Speakers discuss Jensen Huang's scandal, succession risks at NVIDIA, and the emerging private data economy.
Summary
Speakers analyze the fallout from Jensen Huang's personal scandal, highlighting the risk of losing a central coordinator in the AI ecosystem. They debate whether founder-led structures like NVIDIA's are more critical than those of Microsoft or Alphabet in the current AI era. Mia Farnum introduces Yoni Rechtman's thesis that private data is becoming a primary value driver. Charles Hudson reflects on the AI data market, noting his previous underestimation of its size and the risks of stolen data being sold for training.
Key takeaways
- 1 Mia Farnum introduces Yoni Rechtman's thesis that private data is becoming a primary value driver.
Mia Farnum · 12:12
Episode sections
- 0:00 – 7:57
Jensen Huang Scandal and Ecosystem Risk
Speakers discuss the fallout from Jensen Huang's personal scandal, noting his central role in coordinating the AI ecosystem and chip access. They highlight the risk of losing a key coordinator who understands the entire field, potentially disrupting relationships and financing.
- 7:57 – 12:12
Succession and Founder-Led Company Dynamics
The discussion shifts to succession planning, contrasting founder-led companies like NVIDIA with those that have transitioned to non-founder CEOs like Microsoft and Alphabet. They debate whether founder-led structures are more critical in the current AI era due to the concentration of critical constraints and capital.
- 12:12 – 21:17
The Corpus is the Economy and Data Markets
Mia Farnum introduces Yoni Rechtman's thesis that private data is becoming a primary value driver. Charles Hudson reflects on the AI data market, noting his previous underestimation of its size, the risks of stolen data being sold for training, and the eventual normalization of pricing and terms.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| NVIDIANVDA | equity | neutral | potential Chinese secret operative, Alice Wang. We're obviously collecting details every day that are new, but the NVIDIA CEO and Alice Wang apparently. began their relationship over a year ago. Alice Wang was reported to be a former Chinese state media reporter, turned executive at Megaspeed, a Singapore data center company later |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
Jensen Huang had an affair with Alice Wang, a former Chinese state media reporter turned executive at Megaspeed, for over a year.
Mia Farnum ·1:04
How to verify: Primary source: Wall Street Journal article 'How Jensen Huang's Affair Spiraled Into a National Security Nightmare' cited in the episode description; verify via WSJ archives or SEC filings if Megaspeed is a public entity or through corporate press releases. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Primary sourceAlice Wang was investigated for allegedly diverting higher quality NVIDIA chips into China.
Mia Farnum ·1:24
How to verify: Primary source: US Department of Justice press releases or court documents regarding the Megaspeed probe; verify via DOJ.gov or PACER. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Primary sourceNVIDIA's board launched an independent investigation while stating it has full confidence in Jensen Huang and no reason to believe secrets were compromised.
Mia Farnum ·1:44
How to verify: Primary source: NVIDIA Corporation press release or 8-K filing with the SEC; verify via SEC EDGAR under ticker NVDA. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Primary sourceJensen Huang is one of the White House's closest AI advisors and lobbied to sell advanced chips into China.
Mia Farnum ·1:44
How to verify: Primary source: White House announcements or public records of AI advisory councils; verify via WhiteHouse.gov or official government transcripts. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
NVIDIA's stock price dipped around 10% since the news broke.
Mia Farnum ·2:23
How to verify: Primary source: Historical stock price data for NVDA from financial data providers like Bloomberg or Yahoo Finance; verify by comparing closing prices around the news date (Nov 2026) to pre-news prices. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Primary sourceJensen Huang dropped out of a handful of conversations and conferences, including 11 Labs and Anthropic.
Mia Farnum ·5:06
How to verify: Primary source: Conference schedules, social media announcements from 11 Labs and Anthropic, or news reports covering his absence; verify via company blogs or event archives. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Elon Musk paid $60 billion for Cursor.
Unidentified speaker (speaker_0) ·11:45
How to verify: Primary source: Press release from SpaceX or Cursor, or financial filings if publicly disclosed; verify via official company announcements or major financial news outlets. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Primary sourceThe spread for turning raw data into tasks and training environments can be anywhere from two to five times.
