ASK: I’m looking to talk to people in private equity or at PE backed companies about GTM transformation. If that’s you, just reply! Or find a time.

OpenAI projects $200 billion in revenue by 2030, burning $115 billion to get there. Here’s a chart, in case picturing the $3.7 billion in revenue growing over 5,000% in 5 years is difficult:

OpenAI forecast, courtesy of Titan

To put that in perspective, it is almost as much as Microsoft’s revenue today ($282 billion in 2025). I get that growth happens faster today, but it took Microsoft 43 years to get there. OpenAI is going to do it in roughly five. How?

Where is the revenue coming from?

According to internal documents obtained by The Wall Street Journal and The Information, OpenAI has a plan. By 2030, they expect:

• ~$90 billion from ChatGPT subscriptions (up from $10 billion ARR in 2025)

• ~$29 billion from AI agents—products like "Operator" priced at $2,000 to $20,000 per month

• ~$25-110 billion from monetizing free users through ads and affiliate commerce

• ~$22 billion from API/enterprise sales (up from $2 billion)

Let's start with the biggest bet: that companies will suddenly spend massive amounts on AI agents and enterprise software. From 2029 to 2030, OpenAI grows by around $45 billion—which is more than all of Salesforce's 2025 revenue. Is the idea that people just turn off Salesforce, and move all that spending over to OpenAI?

Or maybe they shut off Oracle, which did $57 billion in revenue in 2025?

Given the stickiness of CRM and ERP software I find that exceptionally unlikely. So the revenue would have to come from somewhere else–and eclipse, by a lot, what companies spend on their systems of record.

Typically large companies spend 2-3% of revenue on ERP or CRM. Were OpenAI to replace that spend, this would imply that OpenAI is serving companies doing a combined $6.67 trillion in revenue–representing, pick your relationship, 35% of the Fortune 500 by revenue, 24% of US GDP, or 6.4% of global GDP.

Perhaps I'm oversimplifying. Open AI says only about a quarter of the revenue will come from enterprise–$29 billion from AI agents and $22 from API and enterprise sales. Where does the other half come from? Consumers? Let's dig in there.

What about consumer?

Let’s say just for fun that only half of OpenAI’s revenue comes from enterprise. That leaves $150 billion to come from consumer.

Some will come from ChatGPT subscriptions–but the rate of paying there is only 5%. So while OpenAI forecasts $90 billion, up from $10 billion today, how exactly would that work? At some point you run out of people who can afford a $20 or $200 monthly subscription.

In fact–$90 billion in subscribers would be more revenue than Amazon Prime and Netflix generate globally–combined. In five years (versus 20 and 27 years respectively).

How about this? Let’s just say every Amazon Prime subscriber quits and spends that money on ChatGPT. That’s $44 billion–we can round up to $50. Now we just have to find the other $50 somehow.

If OpenAI consumes all the consumer markets–how much revenue is there? I don’t mean affiliate links, I mean just take all the revenue. This is a question that can be answered. Let’s look at the two biggest that it’s most likely to go after: travel and ecommerce.

In travel, Booking Holdings does $24 billion in revenue, Expedia $14 billion, and Airbnb $11 billion. So if OpenAI just absorbed all of that revenue, and those companies (or the consumers who have been using them for years) just stopped–that’s $49 billion. Those companies go away. Consumers only buy travel through OpenAI. Nobody puts up a fight, or retains any share. I don’t buy all of it–maybe some? Half? That’d be $25 billion, so we just need to find $25 billion from ecommerce.

AI takes over ecommerce. This one is tricky, as it’s more than pure software. There are pesky warehouses, logistics hubs, and trucks with “Amazon” written on them to be worried about.

Let’s simplify and start with software. Shopify–one of Canada’s biggest tech successes ever–did around $9 billion in 2024 revenue, roughly 3% of GMV on the platform (around $300 billion). If ChatGPT can convince consumers to buy from it instead of Shopify, including hosting their websites and handling all of the backend of running a retail business, let’s say it gets that full $9 billion.

But Amazon has the lion’s share of ecommerce revenue–$450 billion. But at massive capital cost–warehouses, logistics, and labor–and low margins. Not the kind of margins in OpenAi’s forecast for 2030. Could OpenAI get Amazon to pay it a referral fee?

No. No it will not. Amazon is not run by fools.

Putting this all together, if OpenAI eats the entire publicly traded travel market and all of Shopify’s revenue, it still comes up short–but within spitting distance of $100 billion. Maybe ads close the gap. Maybe. Thing is if Reddit has long struggled to figure out advertising, partially due to how unpredictable the content is, I’m not sure ChatGPT–much of which is based on Reddit content, with some added hallucinations–is going to appeal to advertisers on such a tight timeframe.

The other elephant

The revenue growth does not happen in a vacuum. While OpenAI is planning to consume roughly 10 or 20% of the entire global software market (at least, of public company numbers), Anthropic is a not-so-distant number two.

Anthropic forecast, courtesy of Titan

OpenAI has gone after the consumer market aggressively. While Anthropic’s Claude performs many of the same functions as ChatGPT, it is not nearly the same household name. Even though it’s vastly better (in my experience) at certain tasks. Still, it would be foolish to think that Anthropic doesn’t take any bite out of the markets OpenAI is going after.

