OpenAI’s Explosive Growth Continues

**SPEAKER_1** (0:02)
Bloomberg Audio Studios, Podcasts, Radio, News. Bloomberg Tech is live from the heart of Silicon Valley, with Ed Ludlow in San Francisco.

**Tim Stenovec** (0:22)
This is Bloomberg Tech, I'm Tim Stenovec in for Ed Ludlow. Coming up on the program, OpenAI is on track to generate annualized revenue of more than $40 billion. That's roughly doubling its run rate from the end of 2025
Plus a new city bill backed by Mayor Zohran Mamdani calls for Amazon to directly employ their delivery workers. What will Amazon do if the bill passes? We discuss. And Meta's former chief AI scientist is joining a new venture firm to invest in AI startups. We'll be joined by the firm's leader, Sean Johnson.
Well, let's start with today's big number, and that is $40 billion.
OpenAI is on track to generate analyzed revenue that tops that number. That's according to sources. That would roughly double its run rate from the end of 2025 I want to bring in Bloomberg's AI reporter, Rachel Metz, who broke this story. Rachel, what specific products are driving this annualized revenue run rate that is such a jump from last year?

**Rachel Metz** (1:20)
So according to what we've been told in our reporting, it's actually kind of a range of things. The company's consumer business is still by far its largest business, and they recently announced that they had passed this one billion active weekly user mark, which they've been pushing toward for a long time. So they've got more paying users on the consumer side. They also have a lot more businesses that are using it, and this is driven in some part by Codex, which is its coding assistant.

**Tim Stenovec** (1:49)
Okay, well, we can't talk about this as the OpenAI in a vacuum. How does it compare to what the latest is from Anthropic, its chief rival?

**Rachel Metz** (1:56)
I mean, they've been that sort of neck and neck in a way. Okay, I hate to say like it's one versus the other, because I think that it's like much bigger and broader than that, and there are a lot of companies here in this ecosystem. But I think Anthropic definitely has proved over the last year or so that they're a really handy rival to this company, and they have, especially on the coding front, they in some ways really outpaced OpenAI. But OpenAI is showing us its revenues going up, that customers are also willing to pay for its products as well.

**Tim Stenovec** (2:27)
Okay, we're talking top line here. Do we know anything about the bottom line and what expenses are and cost of revenue here and what it's spending?

**Rachel Metz** (2:36)
Ooh, we're still looking into that as far as like specific numbers. But what we do know generally, and I feel like we've talked about this so many times. We have.
It's so frustrating, but it's the same thing over and over. Compute is really expensive. It costs a lot. OpenAI has worked very hard to amass a war chest, essentially, of computing power, and it has that at its disposal. And yet the company also will still say that its biggest obstacle is getting enough compute. But that is going to take up a huge chunk of any revenue that it brings in. So other than that, I would say that's probably by far its largest expense.

**Tim Stenovec** (3:15)
Bloomberg's Rachel Metz out there on the West Coast. Rachel, thanks for the update. We're going to keep asking you those questions about top and bottom lines at these companies. Appreciate your time. Well, strategists at Citigroup and Bank of America are finding that the CHIPS index is in quote, bubble land after it rose over two standard deviations above its long-term trends in real terms. As risks develop in the AI trade, our next guest argues that early winners aren't necessarily winners forever.
And early winner NVIDIA is at the center of it all. Grenadilla advisory founder and CEO, Anna Rathbun joins us now. Anna, you talk about this quote, duration mismatch between the realities of AI infrastructure and investor expectations. How do you quantify that?

**Anna Rathbun** (3:59)
Yeah, you know, we're looking at quarter to quarter numbers, and perhaps that's not the right time frame, even though that's what we're used to as equity investors.
The hyperscalers and semiconductor companies are telling us, you know, look out to 2027, 28, 29 I mean, they're issuing bonds that go all the way out to 2066 I'm not saying that we have to wait until 2026, but we definitely have a mismatch in expectations of when we might actually see return on some of these investments. The revenue expectations that are being promised years out, they have to be able to meet it years out. And that takes a lot of investment today in numbers that we're totally not used to, right? So I think we have to, sort of, adjust our expectations to be a little bit more longer term focused than we are today.