Mia Farnum ·12:52
How to verify: Primary source: Yoni Rechtman's article 'The Corpus Is the Economy' cited in the episode description; verify via the original publication on X or Slow Ventures blog. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
The public data corpus is already inside AI models, making private data the next value driver.
Mia Farnum ·12:32
How to verify: Primary source: Yoni Rechtman's article 'The Corpus Is the Economy'; verify via the original text. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Some portfolio companies are getting quotes per record an order of magnitude larger than standard for highly specialized data.
Unidentified speaker (speaker_0) ·20:42
How to verify: Primary source: Internal data from Precursor Ventures or anonymized case studies shared by the firm; verify via Precursor Ventures' public reports or direct inquiry if available. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.
Do your own research
- Review NVIDIA's recent SEC filings for any disclosures regarding Jensen Huang's status or succession planning.
- Analyze historical stock performance of founder-led tech companies versus non-founder-led tech companies during leadership transitions.
- Research Yoni Rechtman's published thesis on private data valuation.
- Investigate current market prices for legitimate AI training datasets versus black market data sales.
- Check NVIDIA's investor relations page for official statements on corporate governance and leadership continuity.
- Review industry reports on the growth trajectory of the AI data market.
- Look for counter-arguments to the thesis that private data is the primary value driver in AI.
- Examine legal precedents regarding the sale and use of stolen data for AI training.
Full transcriptexpand
- 0:00
- Unidentified speaker (speaker_0): If I were somebody who had been dependent on my relationship with Jensen Wong to kind of navigate all this, I'd be like, well, who do I call now? Who can pick up the phone and can speak for him or do the things that he did for me and my business? He's an important player in the coordination of that entire ecosystem, including access to chips, access to financing. I think this is such a complicated
- 0:19
- Unidentified speaker (speaker_0): ecosystem. We've taken out to some extent one of the few people who sort of sees the whole field and knows how it all fits together. I think for once, people are. going to say, oh, there are ways this could go wrong. Not saying it will, but like you can now envision, oh, these are the ways that this could not turn out in the rosy optimistic 100% success scenario that we had maybe envisioned before.
- 0:38
- Mia Farnum: It's the week of November 9th. Welcome to the Learning Corner, a precursor ventures podcast.
- 0:44
- Unidentified speaker (UNKNOWN): Welcome
- 0:44
- Mia Farnum: to the Learning Corner by Precursor. I'm Mia Farnum and I'm joined by our GP and managing partner Charles Hudson. where we will be talking through the articles and good reads that caught our team's attention over the past week. Our first piece is the story that's obviously taken over every single tech news outlet over the past couple days. Of course, it's the news of Jensen Huang's affair with a
- 1:04
- Mia Farnum: potential Chinese secret operative, Alice Wang. We're obviously collecting details every day that are new, but the NVIDIA CEO and Alice Wang apparently. began their relationship over a year ago. Alice Wang was reported to be a former Chinese state media reporter, turned executive at Megaspeed, a Singapore data center company later
- 1:24
- Mia Farnum: investigated for allegedly diverting higher quality NVIDIA chips into China. US officials have since concluded that she was likely passing information to Chinese intelligence. And this past Sunday, NVIDIA's board launched an independent investigation while saying it still has full confidence. in Jensen and no reason to believe any secrets were compromised. Something to note about this was
- 1:44
- Mia Farnum: that all of this unfolded while Jensen was one of the White House's closest AI advisors, lobbying hard to sell advanced chips into China, and the Wall Street Journal reports NVIDIA's general counsel reached out to the commerce officials overseeing the Mega Speed probe before the Justice Department took it over. Jensen has apologized to his family and colleagues. and we've seen
- 2:04
- Mia Farnum: CEO company affairs fall out in public, but not when the company happens to be at the center of the AI ecosystem, with its leader also famously not believing in succession plans as well. There are so many different routes that we can run here, talking about intervention or lack thereof from the US government nearly a year ago on the matter when this was
- 2:23
- Mia Farnum: surfaced, how this resulted in a, I think it's around 10. dip in NVIDIA's stock price since the news. There's been an uptick this morning since we're recording this podcast, talks of succession planning and more. So just wanted to pause for your thoughts here.
