimately turned successful.

There’s Uber, Twitter, Tesla, Shopify, Spotify, Slack, Figma, and whatnot. So maybe, in the realities of the 2020s, waiting that long for break-even is the new norm, huh? Perhaps even if Facebook was not an outlier in its time, today it would have been?

Burn Rate as Startup Health Metric

Did you know that when we’re bleeding heavily, our blood pressure doesn’t change? Intuitively, it should go down, but it doesn’t. The reason is our hearts keep pumping at a higher and higher pace to compensate for blood loss. It does until it can’t keep up, and we crash.

If we look at blood pressure in isolation, it all looks fine all the way till we die. That’s why we look at both blood pressure and heart rate together. Either one is heavily off, and we have a problem.

By the same token, looking at startups’ years to profitability in isolation (and giving ourselves plenty of leeway, because it took Tesla 17 years, and just look where they are) is a straight way to repeat Foursquare history (and, believe me, it ain’t rosy).

So what’s the equivalent of heart rate for startups? Burn rate. How fast they’re spending their cash reserves. Obviously, no company that’s losing money will be overly open about how much they spend (even if it literally has “open” in its name, if you catch my drift).

Yet, we have a decent proxy: a combination of burn rate and revenue (which may be considered a compound metric, making it even better). It’s how much funding companies raise.

Long timelines don’t have to be troubling if the need for external funds is rare and/or limited. That would mean that a startup is either close to profitability or could have been so if needed (Spotify is a good example here).

Unprecedented Funding of AI Startups

We don’t need to look far to gauge how out of touch funding of AI startups has become. Thinking Machines Lab received $2B in a seed round based on funding theater and leadership charisma alone. All before they even announced publicly what they were working on (let alone shown any traction).

Would that be a shocker that their first release was received as underwhelming? The expectations had to be sky high.

Let’s look at more established companies, though. Both OpenAI and Anthropic have already raised more than any other startup mentioned here (yes, that includes the one that’s literally sending people to space). Looking at whatever we can infer about OpenAI's burn rate, they’ll raise more soon enough. I doubt Anthropic is much different.

All that with no signs of profitability round the corner. And we compare AI startups to firms that already are sustainable and their lifetime investment needs.

Funding versus Time to Profit

If we want to understand how much of an outlier OpenAI (or Anthropic) is, we can plot funding against time to profitability.

Even against money-hungry, patience-requiring startups such as Uber or Tesla, they are off. Way off.

And let me reiterate. AI investments are based only on funding to date (and we know they’ll need more), while the profitability timelines are speculative and come from leaders who aren’t known to be reliable with their predictions, to say the least.

So while every other data point is established, the AI ones are bound to move further off the chart.

Each step up or to the right on this chart means one thing: growing expectations of returns. It’s only natural that when investors put more in the pot and wait longer, they want more out of it.

At some point, revenue expectations may grow so absurdly high that it becomes clear the whole thing is not going to be the cash cow everyone expected. By which point, the reference story is not Figma or Shopify, but Fab.com or Foursquare.

The economic realities will finally catch up, even with the biggest bets of the startup ecosystem. Tech startups, after all, are not that different from any other company out there. They need to show profits eventually.

And they can bend realities for only as long.

Business Sustainability Is a Viable Strategy for AI

While there’s probably no going back for OpenAI and Anthropic now, it’s not like it’s the only available option. Even for AI startups.

Let’s juxtapose two vividly different strategies of AI startups. OpenAI bets on super-aggressive growth, raising stakes higher and higher (on borrowed money). If they win, both lenders (VCs) and OpenAI will be good. The thing is, that “win” is a moving target. The bar goes higher with each billion raised.

Conversely, Midjourney focused all their efforts on achieving early sustainability. They never raised money. When revenues couldn’t cover the infrastructure costs, Midjourney changed the product strategy to keep the company in the black. It’s not like they don’t have their challenges, but business sustainability doesn’t look like a major concern for now.

If you had to allocate your retirement money to investments in OpenAI and Midjourney, how would you distribute the funds? At this stage, it gets increasingly difficult to defend the thesis that OpenAI has a huge growth potential. And it’s definitely a riskier bet than Midjourney.

What Can an Average Startup Learn From All This?

It’s all good and fine, but how relevant is all that for a typical early-stage startup? For all we care, there will still be emerging successful products whether OpenAI ultimately wins or loses.

• The faster a startup reaches business sustainability, the better. Initially, VCs would tell you different, but that’s because VC is broken. And the longer you believe the VC mantra, the more dependent on the external funding you become.

• The money flowing to AI startups is off the charts (duh!). When you get people to tell you that Facebook taking 6 years to prove its business model and spending more than $2B in the process is an outlier for how fast and cost-efficient it made it, you can feel we live in a fantasy land.

• The accepted timelines extend mostly when the stakes go up. No one’s giving a decade to a company that raised “just” tens of millions. Yet, it’s a comfort very few startups would ever have.

• The change in investment landscape is massive. Companies raising more than $5B in total used to be an absolute exception. Today, it feels like a back-page story. I leave it to everyone to decide whether it already is a bubble. I know my take, and it’s not optimistic.

For the very end, one interesting observation. Let’s hypothetically assume that we are, indeed, in the AI bubble and valuations are overinflated. Is it that much of a problem? After all, Microsofts and Amazons of this world are wildly successful despite having fallen along with everyone in the internet bubble burst of 2000. So maybe it’s going to be the same with OpenAIs and Anthropics of our times?

By late 2001, Amazon’s valuation had fallen by around 95% from its peak. It took Bezos’ company 10 years to sustainably surpass its 1999 valuation. And it was still record time.

Microsoft, despite falling by mere ~60% took its sweet 17 years to recover. Cisco (falling by ~90%) never did.

And that’s not adjusted for inflation. In reality, the recovery timelines are even longer.

Optimizing for time to profitability and business sustainability may lack glamour, but it’s, by far, the most reliable business strategy for a startup.