Earlier this week, I I wrote an article They claim that there was no obvious AI bubble. I argued that AI companies are making huge investments in data centers due to the surge in demand for their services, and that demand is likely to continue to increase for years to come.
This led to a variety of thoughtful comments on the same basic question. “If there is so much demand for this technology, why are AI companies losing so much money?” As I thought about how to answer it, I decided it would be helpful to explain the intellectual framework I use when thinking about questions like this.
I’m not trying to claim originality here. The ideas discussed below are common in startup finance. However, I suspect that many readers have not spent much time thinking about them.
So this article does three things. First, we’ll show you a stylized example to illustrate some key ideas on how to raise money for a new company. Next, we’ll use real-world examples to explain how to distinguish between healthy startups and doomed companies. Finally, behind a paywall, we apply this framework to OpenAI and Anthropic.
My point is not that these companies are guaranteed to succeed. All startups face risks and these companies can definitely fail. And even if it survives, it may not produce a healthy return for investors.
But I would argue that OpenAI and Anthropic are following a standard playbook in the tech industry. The fact that they are losing more money each year does not necessarily mean that they are on the road to bankruptcy. Or even that it doesn’t mean anything particularly unusual is going on. After all, Amazon lost money for the first nine years of its existence. Today, the company is one of the most valuable companies in the world.
Imagine you start a coffee shop. This space costs $6,000 per month. Coffee beans cost $2 per cup, but will be sold for $4 per cup.
In the first month, you sell 250 cups and make $1,000 in revenue. However, you spent $500 on coffee beans and $6,000 on rent, resulting in a total loss of $5,500.
In the second month, you sell 500 cups of coffee. That is, $2,000 in revenue minus $1,000 in beans. However, it still doesn’t come close to covering the store’s $6,000 monthly overhead. You lose another $5,000.
Despite the early loss, I feel like I’m on the right path. Customers like coffee. They come again and again, and some even bring their friends. In the third month, you sell 750 cups, resulting in a loss of $4,500. In the fourth month, you sell 1,000 cups and lose $4,000.
Projecting into the future, you predict that you will break even at a time when you expect to sell 3,000 cups per year. This yields $12,000 in revenue, enough to pay the $6,000 bean purchase fee and $6,000 in rent. By the end of the second year, you expect to sell 6,000 cups of coffee per month and generate $24,000 in revenue. After subtracting the $12,000 in beans and $6,000 in rent, you’re left with a healthy $6,000 profit.
Starting a business almost always requires a large amount of money upfront before you can make your first profit. Even after launching, it usually takes time to build a customer base. Therefore, it is very common for businesses to incur losses, at least for the first few months, and sometimes for the first few years, until they grow large enough to cover their overheads and start generating profits.
Now imagine that your first store did so well that you decide to open two new stores a year after your first store. Therefore, in month 13, store #1 earns a profit of $500. However, the other two stores each lost $5,500. This is similar to the first store a year ago. The company suffered a total loss of $10,500, the largest loss in its short history.
Customers love the two new stores and they’re growing just as fast as the first. You become very optimistic and decide to open four. more In stores at the beginning of the third year. During the month, store #1 generated a profit of $6,500, and store #2 and store #3 each generated a profit of $500. However, because stores 4 through 7 are brand new, they each lose $5,500. Your company incurred a total loss of $14,500. This is also a record loss.
A financial analyst writes an article claiming that your company is doomed. The bigger the company, the more money is lost.
But you are sure that the analyst is wrong. Sure, the newest store is in the red, but that’s temporary. You expect the new store to be profitable over time, just like the previous store.
This may continue for some time. In the fourth year, you might open eight stores, and in the fifth year, you might open 16 stores. If you’re particularly ambitious and have enough patient and deep-pocketed investors, you might be able to open a new store for 10 years before making your first profit. But eventually you’ll stop (or at least slow down) the pace of opening, at which point you’ll have a large, profitable company.
This is a common pattern in the business world. Once investors believe that a company has a clear path to profitability, they are often motivated to fund the next round of expansion, such as designing another chip, releasing another software version, or expanding to another city, without waiting for the previous round’s investment to pay off. That’s why it’s common to see startups go through larger and larger funding rounds of $1 million, then $5 million, then $20 million, before producing a single dollar of profit.
This is especially common in the technology sector, as it is often a winner-take-all market. Frequently, economies of scale, network effector any other factor that makes the most popular search engines, social networks, or online retailers far more profitable than similarly run ones. You’d rather be Google than Lycos or Ask Jeeves. Therefore, once you (and your investors) are convinced that you have a viable business model, it often makes sense to spend a lot of money to stay ahead of your competitors.
Amazon has famously been doing this for a decade. Losses compounded as the company expanded from books to CDs, DVDs, consumer electronics, and many other products in the late 1990s and early 2000s. the company did not Obtained first full-year profit Until 2003, nine years after its establishment.
In the early days, many doubted whether Amazon would be able to turn a profit. However, the doubters were ultimately proven wrong. Today, Amazon is one of the five most valuable companies in the world. In 2025, it made a profit of $77 billion.
Of course, it doesn’t always work out that way. In 2017, startup MoviePass announced a service that would allow customers to pay $9.95 to watch one movie a day at a theater. Movie tickets are much more expensive than $9.95 for a month. 2018 interviewMoviePass CEO Mitch Lowe admitted that the company was losing $21 million a month on the service. But he insisted he was simply following in Jeff Bezos’ footsteps.
“Remember that Amazon has lost billions of dollars for over 20 years,” he said. “And today, it has become the most valuable company in the world.”
But MoviePass and Amazon differed in crucial ways. Amazon usually sold items above cost. If a CD costs $9.95 on Amazon, the retailer may have paid $7 or $8 for it. Amazon was only losing money because it was rapidly expanding into new markets that were not yet profitable due to start-up costs.
In contrast, the typical customer on the $9.95 MoviePass plan got movie tickets worth more than $9.95. MoviePass was simply buying these tickets from theaters at full retail price and eating their losses.
The technical term for this is “gross profit.”
-
My hypothetical coffee shop had a gross profit margin of 50%. This is because the cost of beans ($2) was 50% lower than the cost of coffee ($4).
-
In 2001, Amazon Gross profit margin 21% — If you buy a CD for $10, the cost on Amazon will probably be around $7.90.
-
in First half of 2018 MoviePass charged customers $121 million for MoviePass subscriptions, but its cost of revenue (i.e., the amount it paid for movie tickets) was $313 million. it works fine -159% Gross profit margin.
If a company has a positive gross margin, meaning it makes a certain amount of profit on every sale, it should be able to increase its profitability by increasing its scale. On the other hand, companies with negative gross profit margins may need a fundamental rethink.