- 2:37
- Unidentified speaker (speaker_0): I think it's probably one of the more complicated things I've ever seen. It's a pretty consequential company. I mean, like we've seen CEO divorces, we've seen CEO affairs, we've seen these things before. I was talking to a few people. this weekend and I was like, look, mostly what you're paying a CEO for is judgment. That's kind of what you
- 2:57
- Unidentified speaker (speaker_0): hire them for, which is to make good decisions. And I don't know how this can be spun as anything other than a terrible decision, both on the personal side with his relationship with his wife and his family, but also on the professional side. In most companies, you know, getting involved with the supplier is a no-no. It's right up there with getting involved with a
- 3:16
- Unidentified speaker (speaker_0): subordinate in terms of like poor judgment. You know, it's also hard to know. The timeline, I think, still isn't fully established. It seems unlikely that nobody warned him that she might be a person who's put him in a compromised position from a national intelligence standpoint. So you would think normally, like, wow, we're having this sort of technological cold war with
- 3:36
- Unidentified speaker (speaker_0): China. This feels bad. The flip side is. it's the most important company in the stock market. I joked this weekend in our Discord. He's like the central banker of GPUs. There's no one more central to what's happening in AI than he is. And my question is how much of all of the neo cloud relationships, like all of the tentacles that Nvidia
- 3:56
- Unidentified speaker (speaker_0): has in the AI ecosystem, how much of that stuff is only in his head? Or where he's sort of the only person who has the complete picture about it? how all this stuff fits together, who's been in all the rooms, who knows all the promises like explicit and implicit they've made to people, and what happens if he goes away. And I have no idea whether he'll keep his job
- 4:16
- Unidentified speaker (speaker_0): or whether he'll get removed, but I suspect there will be some cooling off period where he's less visible, less sought thought after, and less central to what happens in AI. And I don't know who will come and take his place if he were gone for a prolonged period of time. So look, I think that the truth is, of course other companies will try to take
- 4:36
- Unidentified speaker (speaker_0): advantage of this moment in time, but they still are the undisputed kings of GPUs. Yeah. And even his departure will not immediately change that. Over time, sure, but not immediately.
- 4:46
- Unidentified speaker (UNKNOWN): And
- 4:46
- Mia Farnum: I think I was listening to a podcast this morning, or maybe it was squawk on the street. I think people have admitted no company is too big to fail at this point. It's basically the question is what companies will come up and take advantage of this in a way that does feel like it could be a game changer, AMD, etc. The succession part is the part that I think a lot of people are talking about. Everyone has an idea
- 5:06
- Mia Farnum: of who could possibly be in the runnings, maybe not to replace Jensen altogether, but to maybe take a more forward forward-facing facing role. We've already seen Jensen drop out of a handful of conversations. conferences, 11 labs. I think Anthropic was having some as well, no longer obviously going to be on the stage there. A lot of people have said like this is a very specific mix of skills that this person has
- 5:26
- Mia Farnum: to have. They can't just be product focused. They have to be an extremely talented engineer. That's synonymous with how Jensen has brought NVIDIA to life over the past what 30 years? It's crazy. Nvidia has been roughly basically unscathed from a media perspective over these last handful of years. So it's the first time
- 5:46
- Mia Farnum: people have had to talk about succession plans. Wanted to ask you from your perspective, do you think that, I don't know, even with government intervention, there will be pressure placed? I think, and I know we talked about this yesterday internally. I don't, we just had a wave of things that have shook the way that people think about AI. especially in the United States, throw in this AI arms
- 6:06
- Mia Farnum: race question. I think people have a duty to at least showcase that there's some control, even maybe if it's performative.
