# OpenAI’s research progress does not prove its business model works

Markdown copy of https://realalpha.app/s/openais-research-progress-does-not-prove-its-business-model/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

- Speaker: [Ed Elson](https://realalpha.app/people/ed-elson/) (Host)
- Show: Prof G Markets
- Episode: [SpaceX Raised $111B — It Already Needs More](https://realalpha.app/v/9SAgLMahZm4/), Oct 8, 2026
- Moment: 32:06 to 35:50 of the episode (4 min), playing on https://realalpha.app/s/openais-research-progress-does-not-prove-its-business-model/; the episode from there: https://realalpha.app/v/9SAgLMahZm4/?t=1926

## Quote

> “scientific research and enterprise business are **two very different ballgames**.”
> — [Ed Elson](https://realalpha.app/people/ed-elson/) at [35:15](https://realalpha.app/v/9SAgLMahZm4/?t=2115) · Prof G Markets

## Takeaways

1. The math proofs have not yet been officially verified.
2. Research breakthroughs do not establish that OpenAI can turn a profit.
3. Ed cites expanding losses and a $1.5 trillion valuation.

## About this segment

Ed Elson treats the math results as a research milestone, but says they do not answer whether OpenAI can be profitable. He points to expanding losses and a $1.5 trillion valuation against Walmart.

From “SpaceX Raised $111B — It Already Needs More”, Oct 8, 2026. [Watch the full episode](https://realalpha.app/v/9SAgLMahZm4/)

## Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

[32:19](https://realalpha.app/v/9SAgLMahZm4/?t=1939) OpenAI just dropped a bomb on the world of math. On Tuesday night, the AI Lab published hundreds of proofs that appeared to solve many of the world's most difficult math problems. These findings spanned problems across numerous fields, including number theory and algebraic geometry, and they even claimed to have reached a breakthrough on the Riemann hypothesis, which is a Millennium Prize problem known as one of the hardest to solve.

[32:44](https://realalpha.app/v/9SAgLMahZm4/?t=1964) According to Alex Kontorovich, a distinguished professor of math at Rutgers University, this would have been an instant Fields Medal winner had it been accomplished by a human. Now, I want to be clear, these proofs have yet to be verified. They were essentially dumped onto the internet via GitHub, so they haven't gone through any of the official processes of substantiation. And that is important because it means they haven't officially been solved.

[33:10](https://realalpha.app/v/9SAgLMahZm4/?t=1990) However, based on the reactions that I'm seeing from the mathematics community online, this does appear to be a very major breakthrough. These are problems that went unsolved for decades. And while some might criticize the way in which OpenAI tackled them, specifically drawing on existing research without proper accreditation or attribution, and also burning billions of dollars in the process, the truth is they are pushing the boundaries of math.

[33:39](https://realalpha.app/v/9SAgLMahZm4/?t=2019) and if our goal is to better understand the universe well i don't think there's any denying at this point that the company is helping to achieve that because i do want us to uncover the true nature of reality and i do want to find out what is actually going on out there and a deep understanding of math is a fundamental building block in pursuing that goal. Having said that, this still doesn't answer the question of whether OpenAI can become a viable business, because what we know about intellectual research is that it is very expensive and usually very unprofitable.

[34:19](https://realalpha.app/v/9SAgLMahZm4/?t=2059) And if you don't believe me, just go look at the income statement of any research institution. They are largely funded not with profits, but with donations. That is how the scientific discovery process generally works. So if your position is that OpenAI is pushing the boundaries of research, I 100% agree, and I am excited about it. But if you also believe that that is the path to becoming a profitable business, well, then that is where I run into problems.

[34:47](https://realalpha.app/v/9SAgLMahZm4/?t=2087) Because being good at math does not equate to being a good business. They are two fundamentally different pursuits that are being conflated as one and the same. So we can acknowledge the scientific accomplishments of this company, and we can recognize that they are groundbreaking. At the same time, though, we can also recognize that this is a business model that so far doesn't work. The company is still immensely unprofitable, and it appears that their losses are expanding.

[35:16](https://realalpha.app/v/9SAgLMahZm4/?t=2116) And despite that, they are still pursuing a $1.5 trillion valuation, which would be make the company twice as valuable as Walmart while generating less than a tenth of the revenues. In sum, scientific research and enterprise business are two very different ballgames. And unless OpenAI wants to burn billions until the end of time, eventually, they're going to have to choose.