The biggest risk to the forecast is enterprise. Anthropic generates 80% of its revenue from enterprise customers. It’s also doing so in anticipation of profitability much sooner, but why would we talk about that? Will OpenAI get all of the enterprise agent budget, and all of the market for AI APIs? Doubtful. Because Claude is already beating it there.

The other other elephant

Did I forget to mention Gemini? They only have 20% of the enterprise market anyway, and a meager ~90% of the consumer search market.

I’m sure Sam knows something I don’t: that Google will just roll over and let OpenAI take share, on search, ads and enterprise business. And while OpenAI spends $115 billion to achieve its full potential, Google won’t do anything to fight back with the ~$72 billion it generates in free cash flow every year.

To profit or not to profit

Veteran entrepreneurs point to how AI has transformed starting a company. You can do so much more with so many fewer people–and less capital, in general–that it feels like the startup old days. When raising $1mm was a seed round, not a joke. When you could feed your whole team with pizza–for most meals, most days.

And that’s what’s so weird about this forecast. Back in the old days, startups were profitable almost from the get go. Microsoft was profitable from the beginning. Their motto could be “Printing Cash Since 1982”. Google was profitable in year 3, and throwing off hundreds of millions by the time it went public in 2004.

While you might argue that OpenAI is different–it isn’t, actually. Google was a famously server-intense business. But instead of spending ahead of demand, it built its own servers from reject parts it bought on the cheap from manufacturers. It had a team of engineers building BIOS (the software that makes all of the parts work, usually provided by the manufacturer) because the manufacturer had abandoned them. It was a lot cheaper to buy the scrap parts for pennies on the dollar and pay very expensive developers to build software to run them than it was to pay for top of the line hardware.

OpenAI seems happy to just pony up for the top of the line hardware. And datacenters to hold them. And power to run them. With seeming disregard for costs. While spending on even more expensive, less profitable undertakings like image and voice applications.

Bigger than Microsoft

The forecast puts OpenAI close to Microsoft’s revenue today–$281 billion in 2025–just five years out. That means forecasting an increase of 5,400% from $3.7 billion in revenue in 2024 to ~$200 billion in 2030. On 2 years of data. Has any company ever done that? I’m sure many have forecast 50x growth on meager data–I mean actually achieved it.

Microsoft has had an impressive run of growth. The key difference is Microsoft was always profitable. In 2025 it generated $102 billion in profit. (That is, by the way, in the ballpark of how much OpenAI plans on torching to achieve its revenue forecast.)

The revolution AI represents seems at least on the same order of magnitude as the revolution Apple and Microsoft ushered in with the rise of personal computing. Maybe I’m old, and a curmudgeon, but I don’t see how this revolution happens 5x faster or is that much bigger. Or, why one company needs to spend so much money to get there.

Why?

The risks of overcapitalizing and overspending are real, whether you’re a small company raising your first venture round or operating at the scale of OpenAI. The risk is always most acute in that early phase of the hype cycle.

Remember Elon and his “full self driving” talk? That started in 2013. Yes, over a decade ago. I wanted to believe him. Lots of people wanted to believe him. And did believe him. Since then, we’ve had one major lawsuit about it and recently got a handful of Tesla robotaxis in San Francisco (of course all rides cost $4.20–yes, that’s a fact).

The question I ask is: why? Why compels OpenAI to grow this quickly? Why do they have to spend all of this money? What is so urgent here?

Some people have been using AI for the last ~12 months. Most people still aren’t really using it anywhere near its potential. Do we really need to scale this quickly? Is this the reason–that everyone else will start to use it?

And what if all of that demand doesn’t materialize, then what? Will OpenAI turn into the next WebVan, overspending on distribution centers until it went bankrupt?

Meanwhile startups building on AI tools–like Skyp–are more efficient and profitable than ever. I realize there might be bigger problems to solve than writing better sales emails (maybe?) but there’s absolutely no reason to subsidize usage to the tune of a hundred billion dollars to get there. Usage that has spiked–but may unspike, one never knows.

The real question

We've been talking about—and waiting for—AGI for 50 years. What's the rush now?

OpenAI is betting $115 billion that demand will materialize at a pace we've never seen before. That enterprise software buyers will abandon sticky systems of record. That consumers will shift all their shopping and travel through ChatGPT. That this happens fast enough to justify the spend.

Maybe the real bet is something bigger: that OpenAI creates an entirely new category where companies that were spending 2-3% of revenue on core systems now spend 4-6%. Which seems almost reasonable—until you realize that companies currently spend roughly 25% of revenue on human labor. If AI is truly replacing humans at scale, OpenAI would be capturing a fraction of that spend. But that raises a different question: what happens to the humans? And if AI isn't replacing humans at that scale, where exactly does $200 billion in new software spend come from?

Meanwhile, startups building on AI—like Skyp—are more capital efficient than software companies have been in decades. You can do more with less. Which makes the scale of OpenAI's bet even stranger.

The irony is that AI's biggest promise is efficiency. Yet OpenAI is pursuing the most capital-intensive path possible to get there—spending more to reach Microsoft's revenue in 5 years than Microsoft spent to build it over 43.

History suggests two possibilities: either this forecast is Elon-grade puffery designed to justify massive capital raises, or we're watching a potential WebVan moment for a technology with genuine transformative potential.

Either way, if Sam's selling shares at a valuation tied to this forecast, they're richly priced.