- 6:13
- Unidentified speaker (speaker_0): I have so many one is like, I think this was the predictable outcome of the desire to shut off all lawful. I don't I'm not and I'm not passing judgment on the policy. I think if you just shut off the most critical component. in the AI ecosystem to China, there are only two responses that you can do. One is
- 6:32
- Unidentified speaker (speaker_0): find other extra legal means to get your hands on chips like Chinese cutouts or things like that, or develop your own capacity to build something competitive. And I don't think it's an either or, I think it's an and. And I think this is an example of like if you can't get it through the front door, sometimes you go through the side door. And second, I think there's something special about founder-led companies. And I think we talked about this
- 6:52
- Unidentified speaker (speaker_0): internally. You know, I think a lot of people had questions when Steve Jobs left Apple. Obviously for under different circumstances, clearly. But I think when you have these long stretches of founder-led success, there is always the question, can somebody else do it? And I think if you look at Alphabet, Alphabet has figured out how to remain it may is it the same company it was
- 7:12
- Unidentified speaker (speaker_0): before? No. It's an absolute behemoth and a monster. I think after a few challenging CEO transitions, it feels like Satya Nadella has figured out how to make Microsoft relevant in a way that it was under Bill Gates. But these things are hard. And then I think the founder to first non founder CEO transition is hard. And I have no idea also what the
- 7:32
- Unidentified speaker (speaker_0): repercussions will be behind the scenes with government I don't think the government's become less interested in export controls and critical technology And I think it's embarrassing for the government too to have a public example of somebody who it doesn't really feel like even tried that hard. It maybe was even known to the government as a intermediary of some sort, if this is all true, being able to so
- 7:51
- Unidentified speaker (speaker_0): easily flaunt what was supposed to be an effective technology export control regime.
- 7:57
- Mia Farnum: From your perspective, does this crisis change the way that you think about? showing up in this space and investing in it? I think this event just shows how fragile the ecosystem in general. I think you said it yesterday. It is really like a Jenga stack. And you just don't realize that there are specific things that you could have never foresaw that were the things
- 8:17
- Mia Farnum: that created an implosion within an industry that created ripple effects that could swing either way. Do I think this is going to completely overhaul all of the progress? we've made? I don't think so. But if we find out more details throughout the coming weeks that show the case that it's far more risky than we've been briefed, then maybe I don't know how
- 8:36
- Mia Farnum: you're thinking about it right now and how it ripples to what we do day to day.
- 8:40
- Unidentified speaker (speaker_0): I feel like I've been like Mr. Neo Cloudberry. Yeah. But
- 8:43
- Unidentified speaker (UNKNOWN): I just think like that whole
- 8:44
- Unidentified speaker (speaker_0): I think that whole ecosystem is I don't want to be overstated. He's an important player in the coordination of that entire ecosystem, including access to chips, access to financing. I think this is such a complicated ecosystem. We've taken out to some extent one of the few people who sort of sees the whole field and knows how it all fits together. And if I were somebody who had been dependent on my
- 9:04
- Unidentified speaker (speaker_0): relationship with Jensen Wong to kind of navigate all this, I'd be like, Well, who do I call now? Who who who can pick up the phone and can speak for him or do the things that he did for me and my business? And the I don't have a good answer for that. So I do think it's between some of the things happening in the CapEx financing market, anthropics valuation on the IPO
- 9:24
- Unidentified speaker (speaker_0): coming down, but it sounds like there's still a lot of interest in it. I think for once people are gonna say, oh, there are ways this could go wrong. Not saying it will, but like you can now envision, oh, these are the ways that this could not turn out in the rosy optimistic 100% success scenario that we had maybe envisioned before.
- 9:39
- Mia Farnum: One last question on this, and it just came to me. Do you think in this era of AI where we are, being this extreme founder-led company is far more important than it was in generations prior? I'm thinking like internet mobile, or do you think, like, no, it's just the same? I think we just have more money being thrown at it. So it does feel a little bit more shaky
- 9:59
- Mia Farnum: when somebody moves.
- 10:00
- Unidentified speaker (speaker_0): I think we've always had companies that felt like they were the leaders of these transitions, but for different reasons. Cisco was kind of like the, but it was different. There weren't like router shortages. It wasn't like you couldn't get your hands on it. It was just more the plumbing was the constraint, and Cisco was the most important plumber in that era. And so Cisco was a company that
- 10:20
- Unidentified speaker (speaker_0): had outsized importance in that era. But I don't remember ever seeing anything. And you know, now you could argue that Nokia, Black. Blackberry, there were lots of other companies that were like really critical to ushering in the mobile era. But I never felt like any of those companies were dealing with like singular control of such a critical constraint. Maybe I'm forgetting something, but I I don't remember there being a company that was
- 10:40
- Unidentified speaker (speaker_0): both so central and dealing with such a manufacturing constrained item.
- 10:44
- Mia Farnum: And I think in this current era, when I'm thinking about how important it is to be like so founder-led, founder led, we've also added in a I think media is just obsessed with knowing the insides of these CEOs. I'm sure like with Enron, I mean there was a similar obsession with like Kenneth Lee
- 10:59
- Unidentified speaker (speaker_0): and
- 10:59
- Mia Farnum: skilling people just wanted to know more about them because they were so odd. And I think Elon Musk really did bring a new fascination into it of who are these people who run these companies. And so I don't think it's more important, but it's certainly more at the center. And we have so much more information to disseminate how these people tick or how we
- 11:19
- Mia Farnum: perceive them to tick. So I think the conversations around succession with Jensen become far more center.
- 11:25
- Unidentified speaker (UNKNOWN): Also,
- 11:25
- Unidentified speaker (speaker_0): I would say like NVIDIA is still a company that in this moment it's in a really dynamic area. It's not like in a stayed steady business where you're like, oh, you just need a good manager and a caretaker to come in and run the company. I think you look at a bunch of the decisions that Sam Altman has made and Dario have made and Elon have made. These are people who are operating at the frontier of tech, making really
- 11:45
- Unidentified speaker (speaker_0): bold like I don't know that a hired gun CEO of SpaceX goes out and pays $60 billion for cursor.
- 11:51
- Mia Farnum: Yes. I
- 11:51
- Unidentified speaker (UNKNOWN): don't
- 11:51
- Unidentified speaker (speaker_0): know if there's a person who would have believed they had the authority to do that. And that's like a pretty transformational Informational acquisition for SpaceX. And there are things I'm sure that the Frontier Labs are doing that are really based on the informed judgment of the founders who've been there since the beginning and have a vision for where those companies are going. And that's what I think makes founder-led
- 12:11
- Unidentified speaker (speaker_0): companies so special.
- 12:12
- Mia Farnum: Our next piece is another one from Yoni over at Slow Ventures titled The Corpus is the economy, getting a lot of traction over on X and written after connecting with dozens of people across the AI data supply chain from brokers and wholesalers to Frontier Lab buyers. His starting point is that the world's public data is already inside the models. So the next leap
- 12:32
- Mia Farnum: depends on private data that today sits on company balance sheets with a current value of zero dollars. There's no standard units, there are no real prices, and value only discovered at the moment of a sale. The money now is moving up the chain, turning raw data into tasks and then into full training environments. in the real world where the spread
- 12:52
- Mia Farnum: can be anywhere from two to five X, he says, which is why the biggest data vendors are quietly becoming AI labs and deployment shops themselves. Long term, Yoni sees the market splitting into two buckets. You've got this synthetic Walmart of tokens on one side and a scarce, higher value market for truly novel data on the other. That's the kind that
- 13:12
- Mia Farnum: only shows up when AI gets deployed into real businesses and hits edge cases no one has seen. He thinks deciding whether to keep your data internal or sell it will become as consequential for CEOs as deciding what to do with their cash. And he closes with this quote The first corpus was the public internet, the next corpus is
- 13:32
- Mia Farnum: the entire economy. Charles, how does this change the game for founders and company leaders as they evaluate? the largest value drivers for their business? It's been pretty common, at least I find, for founders to claim that they have proprietary data sets, but Yoni's arguing that there will be a massive divide between commodity type data and data that uncovers
- 13:52
- Mia Farnum: truly novel edge cases, which is where the real value sits. So wanted to hear your thoughts here.
- 13:59
- Unidentified speaker (speaker_0): Of all the markets in AI that I like really misinterpreted, this is like one of the biggest ones. I underestimated the appetite and opportunity in this category for labeled trader. One thing I think is interesting is it feels very full circle because when we first started talking about these models, even three or four years ago, everyone's like,
- 14:19
- Unidentified speaker (speaker_0): you can't put your data in Claude or Chat GPT, they're gonna like suck it all up and it's totally like risky. And then everyone just basically. said, Oh, I checked the box that said, Please don't trade on my data, and I hope that they're not trading on my data. I have an enterprise plan. And for whatever reason, we sort of collectively just got over that fear. And I think it's because it wasn't known what that
- 14:39
- Unidentified speaker (speaker_0): data was worth. And I think about the math-proof stuff and some of the other things, I do think we're probably gonna swing back to a world where people start to think harder about what information do I have, do I want to expose to these. models? And I would say the other thing is now there are companies I meet that would have previously just wound down and gone to zero and said, Hey, we
- 14:58
- Unidentified speaker (speaker_0): have one final play here, which is all of the data we have about our company, our operations, our customers' support tickets. And we haven't solved all the things about, you know, if you're a digital health company with a bunch of records about people, Hippah hippie Schmidt, schmitt, but like we haven't sort of sorted out like well, what happens to that? Who owns that data? What rights do
- 15:18
- Unidentified speaker (speaker_0): you have to sell it and distribute it? How can we consistently de identify it yet still? There's all of these questions. And I think my sense from talking to our portfolio companies is like with any frontier market, there's a distribution of uh sophistication of market participants. Some parts of this thing feel like the wild, wild west. You can buy data sets, you're like, how did you come across this data set? It And also
- 15:38
- Unidentified speaker (speaker_0): like I have all these other questions. Suddenly, like, what happens if somebody hacks into your system and steals all your data? Not to hold in a ransomware context, but they take it and go sell it. They go sell it to someone else who wants to train a model on your data. If you think about a two by two, like scrupulous buyer, scrupulous seller, I'm not that worried about that. Those people will. But you get into the unscrupulous Scrupulous seller,
- 15:57
- Unidentified speaker (speaker_0): unscrupulous buyer, the world could get very, very, very weird very, very quickly. because suddenly the value of that data for training purposes is very different than like stealing a bunch of credit cards and going to try to go, you know, buy a bunch of iTunes gift cards. That's different. And so I think we're embarking on a whole new world. And I do think, like Yoni said, at some
- 16:17
- Unidentified speaker (speaker_0): point, this data market is too big to not normalize. And at some point, we will We'll come up with something that feels like more consistent pricing and terms of service and things like that, but we're not. there yet. One
- 16:27
- Mia Farnum: of the takeaways that he said was that the budgets that these AI labs have to buy just raw data are gonna run out. Like there's not gonna be as much interest in just pure CSV files. What the big opportunity will look like is how can you pair your data with real-world experiences. He gives the example of Waymo.
- 16:47
- Mia Farnum: You need to both have the volume of runs to show. showcase that you can show an event many, many times, and then also locate the edge cases that those are true value aspects. One of his takeaways of the piece was we are so far from reaching the true peak of this value. I think everyone's sort of like, oh, is it picked over? And he compares it to the cloud market. He's like, this was 25 years
- 17:07
- Mia Farnum: ago. We're still figuring it all out. One thing that I continue to think about is that in order to really find these specific edge cases and failures, they're not there needs to be a shift in trust for the people who are leveraging this inside their businesses. Like it can't just be like, oh, a non AI native business has enrolled the support of an AI native startup to sit on top of their data and start
- 17:27
- Mia Farnum: running things. That's pre training. You need it to run, make mistakes, fail in front of the people that are trusting costing you to find those edge cases. I just don't know if we're there yet. And I think more so about the businesses that Yoni is. like we could do a big AI roll up over, some of the non-AI native businesses. And I think that's why he thinks there's so much opportunity around there. If you can just buy up all those
- 17:46
- Mia Farnum: businesses and then make the decision on how to control data and make decisions off of it. I don't know how you're thinking about that of timing of it all. How much time do we have before the value is kept capped as it pertains to data? I think raw-wise, raw wise we're we're pretty close for budgets to get less exciting. excited.
- 18:02
- Unidentified speaker (UNKNOWN): I think at some
- 18:03
- Unidentified speaker (speaker_0): point the budgets will normalize, probably because they'll be like, well, we have enough for now. And we can also now predict our with some level of fidelity what we're going to need in the future. So we don't have to go crazy and spend all of this money. We can we can do this for less. I think that will absolutely happen. But I also think there will be new frontiers where AI is like,
- 18:23
- Unidentified speaker (speaker_0): oh well now AI can solve this new technical, this new scientific problem. It's the models are good enough. Quick, now we have to go get a bunch of data about protein folding or like gene track. Like there'll be new areas where you're like, oh, well, we didn't need this data before because the models didn't know what to do with it. Now that we know that the models can solve this class of problems, all
- 18:43
- Unidentified speaker (speaker_0): of a sudden the data that you need to get good at solving those classes of the problems will be valuable. So what I think will probably be harder to do will be to build startups that try to anticipate what models can do and say, Oh, I'm gonna go collect a bunch of this data so that when the models are ready, that I can be a big deal. I think those founders will try to run that playbook. I think it will be very hard to do that.
- 19:01
- Mia Farnum: Yeah.
- 19:02
- Unidentified speaker (UNKNOWN): I think
- 19:03
- Mia Farnum: there was one part of me that was like, Oh, is this the the doom that we're entering in where everyone just becomes like a potential seller to a frontier lab company? Is that is that what we've become? I think. I've tried to shift my mind after reading the piece to be more so if you can move up the quote unquote totem pole, like he says, and start like extracting
- 19:23
- Mia Farnum: real value from your data, you can actually protect your own business and become more valuable internally. I think more specifically about Fortune 500 companies that are sitting on a lot of this data that are now probably making decisions of what we can do ourselves in-house so that we're not so reliant on some of the frontier lab companies. Is that the right way to think about it from your perspective?
- 19:40
- Unidentified speaker (speaker_0): Well, I think that the question is like if you're Marcora and you have all of this data and you're like, and we have access to open source models, maybe we should become uh front. Maybe we we have all the we have the data, like we're this intermediate. We're like collecting this data, packaging it, and selling it to these people. Maybe we could go
- 19:56
- Unidentified speaker (UNKNOWN): the other Maybe
- 19:57
- Unidentified speaker (speaker_0): instead of being just a broker, we could become the builder. And I think you will also find some companies that say, Wow, well, if anthropic wants to pay me $10 million for this data, maybe I should go get Kimmy or GLM. Maybe I should get my own model and train out. Maybe we should become some kind of small. So I think you're gonna find people who just say, Well, if the thing I'm sitting on is valuable, is selling it to somebody else to monetize it further
- 20:17
- Unidentified speaker (speaker_0): always the best answer? Or is this the case where I should take my destiny into my own hands and basically be the seller myself, the person who's getting all the value?
- 20:25
- Mia Farnum: I mean, he does make the point. that he thinks that's where, I don't know, like a strategic consulting firm, a professional services business could actually start becoming even more defensible. And like again, probably a year ago, people were saying or hedging this isn't gonna last for those types of businesses. They're sitting on data that can't be replicated or pre-trained on at this point.
- 20:42
- Unidentified speaker (speaker_0): We have a couple of portfolio companies where the quotes that they're getting per record are an order of magnitude larger than like the standard. And it's because it's highly specialized data that's very hard to get. And there'll be all these other weird businesses. There'll be people who will take your data set and shop it around and try to figure out who will pay you the most for it. There'll be some kind of that they'll we'll have a whole ecosystem around this. It's inevitable. Well,
- 21:01
- Unidentified speaker (speaker_2): we are all out of time today. Thank you, Charles, for hopping in. Thank you to our listeners for hopping in as well. Please like and subscribe, and we will talk with you all next week. Thank you. Bye.